March 2004 Letters To The Editor

or return to March 2004 Contents

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.

Address your correspondence to: Editor, STOCKS & COMMODITIES, 4757 California Ave. SW, Seattle, WA 98116-4499, or E-mail to editor@traders.com. 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

ON-BALANCE VOLUME

Editor,

I enjoyed D.W. Davies' January 2004 article, "Daytrading With On-Balance Volume." As a point of reference, I need to point out that while Joe Granville (for whom I have utmost respect as a technician and friend) popularized the term on-balance volume, the idea was originally called cumulative volume and was presented in a course written by Woods and Vignolia in San Francisco in 1946.
Larry Williams

Thank you for pointing this out.

Editor's note: Larry Williams is author of Definitive Guide To Futures Trading, as well as The Secret Of Selecting Stocks For Immediate And Substantial Gains, and other works. He was an early contributor to STOCKS & COMMODITIES with "The Ultimate Oscillator" (August 1985) and was interviewed in the June 1987 S&C ("Larry Williams: Where Will He Go Next?") and again in July 1997 ("Contextual Trader Larry Williams").-Editor

BOLLINGER BANDS

Editor,

Regarding the letter to the editor that appeared in the January 2004 S&C from Steven Schon on the topic of Bollinger Bands:
Low bandwidth, which can be defined as:

... is indeed useful in predicting higher volatility, but as Schon correctly observes, bandwidth in and of itself contains little in the way of a directional forecast. We have found volume indicators such as David Bostian's intraday intensity or the money flow index (detailed in STOCKS& COMMODITIES, March 1989, Vol. 7, No. 3) to be useful in helping to anticipate direction in the wake of a "squeeze." In addition, certain price patterns, such as the head fake, have proven to be helpful too.
In his letter, Schon asked you for suggestions for websites that might feature lists of stocks with low bandwidth. We wanted to mention that our website, www.BollingerOnBollingerBands.com, might provide what he is after. It offers prescreened lists of typical Bollinger Band setups such as the squeeze -- which is our nomenclature for low bandwidth -- as well as custom technical screening. The site also offers charts, M&W pattern analysis, portfolio monitoring, and a number of other features for the Bollinger Band enthusiast.

Thanks for your fine magazine. I remain, as always, a thorough reader,
John Bollinger, CFA, CMT

John Bollinger has thrice been interviewed in S&C: in July 1993, April 1997, and May 2002, and has contributed two articles to S&C: "Using Bollinger Bands" (February 1992) and "Put Volume Indicator" (March 1993). He is author of Bollinger On Bollinger Bands.-Editor

40-WEEK CYCLE IN THE STOCK MARKET

Editor,

Jay Kaeppel's article "The 40-Week Cycle In The Stock Market" in the January 2004 S&C reveals arguably the best cyclical analysis of the Standard & Poor's 500 I have seen to date. However, I was surprised that the S&P was the only index used as supporting evidence. Using my LIM data/software, I was able to quickly recreate Kaeppel's analysis and then test for normalization across other US stock market indexes. I found very similar (if not better) bullish phase results:

Spiders would have yielded 4.95% (average) since 1/12/2001, and diamonds would have yielded an average of 8.45%! I can't wait to see what happens at the end of the 5/2/03 cycle!
Tomas Kornegay, via email

MORE ON THE 40-WEEK CYCLE

Editor,

In my work on longer cycles, I have found that they are not symmetric, probably because of the ongoing secular or supercyclical bull market we have been in for more than 200 years. I find a four-year cycle, for example, has to be treated as being positive for 30 months and negative for 18 months.

I also find, as Kaeppel did, that for shorter cycles, the negative periods are really neutral, again because of the upward bias.

However, the key point is that the 40-week cycle might really be positive for something like 24 or 30 weeks and negative for the balance. If he hasn't considered that, he should.

Bob Peirce, via email
Venetia, PA

QUESTIONS ABOUT THE 40-WEEK CYCLE

Editor,

Thanks for the interesting article on cycles in the January 2004 S&C ("The 40-Week Cycle In The Stock Market").

On page 32, author Jay Kaeppel states, "Our analysis of the 40-week cycle starts on May 15, 1970..." I want to ask, why that date? What is important about May 15, 1970? Why not another start date? How was this date selected? Was this 40-week cycle backtested and fitted to the best start date? Does the theory hold up if another start date is used? I would appreciate further discussion of the selection process for the start date. Seems to be an important part of the cycle theory.
Mike Ruff Pampa, via email

Jay Kaeppel replies:

I don't recall where I first heard about the 40-week cycle. Cycles get bandied about a lot in financial literature (and much of it does not seem to hold up), and somewhere, somebody mentioned the "40-week cycle is due to bottom" on such and such a date. This was probably 10 or more years ago. I tested this one at the time, and it has held up reasonably well going forward.

I have not tried testing it with other starting dates. My guess is that if you tested every day in 1970 as a start date, you would probably find a better optimum start date, but I have essentially left well enough alone.

One last note: I do not trade based solely on this cycle. I do, however, use it as a factor in deciding whether to be bullish or bearish.
I hope this answers your questions.

DETECTING BREAKOUTS

Editor,

First, I would like to thank Markos Katsanos for those two great articles on detecting breakouts ("Detecting Breakouts," April 2003, and "Detecting Breakouts In Intraday Charts," September 2003). Without exaggerating, I may say that they were two of the best articles I have read in S&C.

I have been studying the markets and developing trading systems over the past two years and I am very interested in gaining experience and new ideas in this field.

After studying the system's code in "Detecting Breakouts In Intraday Charts," I have a couple of questions that were not answered in the article. First, in the simulation report, different values are set in the upper and lower bound fields for each time frame. However, what isn't explained is what those values derive from. Is there a formula as in the case of the cutoff amount and length of "period," or is it simply a result of simulation and optimization? If it is a result of optimization, isn't there a large risk of curve-fitting the system?

One more thing: Which parameters does Katsanos suggest optimizing when testing the system? I guess the number shouldn't be larger than three or four in order for the system to be robust across different markets. Have you tested the system over a range of markets (for example, stocks, futures, bonds, and so on) or do you suggest considering it to trade stocks only?

Finally, why does the indicator only trigger long entries? Is there some technical difficulty, or is it simply that Katsanos designed it for trading stocks?
George, via email

Markos Katsanos replies:

Thank you for your kind words about my article.

The upper and lower bound values for minute charts have not been optimized but are adjusted according to the square root of time (formula 3 in the article) from the corresponding daily value. For example, the upper bound value for the 15-minute chart is 0.6*sqrt(15/390)=0.12, where 0.6 is the corresponding value for daily charts, and 390 is the number of minutes per trading day (6.5*60). You can optimize the linear regression angle values, as these will vary from stock to stock.

An FVE-based system, as long as you have correct volume data, works with indexes, but I have not tested it with futures. Take a look at https://www.trading-konzepte.de/indikator/fve.htm and how FVE predicted the German DAX's spectacular advance three months in advance. (I have no connection with this German website.)

You should also take market conditions into consideration. This particular system worked well at the time I wrote the article (May 2003), but I don't get as many hits now (at least in daily charts), since most stocks have already broken out from bases (the stock under test, ATML, has tripled since). I suspect that a momentum-based system might work better under current market conditions.

As for short-selling opportunities versus long entries: FVE is a leading indicator, and this is a problem if you want to use it for short-selling. If you buy a stock early, you will only need some patience, but if you sell short too early, you also need deep pockets, otherwise you will be stopped out. This is because as the stock goes up, buyers become less enthusiastic, volume diminishes, and Fve turns down and triggers a short sale signal. At this point, momentum players step in and the stock continues to go up for some time.

If you want to use FVE with a short-selling system, it is best to combine it with a lagging indicator such as MACD or moving average.

TRADERS' TIPS

Editor,

Just completed the January issue. I've been reading your magazine for a long time and wouldn't be without it.

My reason for writing is to question your reasoning for including so much programming material. I counted 18 pages in the last issue. In how many of those pages do you think I would be interested? Of the 18, probably 16 are a waste of my time and money.

Most of us have only one of the programs involved. I highly recommend that a reference to each be included but no macros or programming. Think about it. Would you manually copy the material from the magazine, or go to the website to copy and paste? Much less potential for error and a lot easier and faster. Is there an advertiser($) reason for including this material in the magazine? I think this would be an improvement.
RB Millar, via email

We have included code in the magazine since the very first issues, since STOCKS & COMMODITIES readers have always been interested in code. It is part of our editorial format to try to include code, to make implementation of techniques easier. But the code isn't printed in the Traders' Tips section with the intention of it being retyped by readers; in fact, all the Traders' Tips code is posted monthly at our website for copying and pasting. We print it only because some readers like to browse the code to see the logic in the programming as part of studying the technique, and to see how the various software implements a given technique.

But we haven't cut back on articles to include code; instead, we increased the page count in part to include the section. Two years ago, we typically published a 108-page magazine; these days, the magazine is typically 124 pages or more.

We appreciate your feedback and would be interested in hearing from other readers as well at Editor@Traders.com. We could switch to posting the code at our website only.-Editor

VOLUME-WEIGHTED AVERAGE PRICE

Editor,

An article appeared in the May 2001 S&C titled "Volume-Weighted Average Price" by George Reyna. This article appeared before I subscribed to S&C, but I was so interested in the article that I purchased it from your website. Unfortunately, it did not contain any coding for analysis packages that you usually publish in the Traders' Tips section. I looked on your website in the hope the code would be there, but I can't locate it. Could you direct me to a source of code for TradeStation 2000i? Whilst there is code for VWAP on TradeStationWorld.com, I'm concerned that it is not the code that appeared with the article.
Dave G., via email

The Traders' Tips section that appeared in the May 2001 issue was based on that article; however, TradeStation didn't submit code for that technique that month. On the other hand, Reyna did provide an Excel spreadsheet with code for the VWAP in his article. Please compare that code with the TradeStation 2000i code you found at TradeStationWorld.com and see if they are the same formula.-Editor

HEAD & SHOULDERS PATTERN

Editor,

The head and shoulders pattern is formed not only on price but on momentum as well. But is there any signficance of this pattern on volume behavior? Attached is a chart for your review (Figure 1).

FIGURE 1: Chart submitted by reader.
Syed Rehan Ali, via email

A head and shoulders pattern can be either a reversal or continuation pattern. In the case of a reversal to an uptrend, the volume on the head and the right shoulder should be less than the volume on the left shoulder. Otherwise, similar volume on the head and right shoulder should alert the analyst that a continuation of the uptrend is possible. The key is that uptrends that exhibit subsequent peaks on lowered volume are likely to either reverse or go sideways -- hence, the volume would not have the characteristic head and shoulders profile.-Dennis Peterson, Staff Writer

ERRATA: TRADERS' RESOURCE

In the Traders' Resource section on page 111 of the February 2004 S&C, the "Top 10 Viewed" box incorrectly refers to the Trading Systems category instead of the Exchanges category. We regret this error.

ERRATA: ANATOMY OF A CANDLESTICK

Editor,
In my article "Anatomy Of A Candlestick" in the December 2003 S&C, the wrong candle was highlighted in Figure 1 when my article was published. The doji from December 30 was circled, not the December 31st hammer candle, as it should have been.
Clive Lambert, via email

Back to March 2004 Contents