THE TRADERS’ MAGAZINE SINCE 1982
Schrödinger’s Cat Finding information in market data
8
Filtering Price Movement
Introducing a new zigzag indicator
12
Predicting The VIX By reordering the data
26
10 Selling Tips
Knowing when is “when” 30
INTERVIEW
Technical analyst Boon Chin Low
REVIEWS
n Haguro Method n MetaStock XIV MAY 2015
34
www.traders.com
MAY 2015
Take Control of Your Trading with the
Professional Traders’ Starter Kit
™
*Based on the 5-year subscription rate of $299.99. Shipping extra for foreign addresses.
The Foremost Collection For Traders
This massive collection packages the best tools for trading and investing in any market! 1. Technical Analysis of Stocks & Commodities, the Traders’ Magazine™. The premier magazine for technical analysis. You’ll get five years — 65 issues — including our annual Bonus Issues with our Readers’ Choice Awards.
4. Traders.com Advantage™, premium website content delivering real-world technical analysis to you! You’ll get five years of charts, indicators, and “how-to” advice for specific markets, currencies, stocks, and commodities; near-term opportunities; price movement; new techniques. Posted in real-time with an archive of thousands of articles.
5. Working Money™, the Investors’ Magazine online. You’ll get five years of market observations; explanations of charts, 2. S&C Digital Edition. Recent complete issues available in their markets, and market sectors; money management; and entirety as PDFs for you to either download or read directly interviews with money people that will help you trade and in your browser. No more waiting for the mail to deliver your invest wisely. Articles added several times a month. magazine! You will still receive the printed magazine unless you opt for a digital-only subscription. 6. Article Code. Download or copy & paste code presented in past issues of Stocks & Commodities — no need to type it 3. Complete Digital Archive. The complete archives as PDFs out manually. — more than 17,000 pages — from Technical Analysis of Stocks & Commodities from 1982 through the present. The 7 . Optimized Trading. The optimized indicator values can be articles can be read in your browser or download to your used as starting points when trying to decide what values computer (or any device with Internet to input into your charting software. Search for a certain access and the ability to read a PDF)! symbol or company or build your own portfolio.
ALSO RECEIVE A FREE✝ BOOK WITH YOUR PTSK ORDER!
Charting The Stock Market: The Wyckoff Method, edited by Jack K. Hutson, is yours free✝ with your order for the Professional Traders’ Starter Kit. ✝
Shipping and handling charges apply for foreign orders.
Join us on Facebook at www.facebook.com/STOCKSandCOMMODITIES
Online: www.TRADERS.com Toll free: 1-800-832-4642 206-938-0570 • Fax: 206-938-1307 Email:
[email protected] Follow us on Twitter @STOCKSandCOMM
BREAK FREE FROM HIGH BROKER FEES
Keep more of what’s yours with our low cost platform for professional traders, investors and institutions. Interactive Brokers
Interactive Brokers
Commissions are:
Financing Rates1 are:
64-89%
78%
1
lower than
lower than
• Charles Schwab • TDAmeritrade • ETrade
• Charles Schwab • TDAmeritrade • ETrade
For 100 to 1000 shares and 1 to 10 option contracts
For a margin loan of $50,000 to $1,000,000
Interactive Brokers Executions1 are:
Stocks:
8¢
better than the industry per 100 shares
Options:
50¢
better than the industry per contract
Second half 2014*
Interactive Brokers
ibkr.com/iwantmore
LOWER YOUR COSTS TO MAXIMIZE YOUR RETURN
stocks • options • futures • forex • bonds — on over 100 markets worldwide from one account Interactive Brokers is a member of NYSE, FINRA, SIPC – Lower investment costs will increase your overall return on investment, but lower costs do not guarantee that your investment will be profitable. Supporting documentation for any claims and statistical information will be provided upon request. Commissions and Margin Rates as of March 2, 2015. Services vary by firm. [1] For additional information on our commissions and margin rates and execution quality, see www.interactivebrokers.com/compare. * Based on independent measurements, the Transaction Auditing Group, Inc., (TAG). 05-IB15-896CH856
CONTENTS
MAY 2015, Volume 33 Number 6
8 Schrödinger’s Cat
by John F. Ehlers What information is contained in market data? Can you develop an indicator or trading system that can extract this information to provide an edge in trading? Here’s a look.
FEATURE ARTICLE
TIPS
12 Filtering Price Movement
by Giorgos E. Siligardos Here is an alternative to the classic zigzag indicator, which may prove useful to visual technical analysts and chart pattern researchers.
22 Mean Reversion And The S&P 500
by Stephen Beatson It is generally believed that markets tend to mean-revert. But this is true for some markets more than others. Here’s an in-depth look at how the S&P 500 responds to mean reversion.
25 Futures For You
by Carley Garner Here’s how the futures market really works.
26 Predicting The VIX By Reordering Data
by Stephen Butts In recent years, the CBOE Volatility Index (VIX) has increased in importance and use as an indicator of market direction. This article demonstrates how the direction of tomorrow’s change in the VIX might be determined by restructuring readily available market data.
30 10 Selling Tips
by Thomas Bulkowski Do you spend as much time deciding to sell as deciding to buy? Here are 10 tips to make deciding when to sell easier.
INTERVIEW
34 TA For The Longer Term With Boon Chin Low
by Jayanthi Gopalakrishnan BC Low has been a teacher and practitioner of technical analysis since the 1980s. He is one of Singapore’s earliest practitioners to attain the Chartered Market Technician credential. At Singapore Polytechnic, he created and taught two modules of “Technical Analysis and Trading,” the only formal course on technical analysis in Singapore. He was a technical analyst for Merrill Lynch Bank, where he provided currency views to dealers, private bankers, and institutional clients. Currently, he continues to trade his own equity. We asked him about how longer-term investors can apply technical analysis.
REVIEWS 42 • Haguro Method Product review: MetaStock add-on based on the Haguro method 46 • MetaStock XIV Product review: Trading and charting platform
DEPARTMENTS 6 7 44 49 50 56 57 57 58 59 62
Opening Position Letters To S&C †Traders’ Glossary Trade News & Products Traders’ Tips Futures Liquidity Advertisers’ Index Editorial Resource Index Books For Traders Classified Advertising Traders’ Resource
41 Explore Your Options
by Tom Gentile Got a question about options?
AT THE CLOSE
60 Gambling, Speculating, & Investing
by Stella Osoba What do these terms mean as applied to the participant in the financial markets? Let’s have a look to try to come up with some clear definitions.
29 Q&A
by Don Bright This professional trader answers a few of your questions.
This article is the basis for
TIPS Traders’ Tips this month.
n Cover: Jose Cruz n Cover concept: Christine Morrison
Copyright © 2015 Technical Analysis, Inc. All rights reserved. Information in this publication must not be stored or reproduced in any form without written permission from the publisher. Technical Analysis of Stocks & Commodities™ (ISSN 0738-3355) is published monthly with a Bonus Issue in March for $89.99 per year by Technical Analysis, Inc., 4757 California Ave. S.W., Seattle, WA 98116-4499. Periodicals postage paid at Seattle, WA and at additional mailing offices. Postmaster: Send address changes to Technical Analysis of Stocks & Commodities™ 4757 California Ave. S.W., Seattle, WA 98116-4499 U.S.A.
Printed in the U.S.A.
4 • May 2015 • Technical Analysis of Stocks & Commodities
per month*
The math is simple. The benefits are clear. Technical Analysis of Stocks & Commodities, published since 1982, is the savvy trader’s guide to profiting in any market. For subscribers, our content, the heart of the technical analysis industry, is also available online in our digital edition — the complete magazine in digital format (PDF) on our website, Traders.com — available for download. We also offer a digital-only subscription, allowing you to save on shipping charges — no more waiting for the mail to arrive! At Traders.com, subscribers will also find our two online-only publications, Working Money and Traders.com Advantage, plus our entire 30-year online archive of every article we’ve published since 1982, which you can browse by month and year or search by keyword, author, or subject. NON-SUBSCRIBERS
SUBSCRIBERS
✔
✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔
Stocks & Commodities Magazine Stocks & Commodities Magazine, delivered
✔
Basic access to Traders.com Full access to Traders.com, including: Complete access to Working Money Complete access to Traders.com Advantage Complete issues of S&C Magazine in digital format (PDF) for downloading Complete 30–year S&C archive (PDF) Article code to copy & paste Optimized trading systems
6
$ 99 per issue
Subscribers pay less, get more!
$
5 per month!*
✔
Visit Traders.com today to see how we can help you trade better and smarter. Explore the benefits of a subscription to Stocks & Commodities Magazine at Traders.com/tour/tour.html. Online: www.Traders.com 206-938-0570 • Fax: 206-938-1307 Email:
[email protected] Toll free: 1-800-832-4642 Join us on Facebook at www.facebook.com/STOCKSandCOMMODITIES
*Based on the 5-year subscription rate of $299.99. Shipping extra for foreign addresses.
Follow us on Twitter @STOCKSandCOMM
May 2015 • Volume 33, Number 6
Opening Position
The Traders’ MagazineTM
Editor in Chief Jack K. Hutson Editor Jayanthi Gopalakrishnan Production Manager Karen E. Wasserman Art Director Christine Morrison Graphic Designer Wayne Shaw Webmaster Han J. Kim Contributing Editors John Ehlers, Anthony W. Warren, Ph.D. Contributing Writers Don Bright, Thomas Bulkowski, Martin Pring, Barbara Star, Markos Katsanos
OFFICE OF THE Publisher
Publisher Jack K. Hutson Industrial Engineer Jason K. Hutson Project Engineer Sean M. Moore Controller Mary K. Hutson
Advertising Sales
4757 California Ave. S.W. Seattle, WA 98116-4499 1 206 938-0570 Fax 1 206 938-1307
[email protected] National Sales Manager, Classified & Web Sales Edward W. Schramm Advertising Sales Summer Davis
Circulation
Subscription & Order Service 1 800 832-4642 1 206 938-0570 Fax 1 206 938-1307
[email protected] Subscription Manager Sean M. Moore Subscription Sales Carmen Hale
Website
http://www.traders.com
Staff members may be emailed through the Internet using first initial plus last name plus @traders.com
Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by Technical Analysis, Inc. for users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of $1.00 per copy, plus 50¢ per page is paid directly to CCC, 222 Rosewood Drive, Danvers, MA 01923. Online: http://www.copyright.com. For those organiz ations that have been granted a photocopy license by CCC, a separate system of payment has been arranged. The fee code for users of the Transactional Reporting Service is: 0738-3355/2015 $1.00 + 0.50. Subscriptions: USA: one year (13 issues) $89.99; Magazines shipped outside the US require additional postage as follows: Canada, US$15 per year; Europe, US$25.50 per year; all other countries US$39 per year. Single copies of most past issues from the current year are available prepaid at $8 per copy. Prior years are available in book format (without ads) or digitally from www.traders. com. USA funds only. Washington state residents add sales tax for their locale. VISA, MasterCard, AmEx, and Discover accepted.Subscription orders: 1 800 832-4642 or 1 206 938-0570. Technical Analysis of Stocks & Commodities™, The Traders’ Magazine™, is prepared from information believed to be reliable but not guaranteed by us without further verification, and does not purport to be complete. Opinions expressed are subject to revision without notification. We are not offering to buy or sell securities or commodities discussed. Technical Analysis Inc., one or more of its officers, and authors may have a position in the securities discussed herein. The names of products and services presented in this magazine are used only in an editorial fashion, and to the benefit of the trademark owner, with no intention of infringing on trademark rights.
W
ill they, or won’t they, and if so, when? All eyes were on the policy statement released by the Fed on March 18. The takeaway from it was that the word “patient” was not used, implying that there is a chance that we will see a rate hike this year. And rate hikes means that the economy is improving, or that is what we are led to believe. Immediately after the Fed released their statement suggesting they may start raising interest rates sometime in 2015, it was almost as if there was a huge sigh of relief. Stocks moved higher, commodities moved higher, treasuries moved lower, and the US dollar moved lower.
If
you take a moment to analyze what really moves the markets, you’ll find that it’s a lot more than interest rates. Fundamental analysts focus on valuations such as price/earnings ratios, debt-to-equity ratios, EBITDA, and so on, but as technical analysts, we need to look at indicators such as market breadth, advances over declines, and investor sentiment using variables such as TRIN, TICK, and VIX. Keeping an eye on these variables can be used as a barometer to gauge the strength of the market and whether investors are risk averse. Any divergence between the movement of the broader markets and these barometers or a lack of confirmation from all these variables should be considered as a sign to tread cautiously. At the moment there seems to be too much uncertainty in the markets together with too much optimism. The two don’t mix well and that’s a cause for concern. We’re too focused on the central banks and placing importance on their choice of words. First, it was irrational exuberance, then patience, and now reasonably confident. According to the recent statement released by the Fed, it’s inflation, unemployment, and wages that will indicate how well the economy is doing and ultimately be the deciding factor for raising interest rates. But other indicators such as credit spreads, treasury yields, performance of commodities, and performance of the manufacturing/service sector give much earlier signals of the underlying economic fundamentals. But getting a real gauge of the economy is no easy task, especially when it’s been stimulated by funds from the central banks. I seriously doubt we’ll be seeing any interest rate hikes in the next FOMC meeting. We have to patiently wait to see when and if it will happen this year.
6 • May 2015 • Technical Analysis of Stocks & Commodities
Jayanthi Gopalakrishnan, Editor
Miami Downtown Richard Cavalleri/Shutterstock
EDITORIAL
[email protected]
The editors of S&C invite readers to submit their opinions and information on subjects relating to technical analysis and this magazine. This column is our means of communication with our readers. Is there something you would like to know more (or less) about? Tell us about it. Without a source of new ideas and subjects coming from our readers, this magazine would not exist. Email your correspondence to
[email protected] or address your correspondence to: Editor, Stocks & Commodities, 4757 California Ave. SW, Seattle, WA 98116-4499. All letters become the property of Technical Analysis, Inc. Letter-writers must include their full name and address for verification. Letters may be edited for length or clarity. The opinions expressed in this column do not necessarily represent those of the magazine.—Editor
CANDLESTICKS, CONDENSED Editor, I just read Dave Cline’s February 2015 article in Stocks & Commodities , “Candlesticks, Condensed,” and found it quite interesting. I had never thought of using the approach he describes. It’s a nice way to create pattern signatures. I took a course through Coursera on quantitative analysis by Tucker Balch and used Python during the course. One of the exercises was to analyze historical events based on price movements. Adding a candlestick signature could be used as an extension to this. I have also done some basic simulations of crossing EMAs in Excel. When I went to download the Python code associated with Cline’s article from Traders.com, I read that Cline had also done some work in Excel and wondered if he is willing to share a version of the Excel file referred to there. Morley Author Dave Cline replies: I can provide the Excel spreadsheet, although it’s not much, really. I can also provide the Access.MDB file into which I dumped the Excel data for grouping/ consolidation. You can get these files from the following link: https://dl.dropboxusercontent. com/u/29771494/Finance/ CandlesticksCondensed.xlsx You’ll find that some of the problems with the compressed candles approach are:
1. You really need two, three, or four sequential patterns to make the results discrete enough 2. Then you need tons of data to build the dataset to get enough patterns to make their numbers significant. And those two items fight each other. Fortunately, on Quantopian.com, you can use 200 instruments going back to about 2003 to build the source database of patterns. Thanks for your inquiry. MORE ON CONDENSED CANDLES Editor, I was interested in the article by Dave Cline in the February 2015 issue (“Candlesticks, Condensed”), so I decided to see if I could replicate his work. A summary of results follows and the relevant spreadsheet is attached [not shown]. I would be interested in Cline’s comments. I have also written code in PowerBASIC. I analyzed 53 years (1962–2014) of S&P 500 index weekly candlesticks with reference to the past 10 weeks, and each candle was assigned a three-letter code for the body, upper shadow, and lower shadow, as follows:
“A” means +2 SD (standard deviations) “B” means +1 SD “C” means normal “D” means -1 SD “E” means -2 SD
My analysis showed minimal predictive significance, as SD was usually wider than the gap between zero and the percentage gain for the following week. May 2015
The exceptions were limited to patterns that were only seen once over the reference interval (SD = 0) and the following: BDE seen twice CCE seen twice CEA seen twice DEB seen twice EEC seen 5 times EBD seen 3 times
1.43% SD=0.01 1.02% SD=0.64 -1.25% SD=0.3 -1.33% SD=0.03 -1.62% SD=1.34 -1.79% SD=1.37
My conclusion: Single weekly candlesticks were of no value in predicting the following week’s market action for the S&P 500 index. John Rathbun Asheville, NC Author Dave Cline replies: Interesting translation into a StdDevbased variation. The compression technique already is fairly lossy; are you sure you’re not losing any additional information by this technique? Also, you’ve got 2,751 samples, which I would suggest is a somewhat limited set to work with. As you can probably surmise, and I think I alluded to this in my article, single candles have nearly zero predictive information within them. But in sequences, they may provide small probability benefits. Unfortunately, you need tens or hundreds of thousands of sequence samples to build up statistically significant sets. So I would suggest building pairs of candles as patterns. For instance, what is the average return on the CDC-CBA sequence (if it exists)? When I built and tested this mechanism, I ran 10 years of daily data on all the S&P securities through it. I used three-candle sequences. I've also tried two years of hourly data of the same. Within those tests, I could find significant sequences that tested out-of-sample to about one half of their in-sample return. So I think there's value, if tiny and hard to see, in the technique. To me, its just one more layer of probability to add to a list of filters when scanning thousands of instruments for possible trades. Thanks for reading and going through the trouble of testing the theory. It means a lot to me. • Technical Analysis of Stocks & Commodities • 7
Random With Memory
Schrödinger’s Cat What information is contained in market data? Can you develop an indicator or trading system that can extract this information to provide an edge in trading? Here’s a look.
T
by John F. Ehlers
he purpose of technical analysis is to discern what information is contained in market data and, if you are clever enough, to develop an indicator or trading system that extracts this information to provide an edge in trading. On the other hand, there are those who believe in the efficient market hypothesis: that all the information about the markets is known and the effects are purely random due to the law of large numbers of traders. The discussion goes downhill from there. One of my favorite theoretical descriptions of market activity is the drunkard’s walk. When the random variable is position, the partial differential equation solution is called the diffusion equation, and it describes random motion like a particle of smoke in a smoke plume. When the random variable is momentum, then the partial differential equation solution is called the wave equation. Taken together, the drunkard’s walk describes physical phenomena like the meandering of a river, which can be random (trending) or cyclical. Unfortunately, there is no closed solution for the differential equations that can lead to an indicator, because they require
8 • May 2015 • Technical Analysis of Stocks & Commodities
Measuring synthe-
sized market data Synthesizing market data is one thing, and measuring its characteristics is quite another. The problem is similar to that of the “Schrödinger’s cat” thought experiment: Merely measuring the outcome determines the outcome itself. Here’s the problem: When
PATRICK KELLEY
boundary value solutions and there is no definable boundary. In another physical area, Peter Swerling noted that the radar echoes returned from flying aircraft were noise-like. The echoes would vary from pulse to pulse and from one antenna sweep to another. The explanation is that there was a total average power returned, but the total power was the summation of components that were bounced off the fuselage, wings, rudder, and so on, and the changing aspect of the aircraft caused the summation of these components to look like noise. When building deception jammers for radars, I simulated the Swerling noise by using the received radar pulse plus an exponential moving average (EMA) of past pulses. This jamming signal was a remarkably good replica of the real radar echo. This kind signal is called random with memory, and it’s consistent with other phenomena described by the Hurst coefficient. Synthesizing market data using a random number generator and an EMA is simple to do and could be an interesting way to examine the nature of market data. Knowing the nature of the data can therefore lead to the generation of an indicator that possibly can give us a trading edge.
TRADESTATION
the market is modeled as a random variable with memory, the memory is provided by a filter such as an EMA. However, when measuring the frequency content of market data with any technique such as a Fourier transform or a contiguous bank of bandpass filters, they all have filters with memory as part of the analysis technique. Thus, FIGURE 1: MEASURED SPECTRUM OF THE SPDR S&P 500 OVER THE LAST YEAR. The dominant cycle period was between 20 and 25 measuring a truly ran- bars in the fall of 2012, was on the order of 15 bars during most of the uptrend, and was an ill-defined longer cycle period most recently. dom set of data would involve the memory being provided by the measurement daily data). The spectrum shows that there is not much cyclic technique, and the entire process would become self-fulfilling. activity, and the dominant cycle is mostly near a 10-bar cycle Measuring the frequency content of synthesized data must due to aliasing noise. avoid the use of filters. The next experiment is to see the effects of adding memory Interfering with the synthesis of market data is minimized to the random data. For example, Figure 3 shows the data through the use of an autocorrelation periodogram. This and spectrum when the memory low-pass filter has a critiprocess first creates the autocorrelation of the data, a process cal period of 20 bars. Not unexpectedly, the data is much that is basically without filters. Then, a Fourier transform of smoother than in Figure 2. Also, the dominant cycle period the autocorrelation function is taken to extract the frequency in the measured spectrum is near a 20-bar period most of content of the data. On a related note, the autocorrelation the time. periodogram is now my preferred method of frequency meaContinuing with the experiment, the memory of the low-pass surement of market data because it mitigates the effects of filter is changed to have a critical period of 40 bars (Figure 4). spectral dilation. In this case, the data is smoother across the graph. Further, Figure 1 shows what the measured spectrum of real market the measured dominant cycle period has increased. data looks like. The data is approximately one year’s worth of daily bars of the SPDR S&P 500 (SPY). The measured So what does it all mean? spectrum is shown below the price bars as a heatmap. The Dealing with random data is tricky because you can never strength of the cycle amplitude is shown in colors ranging reproduce your results. The best you can do is infer characfrom white hot through red hot to ice cold. The period of the teristics from your measurements. The first observation is that measured cycles is indicated on the vertical scale from zero market cycles are ephemeral—they come and go, and the cycle through 48-bar periods. Figure 1 shows that the dominant periods of the dominant cycle can often change rapidly. cycle period was between 20 and 25 bars in the fall of 2012; was on the order of 15 bars during most of the uptrend; and was an ill-defined longer cycle period most recently. Now that you are familiar with displays of market spectra, let’s turn your attention to the measurement of purely random data with no memory, as shown in Figure 2. The random data is shown as the green ragged line over approximately 250 samples FIGURE 2: MEASURED SPECTRUM OF PURELY RANDOM DATA WITH NO MEMORY. The spectrum shows that there is not much (essentially one year of cyclic activity, and the dominant cycle is mostly near a 10-bar cycle due to aliasing noise. May 2015
• Technical Analysis of Stocks & Commodities • 9
FIGURE 3: MEASURED SPECTRUM OF RANDOM DATA WITH MEMORY HAVING A 20-BAR CRITICAL PERIOD. The dominant cycle period in the measured spectrum is near a 20-bar period most of the time.
Market cycles are ephemeral — they come and go, and the cycle periods of the dominant cycle can often change rapidly. Synthesizing market data as random with memory does gain some credibility because the resulting measured spectra look similar to real market data. Further, the characteristics of the synthesized data can be controlled simply by varying the critical period of the memory component of the synthesized data. Credible replicas of market data can therefore be created simply by making the critical period of the memory
time variable across the chart. But most of all, you can gain the edge in your trading that you sought in the first place. Knowing that market cycles are ephemeral, you can quickly jump on them with predictive filters when they appear. You can get an idea of how this works by looking at the trade setup analyzer on www.StockSpotter. com. A trade setup occurs when the MESA cycle indicator is at or near a cycle trough and the MESA momentum indicator is declining or is at a minimum. S&C Contributing Editor John Ehlers is a pioneer in the use of cycles and DSP techniques in technical analysis. He is president of MESA Software. MESASoftware.com offers the MESA Phasor and MESA intraday futures strategies. He is also the chief scientist for StockSpotter.com, which offers stock trading signals based on indicators and statistical techniques.
Further
reading Ehlers, John [2013]. Cycle Analytics For Traders, Wiley & Sons. [2014]. “The Quotient Tra nsform,” Technical Analysis of Stocks & C ommodities, Volume 32: August. ‡TradeStation, ‡StockSpotter.com FIGURE 4: MEASURED SPECTRUM OF RANDOM DATA WITH MEMORY HAVING A 40-BAR PERIOD. The data is smoother across the graph and the measured dominant cycle period has increased.
10 • May 2015 • Technical Analysis of Stocks & Commodities
‡See Editorial Resource Index
Featuring
Get into futures. No DeLorean necessary. Learn from our veteran futures professionals. Have the confidence to take on futures with thinkorswim.® Explore strategies by practicing with virtual paperMoney.® Assess portfolio risk in real time relative to indexes. And get access to real-time information on commodity and futures markets right from the pit. So you can get into futures, minus the flaming tire tracks. Open an account with TD Ameritrade at tdameritrade.com/thinkorswim
Futures and futures options trading is speculative and is not suitable for all investors. Futures accounts are not protected by the Securities Investor Protection Corporation (SIPC). Futures and futures options trading services provided by TD Ameritrade Futures & Forex LLC. Trading privileges subject to review and approval. Not all clients will qualify. paperMoney® application for educational purposes only. TD Ameritrade, Inc., member FINRA/SIPC. TD Ameritrade is a trademark jointly owned by TD Ameritrade IP Company, Inc. and The Toronto-Dominion Bank. © 2015 TD Ameritrade IP Company, Inc. All rights reserved. Used with permission.
12 • May 2015 • Technical Analysis of Stocks & Commodities
INDICATORS
For Your Digital Eyes Only
Filtering Price Movement Here is an alternative to the classic zigzag indicator, which may prove useful to visual technical analysts and chart pattern researchers.
Applied Micro Devices (daily)
15.0 14.5 14.0 13.5
W
JOSE CRUZ
15.5
13.0
hen there is need for algorithmic iden12.5 12.0 tification of price swings in a chart, 11.5 there is a word that always comes to 11.0 10.5 mind for technical analysts: zigzag. The zigzag 10.0 9.5 indicator is based on the concept from Arthur 9.0 Merill’s 1977 book Filtered Waves, Basic 8.5 8.0 Theory: A Tool For Stock Market Analysis. It 7.5 filters price movements below a cutoff level, 7.0 6.5 that is, a threshold. The threshold is either in 6.0 Zigzag (20%) point terms or in percentage terms. If you were, 5.5 5.0 for example, using a threshold of x points, the Zigzag (20%) 4.5 zigzag would disregard all price movements T1 T2 4.0 3.5 less than x points. If, on the other hand, you May Jun Jul Aug Sep Oct Nov 2014 Feb Mar Apr May used a threshold of x percent, the zigzag would Figure 1: the dynamic nature of the zigzag’s last legs. The red zigzag in this daily chart disregard all price movements of magnitude of Applied Micro Devices, Inc. (AMD) is based on a percentage threshold of 20% and it uses data up to less than x percent. When plotted, the zigzag date T1. The blue zigzag is based again on the 20% percentage threshold but it uses data up to date T2. In other words, the red zigzag is a snapshot from the history of the blue one. Notice how the last is shown as a crooked line connecting peaks two legs of the red zigzag changed when price information from T1 and later were taken into account to the blue zigzag. This chart was created in MetaStock, which plots the zigzag in a way such that and troughs. The line segments of the zigzag create its last two legs are dynamic. In other versions of zigzag, only the last legs are dynamic. are commonly referred to as its legs. Notwithstanding that the zigzag identifies promi- are dynamic and usually change significantly as new nent peaks and troughs, it doesn’t filter the price data comes in. Consequently, the historical values swings the same way a technician’s eye would. In this of the zigzag are based on hindsight. So if you’re article, I will introduce you to a more natural way of using the zigzag in the same way that you use other filtering the price, which is accomplished via what are classic technical indicators such as moving averages, called perceptually important points. This alternative relative strength index (RSI), stochastics, and so to the classic zigzag indicator is closer to the way a on, then zigzag won’t be of much use. However, it can be useful if it’s used to identify prominent price human perceives the movement of price. swings on a chart. Simply put, there is no way to know when the current price movement will pass the Limitations of the zigzag The zigzag is accused of a serious drawback: Its last cutoff threshold before that happens (see Figure 1 for two legs (or, depending on the software, its last leg) an example). In effect, the zigzag is a static tool that by Giorgos E. Siligardos May 2015
• Technical Analysis of Stocks & Commodities • 13
2
1
Figure 2: not all points identified by the zigzag are visually prominent. The zigzag always tries to find and accent prominent price swings based on how high or low these swings go, but this makes it quite stiff. In this iconic example, the zigzag would disregard point 2 just because point 1 is a bit lower. From a visual perspective, however, point 2 was more important than point 1 since it was the pivot that sparked a swift and strong uptrend.
tries to mimic—often in a clumsy way—the eye of the analyst when it looks at a snapshot of a chart. It does so from a more mathematically rigid point of view, concentrating on the major swings of price (as defined by the cutoff threshold). It must come as no surprise then that for the chart pattern
35 34 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4
1 CBS Corp. CL B (daily)
2
3
Zigzag (20%) J A S O N D 2007
A M J J A S O N D 2008
4 5
A M J J A S O N D 2009
analyst, the dynamic nature of the zigzag’s last legs is not a drawback but a merit. For example, in his November 2003 Stocks & Commodities article “The Zigzag Trend Indicator,” Spyros Raftopoulos introduced an interesting binary indicator that he called zigzag trend. The zigzag trend is essentially the zigzag without its dynamic feature, so its strong point is that it can be used and treated the same way as other common trend indicators such as the MACD, with the additional benefit of a low number of whipsaws. From a pattern analyst’s standpoint, however, the absence of dynamic parts makes it completely incapable of identifying visually prominent peaks and troughs in a snapshot of a chart. A more substantial drawback of the zigzag as a tool to represent a chartist’s perception could be its dependence on the threshold parameter. In other words, you can’t use the same cutoff threshold for all charts. A 20% threshold for long-term daily charts of stocks does a pretty good job most of the time, but it might be inefficient for short-term daily charts. So the analyst must first see the chart and then define the threshold that will give the zigzag the opportunity to identify the major swings. That initially negates the usefulness of the zigzag as a representative of the human eye when there is need for identification of major swings in thousands of charts. This is not a serious drawback, however, since there is a simple (albeit not perfect) workaround: You can take the range of values in a chart (highest value minus lowest value) and then define the threshold as a percentage of that range. So what are the essential limitations of the zigzag from a chartist’s point of view? One limitation is that it focuses exclusively on prominent price swings (peak to trough and trough to peak). More precisely, although it indeed identifies meaningful pivots in price, it often misses other pivots that are even more important regarding their role in the visual comprehension of the movement of the price (Figure 2). Also, its bias toward only price swings makes it incapable of perceiving special cases where connection of peaks to peaks or troughs to troughs describes the price behavior in a better way (see Figure 3). Another important limitation of the zigzag has to do with the way it summarizes and ranks information on a chart. More precisely, you can’t force the zigzag to summarize the price action into a specific number of swings. For example, you can’t tell the zigzag to filter the price action and condense it into, say, four swings (legs). You will know the total number of the zigzag’s legs only after it has filtered the price. A M J
Figure 3: the zigzag always connects peaks with troughs. The zigzag has a unilateral way to filter price movements. It always connects peaks with troughs. This means that it is blind regarding changes in the strength of directional movements and so misses important information with respect to the visual perspective of a price trend. In this daily chart of CBS Corp., the 20% threshold zigzag (in blue color) is unable to see the visual importance of points 2 and 3 although they clearly mark changes in the severity of the downtrend. It considered point 4 as significant, but that is not visually prominent. The pink crooked line gives a much better sight of the price movement from point 1 to point 5.
14 • May 2015 • Technical Analysis of Stocks & Commodities
Meet the PIPs method
An alternative method of filtering price fluctuations is one that is based on the idea of perceptually important points, or PIPs. While roots of this method trace back to 1973, it was mainly introduced in 2001 by F.L. Chung et al. in their academic research paper “Flexible Time Series Pattern Matching Based On Perceptually Important Points.” The PIPs method makes it feasible to construct a modified version of the classic zigzag indicator that will
Euclidian distance
Vertical distance
Perpendicular distance
X X X d2 Z Z Z overcome the limitations I mentioned earlier because its filtering process is dx(Y,Z) d1 dx(Y,Z) much closer to the way a technician’s eye scans a chart. This doesn’t mean dx(Y,Z) = d1+d2 that this new method should wholly Y Y Y replace the classic zigzag. It is just a different method serving a different purpose. The PIPs method is more appropriate for representing price Figure 4: the tHree flavors of distance of one point from A PAIr of two points. Three ways to define the distance of a point X from a pair of points Y, Z have been proposed in the literature: The Euclidian, the vertical, and the movement from a visual standpoint. perpendicular. In brief, while the zigzag starts from the left of a chart and creates legs as it moves to the right, the PIPs method identifies important points based B on a holistic approach: All price data is indirectly taken into account Identifying the third PIP for the identification of each and every leg.
{
{
The concept of distance
Before diving into the details of PIPs, it is necessary to define the concept of the distance of one point with respect to two other points. Let X, Y, and Z be three points in a time–price chart in this order: Y, then X, then Z. In their 2008 paper “Representing Financial Time Series Based On Data Point Importance,” Tak-chung Fu et al. proposed three ways to define the concept of distance dX(Y,Z) of X from points Y and Z: n
n
n
Euclidian distance: dX(Y,Z) is defined as the distance of X from Y plus the distance of X from Z. Vertical distance: If ε is the straight line connecting the points Y and Z, then dX(Y,Z) is defined as the vertical distance of X from ε.
C
A
B Identifying the fourth PIP
D
Perpendicular distance: If ε is again the straight line that connects the points Y and Z, then dX(Y,Z) is defined as the perpendicular distance of X from ε.
In Figure 4 you can see pictorial examples for these three flavors of distance.
C
A
Identifying the PIPs
Consider a set of points in a time–price chart that are derived by the values of an indicator such as the MACD or the closing price of a stock. A point from this set will be considered perceptually important when it dominates all other points in terms of importance in the perception of the visual shape that these points create. That’s a loose definition, I know, so let me define the PIPs via a formal inductive procedure using the vertical kind of distance (refer to Figure 5 for a visual aid). Step 1: The first two PIPs are the first and last points in the chart. Name them A and B, respectively. I call these PIPs marginal for obvious reasons. All the other PIPs will be called internal. Step 2: To find the third PIP, calculate the vertical distances of all points of the set from the couple A, B (that is, calculate all dX(A,B) where X runs all points of the set). The point X, which produces the maximum distance, is the third PIP. Let
B Identifying the fifth PIP
E
D
A
C
Figure 5: identifying PERCEPTUALLY IMPORTANT POINTS (PIPs) USING THE VERTICAL DISTANCE. The first two PIPs are the first and last points (A and B). From there on, to designate a point as perceptually important, you go through a procedure that takes into account all price data in the chart. More precisely, you go through calculations of vertical distances involving all data in the chart and lines connecting previously identified PIPs. May 2015
• Technical Analysis of Stocks & Commodities • 15
this point be C. There are now three PIPs that appear in this time order in the chart: ACB. Step 3: Using the same previous idea, run through all set points between A and C and calculate their vertical distances from the couple A, C. Also run through all set points between C and B and calculate their vertical distances from the couple C, B. The maximum distance found from these two runs marks point D, which is the fourth PIP. Step 4: Say that D is between A and C. For the fifth PIP you make three runs of vertical distance calculations: one from A to D, one from D to C, and one from C to B. The maximum distance found marks the fifth PIP (E in Figure 5). Next steps: You can repeat this procedure to find as many PIPs as you like (a new PIP for every step). The procedure stops when you have identified your desired number of PIPs or when the maximum distance in a step is zero (as this would mean that no additional information is gained by identifying new PIPs). Of course, there is always a natural limit to the number of PIPs you can identify—and that limit is the total number of points in the chart.
and, voilà—you have a new zigzag-like indicator. (Note that the number of legs equals the number of PIPs minus one.) I call this indicator zzTOP. The “zz” part of the name comes from it being a generalized kind of zigzag and—what can I say—the “TOP” part comes because I am listening to ZZ Top’s hit song “Legs” as I write this article. The zzTOP indicator requires three arguments (parameters). These are: indicator, LegsNo, and scale. Let’s look at them in detail. Indicator Unlike the classic zigzag, the zzTOP doesn’t rely on cutoff thresholds, so it can be directly applied successfully to any kind of indicator. The indicator parameter is therefore the indicator upon which you want the zzTOP to be applied. It can be the closing price line, the MACD, the RSI, or any indicator you can think of.
LegsNo This is a numeric parameter (a positive integer greater than or equal to 1) that defines the total number of legs you want the zzTOP indicator to have. The number of PIPs equals this number plus 1. For example, a value of 20 for this parameter indicates that you want the zzTOP to have exactly 20 legs (or equivalently, you are interested in 21 PIPs).
As you probably noticed, the inductive procedure used to identify the PIPs has an additional benefit: The PIPs are automatically ranked in descending order of perceptual importance. The mathematically inclined, however, might have already found a possible problem with this procedure: What if there is not one and only one maximum distance among the vertical distances you calculate for a step? This is rare but it can happen. In this occasion, there will be more than one finalist for the next PIP designation, so you either designate all of them as PIPs or, when you need to select only one of them because you want only one PIP, you need a selection convention regarding which one to designate as the next PIP. As a simple solution for the second case, I opt for the finalist, which lies in the right-most side of the chart. In other words, I focus on the most recent data. You could use other methods of selection, but I believe this is the simplest and most efficient for our purpose. A similar procedure could be used to identify PIPs using Euclidian or perpendicular distance. But what is the most appropriate distance to use? A study of various examples shows that from a visual point of view, the Euclidian distance identifies terrible PIPs. Further, the vertical and perpendicular distances produce exactly the same PIPs in most of the real cases. In effect, you can use only the vertical distance and disregard the other two. The indicator I will present uses the vertical distance and the selection convention discussed earlier.
The zzTOP indicator
Now that you know how to calculate PIPs in an indicator’s plot, you can connect them using straight line segments to create legs 16 • May 2015 • Technical Analysis of Stocks & Commodities
Ball Corp. (daily)
50
zzTop (Close,5,L)
50
zzTop (Close,20,L)
1980
1990
2000
2010
Figure 6: zztop performance in the daily chart of ball corp. (BLL). The zzTOP indicator is a nice way to approximate the price action via a predefined number of linear legs. The more legs that are used, the closer the approximation.
Scale This parameter refers to the scaling of the y-axis of the chart, and it has a significant effect on the performance of the zzTOP. The scale parameter can take two values: “A” (arithmetic) and “L” (logarithmic). If you want the zzTOP to filter the movements of the indicator parameter as seen in an arithmetic scale, then you set this parameter to “A.” This instructs the zzTOP indicator to apply its PIPs identification algorithm to the indicator itself. If, however, the indicator is positive and you want the zzTOP to filter its movements as seen in a semilogarithmic scale (in such a scale, the y-axis is logarithmically scaled, whereas the x axis is arithmetically scaled), then you set this parameter to “L.” This latter case is equivalent to first taking the natural logarithm of the indicator, then applying the zzTOP with a scale parameter of “A,” and then applying the exp() function in the result. As an example, zzTOP(close, 30, L) refers to the zzTOP indicator applied on the semilogarithmic chart of the closing price of a security demanding that the zzTOP must have exactly 30 legs. Similarly, zzTOP(MACD, 20,A) refers to the zzTOP applied on an arithmetic chart of MACD and demanding that the zzTOP must have exactly 20 legs. It is important to note again that while the zigzag scans the price series from left to right using a number (the threshold) to classify a price swing as important, the zzTOP uses information from all loaded data in a chart every time it identifies a new internal PIP. This is invaluable from the point of visual comprehension of a chart, but it comes at a price: The zzTOP is much more prone to changing many of its legs when new price data is added to the chart.
Chart examples
It is now time to go through some chart examples. In Figure 6 you can see how the zzTOP(close,5,L) and zzTOP(close,20,L) perform in the same chart. The former scans all prices shown in the chart, finds six PIPS, and summarizes the price action May 2015
• Technical Analysis of Stocks & Commodities • 17
ASML (daily)
100 50
zzTop (Close,20,L) 100 50
Zigzag (20%) 1995 1996
1997
1998
1999
2000 2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Figure 7: zztop and zigzag vis À vis. The upper and lower daily charts of ASML Holding (ASML) are the same. The zzTOP(close,20,L) indicator is overlaid in the upper chart, whereas in the lower chart, the zigzag identifies peaks & troughs based on a percentage threshold of 20%. From a visual standpoint, the zzTOP indicator can effectively render the essentials of the price movement using much fewer legs than the zigzag (see its performance between late 1999 and late 2001). This is mainly because of two reasons: First, it is allowed to connect peaks to peaks and troughs to troughs, and second, it takes into account all price data for the calculation of each leg. The zigzag on the other hand doesn’t look at all price data every time it creates a leg. It processes the data strictly from left to right and it can only change its last two legs during the identification procedure.
into only five legs. The latter finds 15 more PIPs and summarizes the price action into 20 legs. Note that the zzTOP doesn’t have to connect only peaks with troughs. It can also connect peaks to peaks or troughs to troughs and thus it is more flexible in summarizing and expressing the price movement quirks. In this regard, the choice of “zz” in the name zzTOP is perfectly suited because the zzTOP is not limited to only zigzags—it can do zigzigs and zagzags too. In Figure 7 you can see how the zzTOP(close,20,L) differs from the classic zigzag in-
For the chart pattern analyst, the dynamic nature of the zigzag’s last legs is not a drawback but a merit.
Baxter Intl. Inc. (daily) zzTop (Close,20,L)
80 70 60 50 40 30 20
zzTop (Close,20,L)
10
Period 1 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Period 2 Figure 8: zztop usually changes dramatically when new data IS added. That zzTOP identifies an internal PIP taking into account all previously identified PIPs, and that the first and last prices in the chart are always the first two PIPs means that all internal PIPs (and consequently all legs) are indirectly affected by the first and last prices of the chart. So as new data is added to a chart, all the legs of the zzTOP face the danger of change. As more and more data is added, all of its legs will finally change, since the number of legs is constant. This feature of the zzTOP is clearly seen in the daily chart of Baxter International Inc. (BAX), where the zzTOP(close,20,L) is applied to price data for two periods. The red zzTOP is applied to price data for period 1 and the blue zzTOP is applied to price data for period 2.
18 • May 2015 • Technical Analysis of Stocks & Commodities
dicator with a percentage threshold of 20%. Note especially the period from the end of 1999 until the end of 2001. The zzTOP clearly depicts the price movement in a better way than the zigzag does in terms of visual clarity, using just a few legs. Figure 8 shows how the zzTOP may change when you put new data in a chart. Period 2 starts at the beginning of 1983 and ends at the beginning of 2000, whereas period 1 starts at the beginning of 1983 and ends near the summer of 2009. The zzTOP indicator in blue is applied in period 2 only (that is, it doesn’t look outside period 2) and identifies 20 legs for that period. The zzTOP indicator in red is applied in period 1 and summarizes the price action into 20 legs for the entire period. Both zzTOPs have the same parameters except for the time period upon which they are applied. It is obvious that new data can have a significant effect on the performance of zzTOP not only because of the restriction in the number of legs it is allowed to present but also because its algorithm identifies all internal PIPs, starting from the marginal ones (the first and last prices in the chart). In effect, all internal PIPs—and consequently all legs—are affected by the first and last prices in the chart. In Figure 9 you can see why the scale parameter is important. In the top chart you see the weekly price of Caterpillar Inc. (CAT) with the 20-leg zzTOP based on the closing price using arithmetic as its scale parameter. In the lower chart you see the same weekly chart of CAT with the same 20-leg zzTOP indicator, but this time, the scale parameter is logarithmic. The upper chart is arithmetic, whereas the lower one is semilogarithmic. It is clear that the scale parameter is there to ensure that the zzTOP “sees” the chart the same way a chartist would do with his eyes. In the upper chart (the arithmetic one), the price movement before the year 2000 is seen as almost horizontal by the human eye.
Caterpillar Inc. (weekly)
That’s because after 2000, the prices advanced significantly. In effect, the swings of the price after 2000 overshadow those before 2000 from an arithmetic perspective and the arithmetic-scale zzTOP correctly focuses on the price swings after 2000 because that’s what a human eye would naturally do. In the lower chart, though, the semilogarithmic scale makes it possible to see things from a percentage perspective, so the price swings before 2000 are visually more prominent now. The logarithmic-scale zzTOP in the lower chart correctly identifies the 20 most noticeable price swings the same way a human eye would. Most chartists use semilogarithmic charts to plot the prices of trading instruments, so an arithmetic-scale zzTOP is practically useless when applied to the price charts (especially the long-term ones). The charts of common technical indicators (such as stochastics, MACD, and RSI) are nonetheless always arithmetic, so the ability of zzTOP to adapt to scale differences can be useful. In Figure 10 you can see how the arithmeticscale zzTOP performs in a weekly chart of Archer Daniels Midland Co. (ADM).
Automation
Arithmetic scale zzTop (Close,20,A)
90 80 70 60 50 40 30 20 10 0 100 50
Semilogarithmic scale zzTop (Close,20,L) 1990
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Figure 9: arithmetic vS. logarithmic scale. The scale parameter of the zzTOP determines the way the zzTOP “sees” the price. The “A” (arithmetic) scale parameter instructs the zzTOP to see the price from an arithmetically scaled y-axis whereas the “L” (logarithmic) scale parameter instructs it to see the price from a logarithmically scaled y-axis. The results can be strikingly different for these two cases as it is seen in this weekly chart of Caterpillar Inc. (CAT).
The zzTOP doesn’t rely on cutoff thresholds so it can be directly applied successfully to any kind of indicator. Archer Daniels Midland Co. (weekly) 50 40 30
zzTop (Close,20,L)
20 10 4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 -4.0 -5.0
MACD zzTop (MACD,20,A) 2000 2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
The zzTOP requires you to state how Figure 10: identifying the swings of macd in a weekly chart of archer daniels Midland co. (ADM). The zzTOP indicator performs pretty well when applied in indicators without the need to define filtering thresholds. many legs you are interested in. The opportunity to a priori define the number of legs gives you tremendous freedom, but sometimes closer the zzTOPauto line must be to the indicator plot and you may want the indicator to choose how many legs to identify thus the more legs will be needed. Consider, for example: based on a goodness of fit level that you desire. In other words, you might be interested in a hybrid between the zzTOP and the zzTOPauto(indicator,20,A) zigzag. This can be accomplished by requiring the zzTOP indicator to keep finding PIPs and to create legs up to a predefined proximity level (an equivalent to the threshold of the zigzag). and say that the highest value of the indicator is 200 and its lowest value is 40. The range of the indicator is therefore I named this automated version of zzTOP the zzTOPauto. The zzTOPauto indicator has the same indicator and scale R=200-40=160. Since the proximity parameter is 20, you are parameters as the zzTOP does, but instead of LegsNo, it has interested in the required number of legs such that the vertia proximity parameter. So zzTOPauto(close,10,L), for ex- cal distances between the values of the indicator and the legs ample, refers to the zzTOPauto applied to the closing price are less than 20% of 160 (which equals 32). In other words, a of a security on a semilogarithmic chart with a proximity of proximity of 20 means that you want the zzTOPauto to keep 10. Proximity is a positive number up to 100 and represents a finding PIPs and to keep creating legs up to the point where the percentage of the range of values of the indicator parameter. indicator’s values will not divert more than 20% of R from the Its purpose is to give the zzTOPauto a level of goodness of fit zzTOPauto’s plot. Of course, for logarithmic-scale zzTOPauto, you are interested in. Note that the lower the proximity, the the range of the indicator must be measured in a way that will May 2015
• Technical Analysis of Stocks & Commodities • 19
55 50 45 40 35 30 25
Boston Scientific (daily)
perform the backtesting, the typical technical analysis software loads 15 all historical data, then calculates the values of indicators, and then 10 uses these calculated values to simulate the backtesting. This is zzTOPauto (Close,10,L) fine for common indicators such as 5 MACD and RSI, but for dynamic 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 indicators like zigzag, zzTOP, and zzTOPauto (which change their Figure 11: performance of zztopauto with a proximity parameter of 10 in a daily chart of boston scientific inc. (BSX). The zzTOPauto indicator in this chart did a great job in outlining the price movements of BSX. Using historical values when new data a higher proximity parameter would result in fewer legs for the zzTOPauto, whereas a lower proximity parameter would result comes in), this approach produces in more legs. erroneously prettifying results. As a consequence, the zzTOP and take into account the visual idiosyncrasy of the semilogarithmic zzTOPauto indicators must not be used for backtesting in the charts. More precisely, for the logarithmic-scale zzTOPauto, the typical software program using the software’s built-in backtestrange of the indicator is measured using the logarithms of the ing feature. You can, however, use these indicators in a static indicator values instead of the values themselves. In Figure 11 fashion as a digital substitution for your eyes when you want you can see how the zzTOPauto (close,10,L) did a great job in your software to scan thousands of charts. outlining the price movements of Boston Scientific Inc. (BSX). Coding the zzTOP and zzTOPauto indicators requires some Using a higher proximity parameter would result in fewer legs time and effort. For software whose formula language lacks for the zzTOPauto, whereas a lower proximity parameter would looping capabilities (like MetaStock, for example), the zzTOP result in more legs. and zzTOPauto must be coded using a versatile programming language, embedded inside a dynamic link library (DLL) file, Coding and usage and then be called by the software as external functions from The correct way to perform backtesting is the DLL. To plot the zzTOP and zzTOPauto indicators in to recalculate the values of all indicators MetaStock, I created a DLL (named “zzTOPindicators.dll”) involved whenever a new bar is taken into that’s available for download from the Article Code area of account. But that would require too many www.traders.com, or from http://traders.com/files/zzTOPindicalculations. To decrease the time needed to cators.zip directly. In the sidebar “ZZTOP And ZZTOPauto Indicators In Metastock,” you can find information on how to download and use it. ZZTOP AND ZZTOPAUTO INDICATORS IN METASTOCK 20
Readers can download my “zzTOPindicators.dll” file as a .zip archive from http://traders.com/files/zzTOPindicators.zip or from the Article Code of the Technical Analysis of Stocks & Commodities website, www.traders.com. After downloading, you will need to expand the .zip file and place a copy of it in MetaStock’s external function DLLs folder (usually located at C:\Program Files\Equis\MetaStock\ External Function DLLs). The zzTOP and zzTOPauto indicators can be called by the following code: ExtFml( "zzTOPindicators.zzTOP",Indicator ,LegsNo ,Scale)
and ExtFml("zzTOPindicators.zzTOPauto",Indicator,Proximity,Scale)
respectively. For example, the code: ExtFml( "zzTOPindicators.zzTOPauto",CLOSE ,15 ,L )
calls the zzTOPauto indicator for the close price in a semilogarithmic scale with a proximity of 15. —G. Siligardos 20 • May 2015 • Technical Analysis of Stocks & Commodities
Rock & roll
The zigzag’s way of filtering fluctuations, although simple, is not always appropriate for capturing the visual representation of price behavior. The zzTOP and zzTOPauto indicators presented in this article offer an alternative way to transfer your visual perception to your software. Perhaps, if you go through thousands of charts, chances are you will encounter cases where the zzTOPs will miss a few points that your eye would consider as visually important; however, that would generally be rare. So if you are not pleased with the way the zigzag indicator perceives the price movements in a chart, then get ready to rock and let the zzTOPs do their magic. Giorgos Siligardos holds a PhD in mathematics and a market maker certificate in derivatives from the Athens Exchange. He is a financial software developer, coauthor of academic books in finance, and a frequent contributor to Technical Analysis of Stocks & Commodities magazine. He has also been a research and teaching fellow to the University of Crete as well as a teaching fellow to the Department of Finance and Insurance at the Technological Educational Institute of Crete for many years teaching math and financial courses and supervising masters dissertations. His academic website is http://www.tem.
NOTICE OF CLASS ACTION SETTLEMENT
uoc.gr/~siligard and his current views on the markets can be found in http://marketcalchas.blogspot.gr/. He may be reached at
[email protected]. The DLL file mentioned in this article is available from http://traders.com/files/ zzTOPindicators.zip as a downloadable zip archive as well as from the Subscriber Area at our website, www.Traders.com, in the Article Code area. See our Traders’ Tips section beginning on page 50 for commentary on implementation of Siligardos’ technique in various technical analysis programs. Accompanying program code can be found in the Traders’ Tips area at Traders.com.
Further reading
Chung, F.L., and TC Fu, R. Luk, and V. Ng [2001]. “Flexible time series pattern matching based on perceptually important points,” International Joint Conference On Artificial Intelligence Workshop On Learning From Temporal And Spatial Data (pp. 1–7). Douglas, D., and T. Peucker [1973]. “Algorithms For The Reduction Of The Number Of Points Required To Represent A Digitized Line Or Its Caricature,” The Canadian Cartographer, Vol. 10, No. 2, pp. 112–122. Fu, Tak-chung, and Fu-lai Chung, Robert Luk, and Chak-man Ng [2008]. ”Representing Financial Time Series Based On Data Point Importance,” Engineering Applications Of Artificial Intelligence, Vol. 21, Issue 2, pp. 277–300, March. Merrill, Arthur A. [1977]. Filtered Waves, Basic Theory: A Tool For Stock Market Theory, Technical Trends. Phetchanchai, Chawalsak, and Ali Selamat, Amjad Rehman, and Tanzila Saba [2010]. “Index Financial Time Series Based On Zigzag: Perceptually Important Points,” Journal Of Computer Science, Vol. 6, No. 12, pp 1,389–95. Raftopoulos, Spyros [2003]. “The Zigzag Trend Indicator,” Technical Analysis of Stocks & Commodities, Volume 21: November. ‡MetaStock
†See Traders’ Glossary for definition ‡See Editorial Resource Index
If you purchased, sold, or otherwise traded July and/or September 2008 CBOT Rough Rice futures contracts from July 8, 2008 through July 15, 2008 as an opening or closing transaction or otherwise, inclusive, then your rights will be affected and you may be entitled to a benefit. A settlement has been proposed in a class action lawsuit concerning the allegedly improper trading of July 2008 and September 2008 CBOT Rough Rice futures contracts on the Chicago Board of Trade from July 8, 2008 through July 15, 2008, inclusive. The settlement will provide $625,000 to pay claims from Persons who bought, sold, or otherwise traded the referenced futures contracts at any time from July 8, 2008 through July 15, 2008. If you qualify, you may send in a Proof of Claim form to potentially get benefits, or you can exclude yourself from the settlement, or object to it.
Settlement Agreement, available at the settlement website, describes all of the details about the proposed settlement. The exact amount each qualifying Settlement Class member will receive from the Settlement Fund cannot be calculated until (1) the Court approves the settlement; (2) certain amounts identified in the full Settlement Agreement are deducted from the Settlement Fund; and (3) the number of participating Class members and the amount of their Allowed Claims are determined.
The United States District Court for the Northern The number of claimants who send in claims District of Illinois (219 South Dearborn Street, varies widely from case to case. If less than Chicago, IL 60604) authorized this notice. Before 100% of the Settlement Class sends in a Proof of any money is paid, the Court will hold a Fairness Claim form, you could get more money. Hearing to decide whether to approve the settlement. How Do You Ask For a Payment? Who’s Included? If you are a Settlement Class member, you You are a Settlement Class member if you may seek to participate in the Settlement by purchased, sold, or otherwise traded July and/or submitting a Proof of Claim to the Settlement September 2008 CBOT Rough Rice futures contracts Administrator at the address below, postmarked from July 8, 2008 through July 15, 2008, inclusive. no later than November 9, 2015. You may Excluded from the Settlement Class are (i) the Released obtain a Proof of Claim on the settlement Parties (as defined in Section 1(k) of the Settlement website or by calling the toll-free number Agreement), and (ii) any Opt-Outs (as defined in referenced above. If you are a Settlement Class Paragraph 7 of the Settlement Agreement). member but do not file a Proof of Claim, you Contact your futures broker or futures will still be bound by the releases set forth in the commission merchant to see if you purchased, Settlement Agreement if the Court enters an sold or otherwise traded the referenced contracts. order approving the Settlement Agreement. If you’re not sure you are included, you can get What Are Your Other Options? more information, including the Settlement If you don’t want to be legally bound by Agreement, Mailed Notice, Plan of Allocation, the settlement, you must exclude yourself Proof of Claim and other important documents, by July 21, 2015, or you won’t be able to sue, or at www.ricefuturessettlement.com (“settlement continue to sue, Defendants about the legal claims website”) or by calling toll free 800-918-8964. in this case. If you exclude yourself, you can’t get What’s This About? money from this settlement. If you stay in the The lawsuit claims, among other things, that on settlement, you may object to it by August 3, 2015. July 11, 2008, Defendants held 100% of the reported All objections to or requests to be excluded from open interest in the CBOT Rough Rice futures the settlement must be made in accordance with contract expiring in July 2008 and that Defendants, the instructions set forth in the formal Mailed by July 11, 2008, had made large purchases in the Notice. The Mailed Notice available at rice cash market with the purpose and intent of www.ricefuturessettlement.com explains how to limiting the amount of rice that would be available exclude yourself or object. for delivery against the July 2008 CBOT Rough The Court will hold a Fairness Hearing in this Rice futures contracts. Plaintiffs also alleged that case (In re: Rough Rice Commodity Litigation, Defendants uneconomically stood for delivery on Case No. 11-cv-00618) on August 25, 2015, to their July 2008 position during the Settlement Class consider whether to approve the settlement and a Period. Defendants deny any wrongdoing that request by the lawyers representing all Plaintiffs allege in the lawsuit and maintain that Settlement Class members (Lovell Stewart they have complied with their legal obligations. Halebian Jacobson LLP and Lowey Dannenberg The Court did not decide which side is right. Cohen & Hart, P.C.,) for an award of attorneys’ But both sides agreed to the settlement to resolve fees of no more than one-third (i.e., 33 1/3%) of the case and get benefits to potentially affected the Settlement Fund for investigating the facts, market participants. The two sides disagree on litigating the case, and negotiating the how much money could have been won if the settlement, and for reimbursement of their costs Plaintiffs had won at trial. and expenses in the amount of no more than approximately $50,000.00. What Does the Settlement Provide? Under the settlement, Defendants agreed to create a $625,000 Settlement Fund. If the Court approves the settlement, potential Settlement Class members who qualify and send in valid Proof of Claim forms will receive a share of the Settlement Fund, after it is reduced by the payment of certain fees and expenses. The
May 2015
You may ask to appear at the Fairness Hearing, but you don’t have to. For more information, call toll free 800-918-8964, visit the website www.ricefuturessettlement.com, or write to In re: Rough Rice Commodity Litigation Settlement, c/o A.B. Data, Ltd., PO Box 170500, Milwaukee, WI 53217-8091.
• Technical Analysis of Stocks & Commodities • 21
Different strokes
Mean Reversion And The S&P 500 It is generally believed that markets tend to mean-revert. But this is true for some markets more than others. Here’s an in-depth look at how the S&P 500 responds to mean reversion. by Stephen Beatson
M
ean reversion is not a universal phenomenon; some markets have a tendency toward mean reversion, while others don’t. This has led a number of analysts and traders to look upon mean reversion with some degree of suspicion. If mean reversion has a solid statistical foundation, should it not be applicable to all markets all the time?
400,000
Equity ($)
300,000
200,000
100,000
Equity curve 0
-100,000 1/2/1970
8/18/1978
4/6/1987
11/16/1995
7/20/2004
3/18/2013
Figure 1: EQUITY CURVE. Here you see the results of buying the S&P 500 index at a 10-day low and selling 10 days later (1970–2013). The green line shows the strategy’s strong and persistent edge.
22 • May 2015 • Technical Analysis of Stocks & Commodities
n
Buy on the close if the index closes at a 10-day low
n
Sell on the close 10 days later
n
$100,000 per trade, no allowance for commissions or slippage
Figure 1 shows the resulting equity curve. During the 1970–2013 period, the strategy’s win rate was 60.52%
ShredDEd bills: mary981/Arrow: tomwa/collage: JOAN BARreTT
Rhythm Of The Markets
The simple truth is that some markets respond well to mean-reversion strategies, and others don’t. There are several possible explanations for this, including what factors drive the instrument’s price (macro economics, earnings, news, and so on); the number of market participants; the ability to take short positions; the volumes involved; and the average volatility of the instrument in question. It is generally believed that commodity time series respond better to continuationtype systems (trend-following, breakout, and so on) than to mean-reversion systems. The same applies to currency pairs, which are understood to exhibit long- and short-term trending tendencies. The US stock market daily time series, on the other hand, has consistently demonstrated a strong propensity toward mean reversion. In this article, I will try to determine whether this has always been the case. The focus here will be on long-side mean reversion, that is, on a security’s price’s tendency to move upward after a shortterm decline. I looked at the S&P 500 index from 1970–2013 and applied the following strategy:
During both bull and bear markets, a short-term fall in the S&P 500 is more likely to be followed by a bounce than by a continued drop. and the profit factor (total profits/total losses) was 1.71. This indicates that when the index hit a 10-day low, a trader with a long position was generally better off holding his position and exiting 10 days later. The equity curve’s regular upward slope is quite remarkable. Of course, some of this tendency must be assigned to directional bias — after all, the S&P 500 went up in value almost 20 fold over the period, so you would expect the equity curve of this long-only strategy to display a positive edge. However, the upward slope is also stubbornly present throughout the past two decades (1994–2013) that saw some extreme rises and falls in stock market prices. Thus, the data suggests that during both bull and bear markets, a short-term fall in the S&P 500 is more likely to be followed by a bounce than by a continued drop, at least in the first few days that follow. In other words, buying price dips and selling at mean reversion would have been a simple and profitable strategy over the past 50 years. To further understand the nature of this short-term meanreversion cycle, I used the same strategy but applied a much shorter two-day holding period, as follows: Buy on the close if the index closes at a 10-day low
50,000
n
Sell on the close two days later
20,000
n
$100,000 per trade, no allowance for commissions or slippage
Figure 2 shows the resulting equity curve for the two-day holding period. What you see here is a very different chart. The downward sloping red line indicates that from 1970–1987, the S&P 500 exhibited a strong tendency toward short-term continuation, or follow-through — that is, a 10-day low in the index had a strong tendency to be
Where Order Flow Meets Price Velocity
Watch the Cops and Kings chase down the elusive Convict Live Charting Room Free to First 100 every week. followed by further selling, at least in the short term (two days). From 1987 to date, however, and quite consistently for the past three decades, the exact opposite appears to have happened: A 10-day low was generally followed by a quick
40,000 30,000
Reversion
10,000 0
Equity ($)
n
PivotHunter.com®
-10,000 -20,000 -30,000 -40,000
Continuation
-50,000
Equity curve
-60,000 -70,000 -80,000 -90,000 -100,000
Black Monday 1/2/1970
8/18/1978
4/6/1987
11/16/1995
7/20/2004
3/18/2013
Figure 2: EQUITY CURVE OF A SHORTER HOLDING PERIOD. Here you see the results of buying the S&P 500 index at a 10-day low and selling two days later (1970–2013). The strategy had a negative edge until late 1987, then a fairly consistent positive edge thereafter. May 2015
• Technical Analysis of Stocks & Commodities • 23
1.4E+10
1.2E+10
MARKET OUTLOOK
1/4 /1 1/4 965 /19 1/4 67 /1 1/4 969 /1 1/4 971 /19 1/4 73 /1 1/4 975 /19 1/4 77 /1 1/4 979 /19 1/4 81 /1 1/4 983 /19 1/4 85 /1 1/4 987 /19 1/4 89 /1 1/4 991 /19 1/4 93 /1 1/4 995 /19 1/4 97 /1 1/4 999 /2 1/4 001 /20 1/4 03 /2 1/4 005 /20 1/4 07 /2 1/4 009 /20 1/4 11 /20 13
occurring intraday instead of interday. That meant that price stability could be reached before 1E+10 the end of the trading session, allowing opportunistic buyers to step in the following morning, 8E+09 pushing prices back up toward the mean. The third likely explanation for 6E+09 the US stock market’s short-term Volume mean-reverting tendency is the pervasiveness of short selling. 4E+09 Short selling, in its many forms, has been around for a long time and was certainly very much 2E+09 alive in the 1960s and 1970s. But it was only in the 1980s that shorting on electronic platforms 0 became widely available. This rise in the collective power of the shorts has exacerbated the short-rally effect, which is one FIGURE 3: RISING VOLUMES. Stock market volumes have grown exponentially since the early 1960s, reaching their peaks of the main ingredients of mean in the fourth quarter of 2008. reversion. Essentially, when an instrument’s price is falling, bounce, or reversion. And the inflection point of the equity short sellers must buy to cover and take their profits. This curve occurred on a specific date, which was October 19, 1987, buying interest drives prices back up toward the mean. So also known as Black Monday. the greater the short interest, the stronger is the pressure for falling prices to revert back upward. Why the change? Finally, we must also consider the advent of strategy-driven There are arguably four main reasons for this fundamental automated trading systems and high-frequency traders (HFTs). change in the US stock market’s short-term profile from trend- These have, ironically, brought considerable short-term ing to mean reverting. The first lies in the extraordinary rise rationality to the marketplace. Trading systems designed to in the volume of stock market transactions over the past 50 recognize panic selling step in to oversold situations and buy years, as reflected in Figure 3. Buy & hold investing that was into market overreactions. This serves to discourage followthe hallmark of most of the 1900s has made way for active through and favors mean reversion. investing, short-term trading, and hedging. Financial products that were intended as vehicles for investment have become Stay the course tools for speculation. This is the case for just about every We have seen that the S&P 500 index has financial instrument available in the electronic marketplace. exhibited short-term long-side mean reverThis increased volume has brought about an unprecedented sion for at least the past half century. What’s level of liquidity to the marketplace, allowing buyers and more, this tendency has been prevalent in sellers to find each other more efficiently, thereby slowing the bullish and bearish periods. We have also runaway trains associated with illiquid markets. seen what appears to be a quickening in the A second possible explanation is that until the mid-1980s, way the market responds to falling prices. stop-loss orders were often executed on the day following Reversion cycles that in the 1970s and 1980s the stop being hit. That is, if a stop-loss level was touched took several days now occur within a much on Monday, the broker would execute the sale at the open on shorter time frame. Traders recognizing Tuesday morning. These sell orders would serve to compound this phenomenon should be able to profit by the downward effect, resulting in more stops being hit, and using mean-reversion strategies that exploit more sell orders being generated on Wednesday morning. The this edge. downward spiral would continue until value investors stepped in and confidence was restored. The big change occurred in the Stephen Beatson is an investment consultant based in Paris, mid-1980s with the advent of automated systems that allowed France and is founder of the educational site TheMechanistop-losses to be executed instantly when hit. The multiday calTrader.com. price erosion process I described was suddenly compressed, 24 • May 2015 • Technical Analysis of Stocks & Commodities
FUTURES FOR YOU INSIDE THE FUTURES WORLD Want to find out how the futures markets really work? Carley Garner is the senior strategist for DeCarley Trading, a division of Zaner Group, where she also works as a broker. She authors widely distributed e-newsletters; for your free subscription, visit www.DeCarleyTrading.com. Her books—Currency Trading In The Forex And Futures Markets; A Trader’s First Book On Commodities; and Commodity Options—were published by FT Press. To submit a question, post your question at http://Message-Boards.Traders.com. Answers will be posted there, and selected questions will appear in a future issue of S&C.
PIT CLOSURE IMPACT What impact will the CME’s pit closure have on the average trader? In early 2015, the Chicago Mercantile Exchange (CME) announced that it would be closing pit trading for all futures contracts in its Chicago and New York operations, with the exception of the full-sized S&P 500 contract and most of its option pits. The S&P 500 futures contract was the only product that had never been moved to “the screen.” While all other CME futures products traded in both an electronic version and open-outcry version side-by-side, execution in the “big” S&P has always been strictly open outcry. Because of this, many traders moved their speculation from the original S&P contract into the electronically executed emini S&P 500 futures (ES) to avoid the delays and slippage that sometimes came with pit-traded execution. If you are young or new to the trading community, you might not be aware of what trading pits are or about the practice of open-outcry execution. In short, the pits are designated circular areas in which exchange members buy and sell futures contracts through hand gestures known as arb. Although the transactions are dictated by hand movements, they are accompanied by aggressive voices, and in some cases, a degree of physicality. The process of open-outcry trade execution is often referred to as “organized chaos.” If you haven’t seen the movie Trading Places starring Eddie Murphy and Dan Aykroyd (one of the greatest movies of all time, in my opinion!), you should; it manages to capture the essence of pit trading in all of its glory. As another one to mention, the documentary Floored directed by James Allen Smith tells a
candid story of the ups and downs experienced by those on the CME trading floors in Chicago. As I mentioned, along with the fullsize S&P 500 futures, the option pits will continue operation. This is because the complexity of option trading hasn’t translated to the screens as well as futures contracts have. Simple long & short calls & puts can easily be executed via an electronic platform, but in some circumstances, those trading multileg option spreads in high volume still find benefits in using an open-outcry execution broker. This is expected to be the case for the foreseeable future. Nonetheless,
Although the closure of the futures trading pits officially marks the end of an era, there will be little to no impact on the average retail trader. the writing is on the wall: Eventually, the option pits will likely go the way of the futures pits. Although the closure of the futures trading pits officially marks the end of an era, there will be little to no impact on the average retail trader. At the time the CME Group announced the pit closure, roughly 1% of all executed futures contracts on the exchange were traded in the pits. In other words, nearly all of CME Group trades are executed electronically; thus, most of us won’t even notice the change. However, there will certainly be casualties. For starters, all of those working on the trading floor (executing brokers, order clerks, and so on) will be May 2015
Carley Garner
out of a job. Even though the numbers of these individuals has dwindled over the years, there are still hundreds that will be affected. In addition, those who made a living disseminating order flow information (such as which banks and hedge funds were buying or selling in the pits), are finding they no longer have a place on the trading floor. Moreover, despite the clear advantages of electronic trade matching such as transparency, speed of execution and fill reporting, and fill quality (less slippage), the disadvantages are often overlooked. For instance, up until now, when the exchange, or even brokerage firms, experienced technology issues, it was possible to route orders to the pit for execution. These types of events don’t occur frequently, but they do happen. In the absence of an alternative means of execution, it only takes a single instance of halted trades to dramatically affect the integrity of the markets and work against orderly trading. In the aftermath of side-by-side trading (simultaneous futures markets trading) we are left with some residual chaos when it comes to identifying products in a trading platform or on a quote board. This is because when electronic trading was first introduced, the exchange, platform vendors, and brokerage firms opted to identify the electronically executed contract with a different symbol than the open-outcry version. This was because at the time, electronic trading hadn’t fully developed. Consequently, it was important that traders had the ability to choose which venue to route trades to. For example, the open outcry version of crude oil has always been denoted by Continued on page 62
• Technical Analysis of Stocks & Commodities • 25
are likely to come from today and previous days.
The Road Ahead
Predicting The VIX By Reordering Data In recent years, the CBOE Volatility Index (VIX) has increased in importance and use as an indicator of market direction. This article demonstrates how the direction of tomorrow’s change in the VIX might be determined by restructuring readily available market data.
As
by Stephen Butts
usually presented, financial data is ordered in a time series: The data runs from left to right or top to bottom by increasing values of days, months, years, and so on. The data may be altered (with a moving average, lagged values, the log, or square taken, and so forth), but in most cases, data is envisioned and laid out according to the arrow of time. Patterns are then observed or discovered in this time-based framework, and this makes great sense: The factor that may produce tomorrow’s market moves
26 • May 2015 • Technical Analysis of Stocks & Commodities
relationships But ordering data in just one way limits us. If there are patterns in the data that are not related to the one-directional march of time, we may miss them. This article seeks to explore one of many possible orderings of data from a well-studied, readily-available financial derivative—the volatility index (VIX). My objective is to see if I can discover useful relationships that are hidden in a time series view. I’ll begin with a small sample of VIX index data from the 30 days of trading between December 16, 1996 and January 28, 1997 (Figure 1). I arrange this data with the date in column 1, the closing value for each of the 30 days in column 2, and the next day’s change in the closing value in column 3. For example, the VIX closed at 19.27 on January 14, 1997 and I put the difference between this close and that for the following day, 0.13, into column 3 for January 14, not for January 15. I then sort all three columns together by descending value of the VIX close and add another column containing each day’s cumulative sum of the next day’s change (see Figure 2). Note that after the sorting, the dates are no longer in ascending order, and that the cumulative change in the VIX is calculated based on the new order of the 30 days. For the sake of brevity, hereafter we will refer to the change in the closing value of the VIX between a given trading day and the trading day immediately after as the delta, and to the cumulative value of the delta when it is ordered according to the scheme above as the cumulative delta. The chart in Figure 3 presents a graph of the descending VIX and the cumulative delta from Figure 2, and it shows a pattern: While the closing VIX values have been
PHOTO: STUART JENNER/SHUTTERSTOCK/COLLAGE: NIKKI MORR
Recognizing
MICROSOFT Excel
sorted to descend steadily from their highest value to their lowest (the 30 days are no longer in calendar order), the cumulative delta values (that is, the next day’s changes in the VIX) generally, but with several detours, drift down to a low point (-6.84) somewhere around the middle of the chart, and then change direction and move upward (with detours) to the last point in the chart. The low point for the cumulative delta occurs on day 15 of this re-sorted data (December 18, 1996) with a closing VIX value for that day of 19.42. To the left of and including this low point, more cumulative deltas fall (10) than rise (five), and the sum of the deltas for these (-9.93) is larger in a negative direction than the sum of the rises (3.09) is in a positive direction. So for every day in this 30-day period where the closing value of the VIX is greater than or equal to the Figure 2: closing vix sorted by descending vix. Here, VIX value on the day of the lowest Figure 1: closing vix sorted by date (december 16, 1996–january 1997). The the VIX close is sorted in descending order and a column containing point in column 5 of Figure 2, you data is arranged by date, closing value, and next each day’s cumulative sum of the next day’s change is added. would be correct 10 times out of 15 if day’s change. you were to predict that tomorrow’s VIX will be lower than today’s. And for every day when the 1990, for every day with a closing VIX greater or equal to closing value of the VIX is less than the VIX value on the day the VIX on the day with the lowest value for the cumulative of the lowest point in column 3, you would be correct 11 times next day’s change in the VIX (delta), there is a higher-thanout of 15 if you were to predict that tomorrow’s VIX will be average probability that the next day’s VIX will drop. And higher than today’s. In other words, if you knew the closing the reverse is true for all the days when the closing VIX is VIX for every day in the table and the value of the VIX on the less than the VIX on the day with the lowest value for the day having the low point of the cumulative deltas, you would cumulative delta. predict correctly for 21 of the 30 days for a total of 16.64 points, and incorrectly for nine days for 0.00 22.00 -4.72 points, leaving you 11.92 21.50 points ahead. Needless to say, -1.00 21.00 this is a far better performance -2.00 than the simple 30-day change 20.50 -3.00 in the VIX itself, which results 20.00 in a drop of just 1.76 points over -4.00 19.50 the period. 19.00 -5.00 This general pattern for the 19.42 18.50 reordered data is true for most -6.00 18.00 of the 30-day periods since the -7.00 VIX has existed. The low point 17.50 -6.84 for the cumulative delta is rarely 17.00 -8.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 in the exact center of the chart as in the example in Figure 3, Days in order of descending VIX and often is skewed well to the FIGURE 3: DESCENDING VIX WITH CUMULATIVE CHANGE IN TOMORROW’S VIX. While the closing VIX values have been left or the right. However, in all sorted to descend steadily from their highest value to their lowest, the cumulative delta values generally drift down to a low point 30-day periods since January around the middle of the chart. Value of closing VIX
VIX cumulative change
May 2015
• Technical Analysis of Stocks & Commodities • 27
OPTIONS
700.00 600.00
for each 30 days would give a fair estimate of the VIX at the low point. In fact, both of them do: If you sub400.00 stitute the mean VIX value (which Maximum drawdown 300.00 you know for all days, including = 59.57 points today) for the observed VIX at each 200.00 low point in the historical data, the predictions for the next day’s VIX 100.00 change are correct for 3,078 (54%) of the days and produce 652 points 0.00 on the VIX. The chart in Figure 4 shows the cumulative points gained -100.00 Year over the 23 years in this case; note that the maximum drawdown for FIGURE 4: cumulative vix outcome by year using mean prediction. The maximum drawdown for the entire the entire run is less than 60 points. run is less than 60 points. Using the median is almost as good, producing 617 points. How well does this method work? Predicting the direction of tomorrow’s VIX with an advanThe historical database I used for the VIX comprises every tage of 652 points in 23 years may not sound like much until trading day from January 29, 1990 to December 31, 2012 and you consider the alternatives. If you were simply to predict there are 5,749 overlapping 30-day periods in this span. If for each day over the 23 years that tomorrow’s VIX would you predict whether the next day’s VIX change is up or down always rise or always fall, you would be right about 47.6% of for the 30th day in each period as per the method discussed, the time for rising and 52.4% for falling, for a grand total of 3,686 of the total 5,713 predictions (that is, 65%; for 36 days, only ±1.91 points. A 1,000-trial Monte Carlo simulation of the VIX change is zero) are correct for a net gain of 2,571 random predictions for each day in the 23 years shows that VIX index points for the 23 years studied. you would have achieved a gain of more than 300 points just But you want to predict not just the following day’s VIX for eight times in 1,000 tries (that’s 23,000 years), and that you some dates in the past, but tomorrow’s VIX today, in real time. would gain a maximum of 361 points just once. Since you want tomorrow’s change in the VIX, you should begin with the 30 trading day period ending today, and then What will tomorrow bring? reorder the data as mentioned. Then find which day has the In fact, you may get a more accurate estimate of the VIX value lowest cumulative delta, and note the VIX value on that day. needed from the historical data by using more sophisticated However, remember that for each day in the period, you need statistical methods than simply calculating the mean. With a to have the next day’s change in the VIX, and of course, you least-squares approach using multiple variables, I have been don’t know tomorrow’s number today. Thus, you can reorder able to achieve close to a 900-point gain over the period (I the data by sorting them into descending order by each day’s leave an exploration of this to readers as an exercise, and if VIX, but you cannot calculate the cumulative change values you can do better, I invite you to share your method with other because the latest one is missing. readers of this magazine). The only way to overcome this problem is to bypass it. You I hope that the information I have presented here will give need the value of the VIX on the day when the cumulative delta traders a small but useful advantage in predicting short-term is the lowest for the last 30 days. Without knowing the data moves in the VIX, and encourage all analysts to take another from the last day you can use various means to estimate that look at how time-series market data might be rearranged to VIX value. For example, you know from historical analysis produce interesting insights. that the low points for the cumulative VIX changes tend to cluster somewhere near the middle of the sorted VIX values Stephen Butts has a PhD in political science. He has worked for each period. Perhaps the mean or the median VIX value as a statistical and methodological consultant and teacher at Columbia University, the Bureau of Applied Social Research in New York, and the University of Wisconsin-Madison. He may be reached via email at
[email protected]. Cumulative VIX
500.00
The factor that may produce tomorrow’s market moves are likely to come from today and previous days.
28 • May 2015 • Technical Analysis of Stocks & Commodities
Further reading
Gardner, Trent [2012]. “Using VIX To Forecast The S&P 500,” Technical Analysis of Stocks & Commodities, Volume 30: July.
Q&A SINCE YOU ASKED Confused about some aspect of trading? Professional trader Don Bright of Bright Trading (www.stocktrading.com), an equity trading corporation, answers a few of your questions. To submit a question, post your question to our website at http:// Message-Boards.Traders.com. Answers will be posted there, and selected questions will appear in a future issue of S&C. Don Bright of Bright Trading
MARKET LIQUIDITY AND THE TICK SIZE PILOT PROGRAM Mr. Bright, thanks for all the information over the years. Your columns have been very helpful. I have been hearing/ reading about the possibility of changing from the penny tick size in trading. I have to admit that I’m not sure what this even means. Are exchanges going to make changes? Will traders benefit from these changes? Does this mean the end to subpennies? (I’ve read some of your past columns on the topic of subpennies.) Are these changes coming soon?—Ted A. Thanks, Ted, for the nice words. Always glad to help. Please keep the questions coming. To address your question, what you’re referring to is the Tick Size Pilot Program (SEC File No. 4-657). “Tick size” here refers to expanding the minimum tick to five cents (or similar) for certain smallercap securities. Remember, we “old guys” traded with 1/8th tick sizes and what was called “teenies” (12.4 cents and 6.25 cents as the minimum tick size). This was before subpenny denominations and high-frequency trading and other such activities came to be, and before “penny jumping” practices came to be, although some firms did engage in “leaning” on another order, placing their order a “teenie” higher, thus limiting their losses to the 6.25 cents. These minimum price increments have an impact on displayed liquidity and transaction costs. Since the markets adopted decimalization, there have been changes in market structure, the way people trade, and the roles market participants played. How will expanding the minimum tick size for certain stocks help market participants?
The Securities And Exchange Commission (SEC) has been taking public comments regarding this pilot program. We at Bright Trading submitted a comment letter to the SEC on it, prepared by Dennis Dick, CFA (a trader at Bright Trading and a trader lobbyist). I’ll summarize and reiterate some of our submitted comments here. (Note: This information and the link to our comment letter represent the opinion of Bright Trading, and may not be shared by all traders.) One of our concerns is the lack of interest in the small- and mid-capitalization companies. We believe this is primarily due to a lack of liquidity caused by the discouragement of limit-order traders. As we see it, many small-capitalization companies trade with very wide spreads,
The dominance of the penny-jumping program discourages other participants from providing liquidity. which should be attractive for market makers to trade because of the potential profit opportunity from the wide spread. However, algorithmic penny-jumping programs appear to dominate these securities, discouraging other participants from providing liquidity. Here’s an example: Assume stock XYZ is trading with a spread of $25.00 x $25.20. If a market participant places a bid at $25.01 to tighten the spread, the algorithmic penny-jumping program will automatically bid $25.02. If the original May 2015
participant raises their bid to $25.03, the algorithmic program will raise its bid to $25.04. If the original participant cancels their bid, the algorithmic penny-jumping program cancels its bid as well, and the best bid returns to $25.00. The pennyjumping technique is designed to battle for the top of the order queue, so as to be the first to interact with incoming marketable order flow, giving it the best chance to capture the spread. The dominance of this penny-jumping activity discourages other participants from providing liquidity, which keeps the spread on these securities artificially wide. We believe the tick size pilot will help address this issue. Firms will not want to risk 500% more slippage than they face with penny ticks. Another major concern of ours is broker–dealer internalization (that is, when the broker takes the other side of trade versus open market trading). We write in our comments that “The biggest issue that our traders cite is their inability to get filled on their limit orders, even when they are at the top of the order queue. This is primarily due to over-the-counter (OTC) market makers intercepting marketable order flow that would otherwise interact with the trader’s displayed limit order.” In an earlier public comment letter that we submitted to the SEC in June 2010, we cited this issue and recommended that the SEC require an OTC market maker internalizing a retail order to provide meaningful price improvement over the displayed quote. We commend the SEC in attempting to address this issue in the tick size pilot program but believe an exception in one Continued on page 45
• Technical Analysis of Stocks & Commodities • 29
Psst, Here’s A Secret
10 Selling Tips
Do you spend as much time deciding to sell as deciding to buy? Here are 10 tips to make deciding when to sell easier.
T
by Thomas Bulkowski
Landscape: blue67 sign/sign pole: vipman
he stock of Ferro Corporation (FOE) has flatlined like a dead animal since November 2013. It is now 2015, and I have more than doubled my money. Should I sell? Every trader or investor must answer that type of question for their own investments. In this article, I’ll discuss 10 selling tips to help you find your answer.
Tip #1: Use stops. This has to top the list of selling techniques because it makes the process simple. If you use a stop-loss order, you can quit worrying about when to sell. The order will take care of that. All you have to do is locate the stop in the right place. Stop placement is an art beyond this article, but if you are having difficulty, use a volatility stop. I learned about volatility stops from Perry Kaufman’s book A Short Course In Technical Trading. The idea behind a volatility stop is to place a stop far enough away from the current price to avoid being stopped out on normal
30 • May 2015 • Technical Analysis of Stocks & Commodities
price movement. A volatility stop works well when no other stop locations are nearby. To use a volatility stop, compute the high-low difference of price each day for the last month (about 21 price bars). Average the result and multiply by 2 to get the volatility. Subtract the volatility from the current low price to get the stop price. Trail the stop upward as price rises. This type of stop is similar to a chandelier stop, but Kaufman’s stop performs better. I found that a multiplier of 2 works best for my trades, but you may prefer 1.5 or another number. If the stop price, in percentage terms, is too far away from the current price, then skip the trade and find a less volatile stock. The computation need not be a painful experience. One way to do it is to go to http://finance. yahoo.com (Figure 1), type in the stock symbol, and on the left, you’ll see a link to historical prices. I show that circled in red in Figure 1. Click on that link and download the prices into a spreadsheet. On the same Yahoo page, the link for a spreadsheet download is located below the price grid. Subtract the high price from the low and use the average() worksheet function to get the volatility. I am providing a sample spreadsheet at http://traders.com/files/ VolStop.xls.zip that does this math for you. All you have to do is paste in a month’s worth of price quotes from Yahoo. Tip #2: Sell at a target price. When the stock reaches a price target, sell. You should have the target (and a stop-loss) price picked out before trading the stock. Consider placing the target just below overhead resistance, such as at a prior peak or valley, or at a round number (double or triple zeros such as $9.00 or $10.00 are good for daytraders), or at some other price target. If using a round number, consider selling a few cents below the target, like 8.93 and 9.95. You want to beat the crowd to the exit. If you are using chart patterns as the entry signal, take the FIGURE 1: HISTORICAL PRICES. height of the chart Downloading historical prices can be done easily given that there are sevpattern and add it to eral resources available from which the breakout price to do so. Here you see an example to get a target. That using Yahoo! Finance.
method works about 70% to 80% of the time, depending on the chart pattern. You can cut the height in half for a closer target that will work close to 90% of the time. In Figure 2 you see such an example. A double-bottom chart pattern appears at AB and it confirms as a valid pattern when price closes above C, which is the highest peak between the two bottoms. C is also the breakout price. The height from C ($24.50) to the lower of the two bottoms, A ($21.67), is 2.83. Add the height to the breakout price (C) to get a target of $27.33. Line D shows the target. For double bottoms, this method works 68% of the time. Using half the height boosts the success rate to 86%. Line D is also where overhead resistance begins (between the two red lines). Thus, the target and overhead resistance merge to form a good sell target. Target selling is most useful for swing and daytraders because time is money (that is, they want to keep their money in stocks climbing the fastest). Position traders looking for a trend change (a drop of at least 20%) should avoid using this approach. That’s because, as the chart shows, selling at the price of line D limits profit since the stock can continue higher. Tip #3: Sell after an adverse breakout. If you see a chart pattern and buy into it before the breakout, sometimes the breakout is opposite of what you had planned. In that case, sell immediately. This applies to any approach you use to buy a stock. If the stock moves in a surprising direction (down when you expected it to rise), then you need to sell your position. This tip is especially useful during earnings season. A stock suffering a hit in earnings will get slammed by the market. For any trading style other than buy & hold, sell immediately. Otherwise, the stock is likely to be whacked again in three or six months’ time.
Jurik algorithms deliver low lag, low noise analysis Tools for: TradeStation, AmiBroker, Investor/RT, MultiCharts, NeuroShell Trader, eSignal, NeoTicker, Tradecision, TradingSolutions, MATLAB, Ninja Trader, Genesis TradeNavigator, Market Delta, Extreme charts, DLLs for custom software
Jurik Tools on live charts, on the web ! tinyurl.com/jurik-online
Jurik Research
2010 -- 2011 -- 2012 -- 2013 Add-In software
jurikres.com • 800-810-3646 • 719-686-0074
section at the end of this article. Avoid buying or owning countertrend stocks by checking whether they drop when the market makes a big move higher. If so, and if they do that for several days in a month, consider selling those holdings. The days need not be consecutive, but if
Overhead resistance
D
C
A
B Tom Bulkowski
Tip #4: Watch for countertrend moves. Last year I was planning to add a stock to my portfolio, but when I checked its price, I saw that the stock price was dropping when the Dow Jones Industrial Average was soaring. That drop in the stock was a countertrend move. You may have heard the phrase a rising tide lifts all boats, but this boat had a hole in it. I avoided the stock and saved myself from drowning. I did a study on stocks that made countertrend moves. I compared stocks that dropped when the index made a significant move upward (at least 1%). Stocks that dropped underperformed the index for at least three months and underperformed the trend followers (those stocks that followed the index higher) for a year. You can find a link to this research in the “further reading”
Noisy indicators delay your analysis
FIGURE 2: APPLYING CHART PATTERNS TO MAKE SELL DECISIONS. In this case, take the height of the chart pattern and add it to the breakout price to get a target. May 2015
• Technical Analysis of Stocks & Commodities • 31
TRADING TECHNIQUES
If you wake up in the early hours of the night and your first thought is of trading, that’s a clue you have unresolved issues. the countertrend moves occur too frequently, you will be able to do better by selecting a stock that follows the uptrend. Tip #5: Obey indicator sell signals. If you are a system trader, when your system issues a sell order, obey it. If you try to finesse the exit by waiting, that could cause you more grief because you’re creating a bad habit. Imagine that you delay selling and the stock climbs further. By holding on longer, you are rewarded. The next time this scenario unfolds, you do the same thing and make more money. By delaying selling and winning, you create a bad habit that is like a bomb waiting to explode. Eventually, the same scenario will unfold, you’ll hold on, and boom! The stock will plummet, giving you a huge loss. If your system is flawed such that it gives sell signals too early, then that’s fine. Indicator testing can identify this problem. I’m not talking about living with a flawed system. I’m talking about ignoring trading signals. If you are going to use a system, then trust it and obey it. If you don’t trust the system, then research its operation to identify its weaknesses and become confident that the system works. Then avoid trading when those weaknesses are apparent (say, during bear markets or during the afternoon session when volume tapers off). Tip #6: Worried about a trade? I remember planning to daytrade Intel (INTC) by shorting 2,000 shares of the stock at the open. I went to bed and worried about the trade. Why? Because the position size was double what I normally used. So I got up and reduced the size of the trade. That turned out to be a good idea because the trade lost money (I was right in the direction, but was flushed out when the stock spiked upward within minutes of the open). A friend of mine had a similar problem. She was having difficulty sleeping through the night because of worries about her position in a mining stock. My advice to her was to cut the position size until her worries disappeared. If the stock still bothered her, then she should sell it completely. If you wake up in the early hours of the night and your first thought is of trading, then that is a good clue you have unresolved issues. This is especially true if sleepless-
ness happens repeatedly. Do what it takes to eliminate that stress by making adjustments such as reducing your trading and investing more or moving into cash for a while. Or perhaps it’s time for a vacation. Tip #7: Sell when the market says you’re wrong. Keep that phrase in mind when looking at your losing positions. If the stock did not act as you expected it to, then sell. I like to have a diversified portfolio by owning several securities and diversifying by trading style. I may have some stocks that I am swing trading, some that I am position trading, and some that I am buying & holding. The buy & hold stocks do not require much attention. That inattention can lead to problems when a cyclical stock goes out of favor. Perhaps the reasons for ownership have changed. If the company sells a division that removes a source of big profits, the stock could suffer, and it’s time to sell. What sometimes happens is that I’ll buy a stock and it’ll act like I expected it to for a few weeks by moving higher. As soon as my focus turns elsewhere, the stock drops. The decline is not an in-your-face plunge of 50% in one session. Rather, the stock eases lower day by day, eating away the meager profit it once enjoyed. Soon, the position is losing money. The market is telling me I made a mistake. The transition from profit to loss is a wake-up call—a sell signal. Obey it. Tip #8: Price drops 10% from a high. A financial consultant I know told me that she uses a 10% rule to sell her blue-chip stocks. A blue-chip stock is a low-volatility, high-quality, highvolume security priced above $5. If it drops by 5% from a peak, it piques her interest. She will look at the fundamentals
B
C
D
A FIGURE 3. PRICE BREAKS SUPPORT. When a stock breaks support, as in the case of Ferro Corp. (FOE) in this weekly chart, consider selling it.
32 • May 2015 • Technical Analysis of Stocks & Commodities
and try to understand the reason for the decline. If the market is dropping as part of a bullish move higher (a retrace in an uptrend), for example, she will ignore the temporary slump. If the drop turns into a 10% decline from the peak, she will sell at least half, but most often, the entire position will go. Tip #9: Market relative strength. If you manage money for a living, either privately with your own portfolio or professionally by steering a hedge fund, your goal is to beat the competition. The competition is the market indexes. One way to do that is to find stocks that are outperforming the market. Here’s what you do. Divide the closing stock price each day by the corresponding close in the S&P index (or the index of your choice, like the utility index for a utility stock, as an example). You should see one line on the chart. If the line slopes upward, your stock is outperforming the index. If it trends lower, the stock is performing less well. This technique is what I call market relative strength since you are comparing the stock’s performance to the overall market. Sell stocks that are underperforming the market. I apply a moving average to the line (22 days is what I use) to smooth out the bumps so I can see the overall trend. The direction of the trend is what’s most important. Ideally, you want to have a portfolio of stocks that are climbing faster than the market. Tip #10: Price breaks support. I mentioned this stock (Figure 3, weekly scale) at the beginning of this article. The stock hit bottom at A in November 2012 and then it went on a tear. The stock climbed in a stairstep fashion until it bumped up against a ceiling. It slid horizontally for months, resting on a new floor at $12. After such a big gain from the low at $2.38, I expected it to take an extended break. My hope was that the stock would continue higher, resuming the stairstep move up. When the stock poked its tail through support at C, it was not an automatic sale. I know that these types of false signals happen from time to time when the smart money tries to shake out the uninformed. I monitored the stock closely and decided to hold on. Fortunately, the stock recovered but still continued moving sideways. Then the stock pushed through support at D and congested just below $12. It sat there for about two weeks. Again, I sat and waited just in case the move was trying to shake me out of a winning position. However, with the stock resting below support, it pushed me closer to the exit. I just needed a shove. The push came when the stock dropped 5% in one day but
closed slightly above the open. I decided to sell the following day. After buying the stock at $4.93, I sold it at $11.53 for a gain of 134%. The point of this anecdote is a simple one. When a stock breaks support, consider selling it. In Ferro’s case, it was time to let it go, so I sold. Timely selling is what this game is all about. S&C Contributing Writer Thomas Bulkowski (who may be reached via email at
[email protected]) is a private investor and trader with more than 30 years of market experience and considered by some to be a leading expert on chart patterns. He is the author of several books including Getting Started In Chart Patterns, Second Edition and The Evolution Of A Trader trilogy. His website and blog, www.thepatternsite.com, have more than 600 articles of free information dedicated to price pattern research. The spreadsheet file mentioned in this article is downloadable from http://traders.com/files/VolStop.xls.zip as well as from the Subscriber Area at our website, www.Traders.com, in the Article Code area.
Further reading
Kaufman, Perry [2003]. “Short Course In Technical Trading,” Wiley & Sons. • http://thepatternsite.com/CounterTrends.html (note this URL is case-sensitive)
May 2015
• Technical Analysis of Stocks & Commodities • 33
INTERVIEW
Knowledge, Patience, Discipline
TA For The Longer Term With Boon Chin Low
What led to your being interested in technical analysis? It was 1986 when I began trading in Japanese commodities that I started in technical analysis. There was little by way of fundamentals, so charts were the only thing that made sense. Technical analysis was on the cusp of a new era in Singapore, spurred by the publication of John Murphy’s first book, Technical Analysis Of The Futures Markets, in 1986. Still, it was tough trading commodities with huge paper charts, handdrawn candlestick bars, and trendlines, plus moving averages calculated manually. I began to feverishly learn technical analysis from as many sources as I could. When I joined Merrill Lynch Singapore in 1989 as a technical analyst, I had my first encounter with technical analysis software, CompuTrac. From then on, I was hooked on TA!
Understanding the relationship between the different time frames helps the investor to know how the bigger time frame trends change.
That’s interesting. You know, the premiere issue of this magazine was distributed at a CompuTrac seminar. From your experience, do you think technical analysis taught you things about the market that you may not have known if you hadn’t applied technical analysis? Certainly. Technical analysis taught me that all markets can be reduced to the common denominator of price. And as Charles Dow said more than 100 years ago, charts show that markets move in trends. With technical analysis, it is possible to forecast market direction, something much needed by all investors and traders. What I hope to add to this premise through my book Integrating Technical Analysis For The Investor is that it is also possible to forecast farther into the future with the consistent application of technical analysis into the larger time frame charts such as weekly, monthly, quarterly, and even longer time frames.
34 • May 2015 • Technical Analysis of Stocks & Commodities
Applying technical analysis for investing or on longer time frames is contradictory to what most people do. Do you combine technical analysis with fundamental analysis and if so, how? I think the reason technical analysis has not been used for the longer time frame is because there has not been sufficient focus on using it that way. Technicians tend to be traders and not investors. So the use of technical analysis tended to be skewed towards the shorter time frame. And in recent times, time frames have gotten even shorter. But there is no reason for technical analysis not to work in the longer time frame—it is, after all, still about price, albeit on a longer time frame. And investors should discover that technical analysis works just as well in those time frames. Some mental adjustment is needed to get used to the larger dimensions in time and space, but it can be done.
Singapore skyline: joyful/Shutterstock
BC Low has been a teacher and practitioner of technical analysis since the 1980s. He is one of Singapore’s earliest practitioners to attain the Chartered Market Technician credential from the Market Technicians Association, which is based in New York. From 1991 to 2011, Low was a senior lecturer at Singapore Polytechnic, where he introduced education in technical analysis to the curriculum. He created and taught two modules of “Technical Analysis and Trading,” the only formal course on technical analysis in Singapore. Low has been active in the Singapore Technical Analysts & Traders Society (STATS) since the 1990s. He launched the Certificate in Technical Analysis (CTA) through STATS in 2004; he launched the society’s Diploma in Technical Analysis (DTA) in 2010. Prior to his work at Singapore Polytechnic, Low was a technical analyst for Merrill Lynch Bank, where he provided currency views to dealers, private bankers, and institutional clients. Currently, he continues to trade his own equity. His recent book release is Integrating Technical Analysis For The Investor. S tocks & Commodities Editor Jayanthi Gopalakrishnan communicated with him via email in early March 2015 to find out more about how longer-term investors can apply technical analysis.
"Best Technical Analysis Website" for 14 Straight Years!
Create high-quality technical charts with just a couple of clicks No software to install - works with any browser Watch your charts update automatically in real-time Our charting tools are preferred by hundreds of online analysts Access your charts anywhere using any web-enabled device Learn about chart analysis from experts like John Murphy and Martin Pring NEW! Free, insightful weekly webinars hosted by expert analysts
New! Free Weekly Webinars
"In our opinion, it is the ‘best of the best’ among all online technical analysis and charting services." - American Association of Individual Investors, February 2013
“Best Technical Analysis Web Site” for 14 Straight Years!
as voted by the readers of Stocks and Commodities magazine
Get started today at http://stockcharts.com
®
NeuroShell Trader Now Works with FXCM Data and Brokerage
Winner 13 years in a row!
www.NeuroShell.com 301.662.7950 As far as possible, I try to avoid fundamental analysis. But once I am given a fundamental view, I will look at my charts to evaluate the outlook based on the fundamental analysis and see if the fundamental view is supported by the technical view on the longer time frames.
If the fundamental view is divergent with my technical views, naturally I will not act on the fundamental view. You integrate trend, price, and timing. How do you determine if a trend is just beginning or is close to maturing? The end of one trend may signal the start of a new trend; that is, the end of a downtrend may imply the start of an uptrend. There are indicators that help investors ascertain the timing of the start and end of a trend. One of my favorites is the combination of a 10- and 40-period exponential moving average (EMA). In the simplest terms, the crossing of the 10 EMA above or below the 40 EMA can be interpreted as the start of an uptrend or downtrend, respectively. I have used this combination for all markets and all time frames, and it has proved to be invaluable. That sounds simple enough. You also look at price patterns. What patterns do you look at and how are they helpful? My view is that trends change as a result of changing fundamentals. But as we know, economic fundaments do not normally change overnight. As such, larger reversal patterns will reflect such changes in fundaments better than the minor patterns will.
“Calls may be monitored because these are uncertain times.” 36 • May 2015 • Technical Analysis of Stocks & Commodities
Historically, I observed that while reversal patterns—such as double tops or bottoms, or head & shoulders—do not occur frequently, trends do change all the time. So there is a need for reliable indicators of trend change other than price patterns. My use of price patterns is to have them play a confirmation role. Once you can establish that a trend is possibly reversing with another indicator, you can look out for price patterns to confirm the impending change. I find that breakaway gaps have done very well in this respect, as have certain candlestick patterns. I have used this approach of using price patterns as a confirmation tool for many years in various markets, and it has worked well. It places the horse before the cart, so to speak, rather than the other way around. What time frame charts do you look at when you do your analysis? I look at time frames from daily all the way to the quarterly. Sometimes, I even look at the yearly chart just for a very big picture view, but that is rare and I do that only for exceptional circumstances. Typically, the daily and weekly charts are the staples; I prefer to use the weekly chart, as it filters the noise of the daily chart and provides earlier signals than the monthly. As the monthly chart is only evaluated at the end of the calendar month, its impact will be slower, but it is still a necessary tool for investors. My interest in the longer time frames led me to the important discovery that technical signals are consistent in whichever time frame they are used. If, for example, the crossing of two moving averages signals an uptrend in the daily chart, the same is true in the weekly, monthly, and other charts. The difference is that the time and price scales will be increasingly larger. Because of this consistency in technical signals across all time frames, it is possible to achieve longer-term views of markets. This consistency has in fact been proved with charts dating back several decades such as in the Dow Jones Industrial Average. One important issue in using more than one time frame is to understand the relationship between the different time
frames at any one point in time. If the monthly, weekly, and daily trends are at odds with each other, how is the investor to make sense of it and invest? This relationship has not been well defined in the past. The hope is that my book will provide investors with a clear explanation of this relationship so that they can invest better. Understanding the relationship between the different time frames also helps the investor to know how the bigger time frame trends change—it involves the progressive changing of the shorter time frame trends, causing the bigger time frame trend to eventually change. This process, described in my book, can help investors to fathom out the likely change or continuation in the long-term trend of markets such as gold, crude oil, and the US stock indexes. What are some of your favorite indicators? Despite my constant refrain for investors not to look for the holy grail in indicators, my most important indicator is the 10 & 40 EMA combinations, as I mentioned earlier. It has a proven track record and is unique in being able to perform more than one important function well, and that is to forecast future trends, to monitor the process of a change in trend, and to provide important support or resistance levels in a trending market. Few other indicators can achieve that much that well. The other indicator I like a lot is George Lane’s classic stochastic oscillator, because of its ability to signal quick turns in the market under certain market conditions. However, I recommend that investors understand the specific role that stochastics is equipped to play, and not expect it to perform more than it can. A sprinter is not a marathon runner, and vice versa.
ninjatrader.com
Simple Rates. Clear Savings. Deep Discount Commissions as low as $.53 per contract. View Commissions at ninjatrader.com/Commissions You Know our Award-Winning Platform. Get to Know our Next Generation Brokerage. NinjaTrader Brokerage™ equips traders with unmatched benefits including exclusive advanced platform features and single source support.
Go Direct to the Source. NinjaTrader Brokerage. Trading is about probability. How does applying TA for the longer time frame improve your odds of success? I’ll start with Charles Dow’s axiom that markets move in trends. If you look at the longer time frame charts, the trends are more clearly discernible because the short-term gyrations are filtered out. However, as you get into the shorter time frame, say, daily or intraday, the shorter-term gyrations complicate the view of the trend. This greater shorter-term uncertainty increases the probability of making the wrong decision or lowers the probability May 2015
of making the right decision. Most people will think that trading for the longer term is more forgiving. Is it, or do you still have to have as much discipline to gain the higher returns? The use of the longer time frames gives the investor expanded space to place the stops in return for catching the bigger returns from longer-term moves. If the investor focuses on the long term, the day-to-day market gyrations may not impact the bigger trend. Because of that, you can be misled into thinking
• Technical Analysis of Stocks & Commodities • 37
that discipline can be more relaxed, but that should not happen. Key market levels of longer-term charts should still be identified and earmarked for action even though they may be further away. In other words, the same discipline is needed for the longer time frame as it is for the shorter. How much of an influence does fundamental analysis have on your investment decisions? While trends are underpinned by market fundaments, technicals, in my opinion, pinpoint the timing of trends more clearly and visibly. I find that technical signals tend to front-run the fundamentals, such that when the news finally appears, good technical signals would have already been triggered and the move is well underway. In fact, when I am given a fundamental view of a market, I immediately refer to my charts to evaluate the fundamental view. What are the important factors to be mindful of when managing your positions? There are many factors to bear in mind, but since I am advocating the use of technical analysis for investing long
term, I wish to emphasize that investors should not neglect their positions just because they are long term. Investors, like everyone else, lead busy lives—work, family, and social. So when a position is profitable, they tend to just let it run and take no action. But as markets move in trends, they eventually turn bearish and wipe out what was previously a profitable position. This is one of the more common and serious errors in investing—investors do not plan the exit of their positions enough. In my experience, investors are more focused about entering the market, but not so much in exiting. Investors must monitor their investments regularly; and while identifying a good time and price to exit a position may not be easy, neglecting the positions is not an option. From your book, it appears as if you use indicators and patterns that are in the public domain. Do you rely on these tools or do you come up with your own indicators or tools? In my practice of technical analysis the past 30 years, I have focused on achieving better interpretation and integration of indicators rather than creating new indicators. There are enough good indicators out there, so why not understand
them better and make more effective use of them? The three basic needs of an investor or trader are to correctly define the trend, and based on the trend, the optimum timing and price to enter and exit the market. This means assembling a tool box of effective trend, timing, and price indicators, and then integrating them for the best outcomes. However, the investor should also be aware that his technical tool box is a dynamic one and he should constantly look out for better indicators over time. Having said that, I innovated on J. Welles Wilder’s famous indicator, the directional movement index (DMI), a few years ago by creating three clusters each of the ADX, +DI and –DI lines. [Editor’s note: For an explainer on the ADX indicator, see sidebar “The ADX.”] I named my indicator DMI clusters and its function is to identify the timing of the highest or lowest bar in a trend. For example, on the weekly chart of the NASDAQ Composite in Figure 1, you see how the market top is signaled by the ADX cluster peak at 90 and the 3X -DI cluster bottom at 5. I have since used it for five years with very good outcomes. In fact, you published my article “Identify The Start
Directional Movement -DI (15.0000), Directional Movement -DI (12.0000) Directional Movement -DI (8.00000) 50 40 30 20 - DI < 5 - DI < 5 Directional Movement ADX (71.0000) Directional Movement ADX (43.0000) Directional Movement ADX (56.0000)
3X ADX cluster turning down around 90
10
- DI < 5
0 95 85 75 65 55 45 35 25 15
3X ADX cluster turning down around 90
USA - Nasdaq 100 index (2,821.03, 2,823.34, 2,780.24, 2,784.89, -40.2202
Market top
Market top
2500
Market top
2000 1500 1000
Mar
Jun
Sep
Dec 2010
Mar
Jun
Sep
Dec 2011
Mar
Jun
Sep
Dec 2012
Mar
Jun
Sep
Nov
FIGURE 1: MARKET TOP. On this weekly chart of the NASDAQ Composite, the market top is signaled by the ADX cluster peak at 90 and the 3x ‑DI cluster bottom at 5.
38 • May 2015 • Technical Analysis of Stocks & Commodities
Of A Trend With DMI” on this topic in the November 2012 issue of Technical Analysis of Stocks & Commodities. Yes, we did. [Readers can find that article in our article archives at our website, www.traders.com.] Would you say that technical analysis has gained more respect than when you started using it in the 1980s? Yes, TA is better understood today than when I started 30 years ago. There must be hundreds of books on the subject of technical analysis alone. Its use has expanded from end-of-day to intraday markets, from off-line to online and to more markets globally than ever before. There are also many more technical indicators and approaches today compared to the past. Despite the proliferation in the application of TA among trading professionals, I still feel that TA is still not well understood by the investor community. By nature, technical analysis tends to be numeric or quantitative, and as such, it is
Key market levels of longer-term charts should still be identified and earmarked for action even though they may be further away. suited to individuals who are comfortable with numbers. Investors not inclined that way still feel more comfortable with the intuitive logic of fundamental analysis. This is why I have written my book in a concise manner so that it can be understood by investors. My hope is that with this book, the lay investor can make better and more profitable decisions in their investments. Thank you for sharing your thoughts, BC.
Stocks & Commodities, Volume 29: April. Low, BC [2014]. Integrating Technical Analysis For The Investor, Technical Analysis Consultancy. [2012]. “Identify The Start Of A Trend With DMI,” Technical Analysis of Stocks & Commodities, Volume 30: November. [2010]. “Trading, Time Frames, And Trends,” Technical Analysis of Stocks & Commodities, Volume 28: September.
Further reading
Gopalakrishnan, Jayanthi, and Bruce Faber [2011]. “Then And Now With Tim Slater,” Technical Analysis of
THE ADX The calculation of the average directional movement (ADX) indicator is built on the intuitive notion that a trend is a series of price ranges extending in a consistent direction. In sidebar Figure 1, example A, the second day’s trading range is higher than the first day’s trading range, indicating positive directional movement. In example B, the second day’s trading range is below the first day’s trading range, an indication of negative directional movement. Example C is more complicated because the second day’s range is both lower and higher than the first day’s range. Directional movement is only considered to be up, down, or not present. Therefore, the larger part of the day’s range extending beyond the previous day’s range is used to identify directional movement. In example C, the largest part of the second day is higher; consequently, the directional movement is positive. In example D, the largest part of the second day’s range is lower so that the directional movement is negative. In example E, the second day’s range is within the first day’s range so the directional movement is zero. Directional movement for the ADX is expressed as a function of true range (TR). True range is the largest of the following:
Directional Movement Examples
A
B +DM
1
C
D +DM
1
2
2
2 1
1
-DM
2
-DM
12
1 The difference between today’s high and today’s low
E
2 The difference between today’s high and yesterday’s close 3 The difference between today’s low and yesterday’s close. In the Excel 4.0 spreadsheet (sidebar Figure 2), the first calculation for ADX is the true range value. This is performed in column E. The formula for cell E3 is: =MAX(B3-C3,ABS(B3-D2),ABS(C3-D2))
SIDEBAR FIGURE 1: DM EXAMPLES
Next, column F determines the positive directional movement or returns zero if there is no positive directional movement. The formula for cell F3 is: May 2015
• Technical Analysis of Stocks & Commodities • 39
SIDEBAR FIGURE 2: MICROSOFT EXCEL
=IF(B3-B2>C2-C3,MAX(B3-B2,0),0) Column G calculates the negative directional movement or returns zero if there is no negative directional movement. The formula for cell G3 is: =IF(C2-C3>B3-B2,MAX(C2-C3,0),0) The daily calculations are volatile and so the data needs to be smoothed. First, sum the last 14 periods for TR, +DM and ‑DM. The formula for summing the TR is in cell H16: =SUM(E3:E16) The formula for summing the +DM is in cell I16: =SUM(F3:F16) The formula for summing the -DM is in cell J16: =SUM(G3:G16) The smoothing formula for the TR14 column begins at cell H17: =Round((TRUNC((H16-(H16/14)+E17),3),2) The smoothing formula subtracts 1/14th of yesterday’s TR14 from yes‑ terday’s TR14 and then adds today’s TR value.The rounding((truncating function is used to calculate the indicator as close as possible to the developer of the ADX’s original form of calculation (which was done by hand). The smoothing formula for the +DM14 column begins at cell I17:
to calculate the ratios of +DM and ‑DM to TR. The ratios are called the +directional indicator (+DI) and ‑directional indicator (‑DI). The formula for the +DI column begins at cell K16: =Round((100*(I16/H16)),0) The formula for the +DI column begins at cell L16: =Round((100*(J16/H16)),0) The INT (integer function) is used because the original developer dropped the values after the decimal in the original work on the ADX indicator. The next step is to calculate the absolute value of the differ‑ ence between the +DI and the ‑DI. This is done in column M and the formula for cell M16: =ABS(K16-L16) The next column calculates the sum of the +DI and ‑DI. The formula for cell N16: =K16+L16 The next step is to calculate the DX, which is the ratio of the ab‑ solute value of the difference between the +DI and the ‑DI divided by the sum of the +DI and the ‑DI. This is done in column O. The formula for cell O16: =Round(100*(M16/N16)),0)
=Round((TRUNC((I16-I16/14)+F17),3),2)
The final step is smoothing the DX to arrive at the value of the ADX. First, average the last 14 days of DX values. The formula for cell P28: =AVERAGE(O15:O28)
The smoothing formula subtracts 1/14th of yesterday’s +DM14 from yesterday’s +DM14 and then adds today’s +DM value. The smoothing formula for the -DM14 column begins at cell J17:
The smoothing process uses yesterday’s ADX value multiplied by 13, and then add today’s DX value. Finally, divide this sum by 14. The formula for cell P29:
=Round((TRUNC((J16-(J16/14)+G17),3),2)
=Round((((P28*13)+O29)/14),0)
The smoothing formula subtracts 1/14th of yesterday’s ‑DM14 value from yesterday’s ‑DM14 and then adds today’s ‑DM value. Now we have a 14-day smoothed sum of TR, +DM and ‑DM. The next step is 40 • May 2015 • Technical Analysis of Stocks & Commodities
—S&C Readers can find a copy of this ADX spreadsheet file at our website, www. traders.com, in the Article Code area in the May 2015 issue listing.
Explore Your Options Got a question about options? Tom Gentile started his trading career on the floor of the American Stock Exchange in 1994. He has appeared on many financial TV and radio shows, as well as hosting a weekly talk show himself, and has co-authored many books on the markets. He can be found at www.tomgentile.com. To submit a question for Tom Gentile, post it to our website at http://Message-Boards.Traders. com. Answers will be posted there, and selected questions will appear in a future issue of S&C. Tom Gentile
BUYING INSURANCE Which way will oil and energy prices go as we head into the spring and summer months, and how can you trade these markets? As spring creeps up on us, so does the winter thaw out of the Northeast. For much of the winter season, record amounts of snowfall blanketed the northeastern part of the United States, with some storms so fierce, they were given names such as Pandora and Thor. Those poor Nor’easters, as I like to call them, are happy that the winter season of 2014–2015 is finally behind them. So much so that “cabin fever” is in full effect from North Carolina up to Maine. People are eager to get out and travel, and this should bode well for oil and gas demand. This year presents us with different problems that were not around last season. For instance, there is a much greater supply of oil worldwide, so perhaps prices will shift up and down more on the economy and off of international growth than before. The move in oil over the last year presents risks that were unseen in years, no matter how much the demand for oil might occur this summer. How can a trader capitalize on this potential opportunity with minimal risk? If there’s one thing that oil markets are yielding these days, it’s uncertainty. All of this uncertainty about the future price of oil has option premiums exploding this year versus last year. Looking at the chart in Figure 1, this tells an option trader that premiums on USO options look to be three times higher than at this time last year. That’s like seeing your car insurance go up triple for the same coverage. Good for the insurance salesman, bad for you. The one thing we don’t want to do as option traders is buy expensive
options. So what’s a trader to do, when opportunities present themselves in a market such as oil and energy? Become the insurance salesman The best way to capitalize on this opportunity is to become an option seller. Selling options increases your probability of being right over time, but the risks if you are wrong increase as well, and in some cases can be unlimited risk. So how do insurance companies stay in business after a hurricane blows through and everyone
starts calling in claims? Easy. They cut their losses by reinsuring with a bigger company. So for instance, let’s say that ABC insurance sells $50 million in home insurance and they want to hedge their risk. What they do is perhaps reinsure so that if claims go over $100 million, they are covered from that point on. Their risk is between $50 and $100 million, but no more. You want to do the same thing with options. In Figure 2 is an example Continued on page 45
Figure 1: exploding option premiums. Premiums on USO options look to be three times higher than this time last year.
Figure 2: hedging your risks. Here you’re selling the 16 puts and hedging the 14 puts for a combined credit of 0.50 or $50 per spread. A credit spread has limited reward and risk. The risk is that USO drops in value, causing the price of puts to rise. The most you can lose is the difference between strikes minus the credit received. May 2015
• Technical Analysis of Stocks & Commodities • 41
16
product review
Haguro Method METASTOCK 90 South 400 West, Suite 620 Salt Lake City, UT 84101 Phone: Sales 800 508 9180, support 801 265 9998 Email:
[email protected],
[email protected] Product: MetaStock add-on based on the Haguro method System requirements: Windows 7 and above, 2.4GHz or higher, 4GB RAM, 1.6GB disk space High Price: Free price
T 15
by S&C Staff Middle
he Haguro method ofprice chart analysis was written about by Seiki Shimizu in his book The Japanese Chart Low price Of Charts, and we published by 13an article 12 10on the 14 method 11 Gary Burton in the April 2015 issue of Technical Analysis of Stocks & Commodities. The method has now been adapted as a MetaStock add-on, which is available to MetaStock users at no additional cost. You can download the add-on from www. metastock.com/haguro. Burton, director of the Australian School of Technical Analysis, had translated Shimizu’s work, and Jeff Gibby, Business Development Manager for MetaStock, saw a presentation by Burton on the method while attending a conference in Australia. Since Gibby manages the development of new software for MetaStock and also trades the market himself, Gibby is well-qualified to evaluate new methods and strategies. What he saw in the Haguro method was unique, he thought, and could be a potentially profitable method to use.
Overview High
High
Here, we see the open & close below the midpoint of the line.
Open
Close
Close
This example shows the open & close above the midpoint of the line.
Open
Low
Low
Figure 1: Candlestick Versus Midpoint. Candlestick patterns are given a specific number to identify them. There are two groups with eight patterns in each. One group will be based on bodies where the close is greater than the open (the green candle on the left). The other group uses candlesticks where the body is red (that is, closeopen) or red (close
