Friday, April 28, 2006

Getting Started In Chart Patterns by Thomas Bulkowski: a book review and interview with the author

Getting Started In Chart Patterns by Thomas Bulkowski. A book review and interview with the author:

Pro's:

  • Chart pattern discussions in the book is grounded in statistical performance numbers. The chart pattern performance statistics can also be viewed in table form in the back of the book and on the companion website.

  • "Index of Chart and Event Patterns" near the end of the book is a very useful reference. Maybe we can petition Tom to make a wall poster out of it.

  • Numerous trading case studies of Tom's real trades, from entry to exit. These include excerpts from Tom's personal trading journal and round trips plotted on charts.

  • A chapter on "Busted Chart Patterns" that is probably new material for most seasoned chart pattern traders.

  • Tom identifies many potential traps and pitfalls throughout the book, and also concentrated in the chapter, "The Art of Trading". I found myself agreeing with basically everything Tom discussed in this chapter.

  • Chart pattern recognition software available on the companion website.



Con's:

  • Some sections in the middle of the book are difficult to follow because the referenced chart figures appear on the following page due to space restrictions.

  • Not much discussion of why chart patterns form or similarities between patterns.

  • No "find the pattern" quizzes for the reader in the book, but there are quizzes at the companion website.



Here's a link to the Tom's chart patterns companion website. It is also in the links on the right side of the page.

I use chart patterns for "digging in" to trending stocks, so Tom's book was a useful one for me. My biggest suprise was the chapter on "busted" chart patterns, because like any other chart pattern trader, I've been victimized by busted patterns dozens of times. This chapter is a good study of how to turn these events from losses into wins, especially if you've never been introduced to them formally.

To give you a flavor of the book, here is a quote from the chapter "The Art of Trading" (page 261) where Tom describes his trading style as an introduction to the useful guidelines that follow:

  • "I'm an end-of-day trader, a position trader. I don't day trade stocks but hold them for weeks, months, and sometimes years. I review the stocks that I follow, and if I don't see anything interesting, I'm done for the day. I can go weeks without trading a stock, and I usually spend about one hour each day looking for trades or updating my database. The rest of the day is free time."


Sounds familiar... It takes a lot of discipline to trade this way, but that is why the rewards are so great.

Here is a quote from the section on flags and pennants (page 154), which I think is particularly relevant for readers here:

  • "Patterns with throwbacks or pullbacks have postbreakout performance that suffers. For example, when a flag with an upward breakout has a throwback, prices rise 14% after the breakout. Those without throwbacks climb 26% on average."


I asked Tom if he was interested in doing an interview with me to go along with this book review, and fortunately he accepted! Thank you for the opportunity, Tom.


  • Jon: You have compiled a table of statistical expectations for dozens of chart patterns in the book ("Getting Started In Chart Patterns"). When trading these patterns, how big of a factor are the statistical expectations in your position sizing relative to total equity?

    Tom: None. I take $15,000 and divide it by the stock’s price and then round up to the nearest 100 shares. That tells me how many shares to buy, and it places the trade in the $15k to $20k range. Nothing fancy here. I have found that I can buy enough positions such that I begin to sell stocks as I buy new ones (rotation), leaving me with a comfortable float for new positions without running out of cash.

  • Jon: Some of the chart patterns discussed in the book occur more frequently than others. When you find a rare gem like the high, tight flag, will you allocate a larger position size than you typically do to trade common patterns such as the symmetrical triangles?

    Tom: No. I want to maintain diversity among my stock picks/portfolio so I’ll just buy the usual amount regardless of the type of pattern. Occasionally, I’ll increase my holdings in the stock, 2x, 3x, or even higher if I see value, usually averaging in at a higher price but not always. Sometimes, if I feel the market might tumble or doubt whether the trade will work as expected, I’ll trim the position size (I usually round down instead of up in the shares calculation).

  • Jon: Do you pyramid in or average out of your positions? If so, does it depend on the pattern, and can you describe how you prefer to do it?

    Tom: Scaling in or out (pyramiding) is not pattern related. I do scale into a trade but that’s rare, and I almost never scale out (the exception being a long term holding. I scale in and out frequently, selling the most recent shares first to keep the taxman away from potential long-term capital gains). If I buy a stock on the breakout and it throws back to the breakout price and then starts moving up, I might buy another $15k. But, once I get the sell signal, I’m out. I’ve found that scaling out means I lose more money than if I just sell the whole thing at once.

  • Jon: When trading a familiar pattern, you have a pretty good idea going into the trade how it has performed in the past, as described in your book. In your own trading, have you found that you do better overall by going with more or less aggressive targets than the average? How about with stops?

    Tom: It depends on what you mean by target. There’s a price target and there’s a chart pattern target, searching for a particular chart pattern to trade. For a price target, I just use the measure rule (the height of the chart pattern added to the breakout price) as a price target tempered with how close overhead resistance is. Resistance is a better predictor of how the trade will do than most any mechanical calculation.

    For chart pattern targets, I don’t scan for a particular chart pattern. Chart patterns are just buy and sell signals. They are a starting point to an analysis of each trade. I have learned to dislike ascending triangles, for example. They tend to fail more than anyone expects despite breaking out upward 70% of the time (that’s not a guess: I measured this). Now, a busted descending triangle with a downward breakout, meaning price drops less than 10% before reversing and shooting out the top of the triangle, is better than having sex. Well, maybe not. I’ll have to test that some more and get back to you.

    Stops: Setting a stop is an art. I found that you just can’t use one type of stop. In a study I did recently, I found that using a volatility stop improved 63% of my trades. Almost 2 out of every 3 trades would make more money or lose less if I just used a volatility stop. The bad news is that a volatility stop would have cut profits almost by a factor of 10. Ten! In other words, it helps, but it also can cash you out of a potential winner.

    I like to hide under minor lows. If the stop is too far away, then I’ll use a volatility stop (I compute the high low range of EACH day over the prior month, average the results, and multiply by 2. Then I subtract the result from the current low price to get the stop price – that’s a volatility stop. And before you ask, I found that the high-low calculation performs better in my tests on actual trades than the average true range and standard deviation). My Patternz chart pattern recognition program (free on my website) calculates it automatically with a mouse click. Sometimes I’ll duck under a 62% retrace of the prior up move (think Fibonacci) providing it doesn’t place the stop too far away.

    A trader wannabe I know puts his stops 1% or 2% below the current price and keeps being stopped out. He’s losing a fistful of money, but the brokerage firm loves him for the commissions. I tried to get him to widen his stop, but he won’t. I haven’t heard from him recently, so he may have flamed out. Moral: Learn from your mistakes.

    As to aggressive stops, I’ll start out with them wide, say 10% or 15% not out of choice, but closer targets in low priced, highly volatile stocks are hard to find and you risk being stopped out quickly. As price climbs, I’ll move the stop up so my potential loss closes fast. The bad news is that when I’m wrong, it could cost me dearly. Last year, my average loss was 6.9% and my average winner was 19.5%. That’s almost 3 to 1 win/loss ratio. At one point, I had it up to 5 to 1 but lost control of one trade. That happens.

  • Jon: You mentioned that market trend is a considerable factor with how you place your trades, stops, and targets. Are there any trend analysis resources (books, websites, etc.) that you have found to be particularly helpful?

    Tom: Let’s exclude your blog…My favorite site is finance.yahoo.com. That’s just for general info on a company, not market trends. I do my own research, well, guess really, and take it from there. I post my research findings on my website (http://mysite.verizon.net/resppzq7/). It’s a great resource for studies, chart patterns, and my Patternz program. Other than that, I don’t go search the web for that trend analysis information. Price tells almost everything you need to know, so if you look at the S&P 500 or Nasdaq trends, and trade in the direction of those trends, you should do well.

  • Jon: Do you find chart patterns to be more predictable of future price direction in large cap or small cap stocks? If so, is there anything noteworthy that you attribute to the difference in reliability?

    Tom: My Trading Classic Chart Patterns (Wiley 2002) book explores this question for the major chart patterns. I found the obvious: in a rising market, you want to be in the small caps (less than $1 billion in market cap). In a falling market, the large caps (over $5 billion in market cap) hold up best.

    For example, most people know what a double bottom is. On page 163 of the TCCP book, the table shows that for an Eve & Eve double bottom (a special type of double bottom describing wide, rounding turns for each bottom), small caps rose 39%, mid caps climbed 38%, and large caps climbed 33%.

    A head-and-shoulders top, which is a bearish pattern, shows small caps dropping 22%, mid caps 21%, and large caps 19%. Large caps hold up best, small caps drop farther. The sample counts are smaller than I would like, but that’s how the trends play out, on average.

    At the back of the book, page 425, I show a table of which market cap does best in a falling market. Small caps win 8 out 13 times, mid caps four times, and large caps once. That means small caps decline most in a falling market, mid caps are in second place and large caps decline least (assuming that you wanted to short a stock, so a large decline is good).

    On page 416, I shows the results of upward breakouts: small caps tie (twice) or win 11 out of 13 cases, mid caps win twice and large caps tie twice. Small caps rise most, large caps barely register.

    Knowing these types of statistics give you a trading edge over other traders.

    As to the reliability question, you’re talking about failure rates. I define a failure as a count of how often price after a chart pattern breakout fails to rise or decline more than 5%, what I call the breakeven failure rate. And, gulp, I haven’t looked at failure rates according to market cap. I only looked at it in terms of performance. My guess is the results will be the same because the failure rate is just a measure of performance anyway (price climbs by less than 5%).

  • Jon: If you see a similar pattern developing at the same time on industry related stocks, does that give the pattern a higher chance of following through and reaching its target?

    Tom: The honest answer is I don’t know. Certainly, when I see multiple bogies (chart patterns) flying around, I can pick and choose which ones to shoot at (trade). If many of them are bearish patterns, then I’ll want to tighten my stops of any holdings or avoid buying anything in that industry.

  • Jon: Is there anything you would like to say to the critics of technical analysis and chart patterns who might be on the fence about the predictive value of those techniques?

    Tom: Chart patterns, like technical analysis in general, works for some people and not others. Find something that works for you and use it. When it drifts (begins to lose profitability) then tune it. You’ll have to do that from time to time. The markets are not static and your strategy shouldn’t be either.

    A novice trader once emailed me and said “I made 10 triangle trades and 9 of them were profitable!” What did he do? He threw away his winning strategy and sunk his money into one stock. That stock was a gold mining scam, and he lost everything. Trading isn’t entertainment, and that’s not the type of tuning I’m talking about. Try small changes with lots of testing.

2 Comments:

At 9:41 AM, Blogger Michael Taylor said...

Nice review and interview.

Great job!

MT

 
At 7:41 PM, Anonymous Bert said...

I just read Thomas N. Bulkowski's book titled Getting Started in Chart Patterns. In his book and on his website he described how he calculates a volatility stop but I couldn't find an automated calculater to do this anywhere on the web so I created one. Here's the link to my Stock Volatility Stop Calculator.
-Bert

 

Post a Comment

<< Home