 |
|
August 1998 Contents
DOW THEORY
Editor,I especially enjoyed your interview in the June 1998 STOCKS & COMMODITIES ("Before and beyond technology: Michael Sheimo on Dow theory") because Sheimo comments on the Dow theory in light of the much larger volume and velocity of trades today as compared with the late 1800s, when Charles Dow developed his theory.
Based on other articles that have been published in your magazine and based on my own research, I've concluded that Dow theory has not lost its relevance to market behavior in spite of massive changes in trading volume and velocity since the 1800s. Although electronics and an increase in the number of traders and size of trades have increased volume and velocity, the behavior of market trends still follows the old rules, although perhaps at faster rates. Thus, market history and some of the older trading rules remain relevant to traders today. I believe that studying the famous market events, excesses, and theories of the past can make one a much better trader today.
Please continue to offer articles commenting on and updating classical as well as new investment theories.
RANDALL COVILL
via E-mail
SOURCES FOR TREASURY BOND PRICES
Editor,I've just read John Murphy's Intermarket Technical Analysis. In it, he shows that an important index to follow is Treasury bond prices. I have looked at data sources and am confused as to what data would best represent T-bond prices. Some I've found include:
- In Barron's, Market Laboratory, bonds
- Lehman Brothers T-bond index
- Ryan Labs Treasury index
- 30-year T-bond percentage yield
- From Dial Data, index.doc
- Five different files for Lehman T-bond index
- Six different files for Shearson T-bond indices
Can you help me determine which one or more of these would be appropriate?
And just to add to the confusion, you should also include the T-bond futures contract traded at the Chicago Board of Trade as an additional source for Treasury bond prices. This contract data has the advantage of being disseminated throughout the trading day, and historical data is readily available.WOODY HAUTER via E-mailBut no matter which price data you use, the key is to use the same price series for all your research, and if you want to follow long-term Treasury rates, make sure the index you follow is a proxy for a 30-year maturity.--Editor
ESTIMATING THE AVERAGE
Editor,I'm a long-time reader of your magazine, and this year I've become a subscriber.
I'm very interested in the article "Applying moving averages" by Technical Editor John Sweeney (STOCKS & COMMODITIES, April 1998), because I've been searching for years for patterns to use for placing low-risk entry and exit orders, also using moving averages.
In his article, Sweeney tells us we can place a buy order the night before at the level of a moving average, leaving us free for the day. But I have one problem: The moving average value is known only after the close of the market, not before. Is this a "zigzag trap"?
Good work and good trading!
You're correct; you cannot know precisely where the average will be, but you can get pretty close, especially if you're using simple moving averages, where you know the value dropping out of the average. Personally, I usually predict the next day's average by looking at the last several days' changes in the average's values. This technique has been criticized as imprecise, but I feel the numbers are so rough that this is an acceptable approximation. Hope this helps!--Technical EditorANGELO VALENZANO Bari, Italy
IMPACT OF VOLATILITY
Editor,To me, the results presented in Technical Editor John Sweeney's April 1998 article ("Applying moving averages") on high versus low system volatility are misleading. In analyzing the effects of volatility, the author uses an input set of trade returns that averaged zero percent, with a range of -50% to 50%. I believe this is an unreasonable set of inputs. After all, who would trade with any system that accepted losses as large as -50% (with no money management rules)? Further, who would trade with a system that accepted a 0% average trade return?
Through the use of stops, a trader can limit losses on a losing trade to, say, 15%, while letting the winning trades continue unhindered until the exit point (such as a profit target or sell signal) is reached. In fact, STOCKS & COMMODITIES has included several articles on this very subject during the last few issues.
A better real-world assumption for a study like this is to allow, say, a -15% to +25% range, with an average trade return of 5%. The results of sequential trades in this case clearly favor the higher volatility system. Using Microsoft Excel, I ran a 20-trade sequence 200 times, and the higher volatility system won 191 times out of 200 (95.5% of the time). I believe the author should have reported the results of a study like this using various ranges of returns, to show the sensitivity of the analysis to varying average trade returns.
While I don't think the article was misleading, I commend using other assumptions to assess the impact of volatility on ending wealth. Perhaps you would like to share your work with our readers via an article.--Technical EditorLOREN FORBUS via E-mail
TIME SERIES SMOOTHING
Editor,A couple of observations have led me to a better smoothing rule for time series of prices, which may be of interest to others. The observations are:
1) When both the high and the low move in the same direction, the immediate price direction of the close appears relatively less important.
2) When the high and low move in opposite directions, the immediate price direction of the close appears relatively more important.
Based on these observations, one can compute a composite value of close (CC) that more nearly represents the true direction of price movement, minimizing noise and whipsaw. When the high and low are trending in the same direction, compute the composite (CC) as the average of the high and low. When the high and low are moving in opposite directions, compute the composite (CC) as the close.
These observations can be easily implemented in Microsoft Excel. Beginning in row 5, column E, compute the composite close (CC) column as:
=IF((B5-B4)*(C5-C4)>=0,(B5+C5)/2,D5)
where high, low and close are in columns B, C and D, respectively, beginning in row 4.
If, in addition, the composite close (CC), column E, is smoothed by applying a two-three period exponential moving average (EMA) or a five-six period end point moving average (EPMA), a superior smoothed signal will result. The EMA has more amplitude smoothing and more time lag. The EPMA has less amplitude smoothing and less time lag. A 50%-50% average of the two gives good results, for adequate amplitude smoothing and minimum time lag.
Another really good smoothing formula to apply to CC is Tim Tillson's T3, which was described in Tillson's January 1998 S&C article, "Smoothing techniques for more accurate signals."
Basing indicator, rate and momentum calculations on the smoothed CC may give better timing signals.
DON KRASKA, via E-mail 75664.2714@compuserve.com
T3 INDICATOR
Editor,I am writing to pass along an observation about Tim Tillson's January 1998 S&C article, "Smoothing techniques for more accurate signals." In my correspondence with him, Tillson suggested I share the observation with other S&C readers.
In reading the article, I noticed that the formula for generalized double exponential moving average (GD) could be restated as GD(n) = EMA(n)*(1-hot) + DEMA(n)*hot. In other words, GD is a weighted average of EMA and DEMA. Since EMA has a lag of (n-1)/2 and DEMA has a zero lag, I inferred that GD's lag is a weighted average of the lags of its components, or (n-1)*(1-hot)/2. I then concluded that applying GD to itself m times has the effect of multiplying the GD lag by m.
So for Tillson's example of n=5, hot=0.7, I estimate the T3 lag at 3*(5-1)*(1-0.7)/2 = 1.8. Tillson noted that this matched his empirical conclusion about the T3 lag.
GEORGE HERBERT via E-mail
ENTRY TIMING FOR DOGS OF THE DOW
Editor,Has STOCKS & COMMODITIES ever published an article giving a backtested statistical study to determine the best time of year (or month) to begin investing using the "Beat the Dow" method (that is, choosing the five highest-yielding/lowest priced Dow stocks)? If not, do you plan to do so in the future?
This sounds like a good article topic that we haven't addressed yet. We welcome article submissions on such subjects, in case anyone reading this has some thoughts on this or related topics.--EditorDAVE MELE via E-mail
CONFIRMING INDICATORS
Editor,Although I am relatively new to technical analysis, I have already read a fair bit about the subject, including about three years of STOCKS & COMMODITIES. In reading books, software manuals and articles, I frequently come across statements that go something like this:
- The Chaikin oscillator should more often be used to confirm other technical forecasting indicators.
- Combining the directional movement index with other directional tools can increase the odds.
- In any market, confirmation signals from other indicators should be noted.
- The indicators should be used in many cases to confirm and be confirmed by another.
- As always, consult other indicators for confirmation.
I need an explanation of how to use confirming indicators. I can't believe that using one indicator by itself is a wise or safe method of making money in the market. I know about confirmation, but what combinations of indicators work well together in software, and how is this best put into practice? My experience is too many signals get returned or the combination is too strict. The quality of trades varied from poor to middling.
If you have done similar articles, maybe you could remind me and other readers of when.
Review most of the indicators in your charting program and gain a solid working knowledge of them. Then use a set of indicators that each measure a different aspect of the market. For example, employ one momentum indicator, one trend indicator and one volume-based indicator to get a complete picture of the market.DONALD DALLEY Toronto, CanadaA simple test of indicator redundancy is to export the daily indicator values for each into Excel, then measure the correlation of the daily values. If two indicators have a very high correlation, then chances are good that the indicators are measuring the same aspect of price action and are redundant; in other words, you're not gaining additional information from using both.
Finally, consider using the same indicator but apply a multiple time frame approach. Check out Robert Krausz's article, "Dynamic multiple time frames," in our November 1996 issue. --Editor
INTERNET BROKERAGES
Editor,How do you check out the validity of an Internet broker? Is there a place we can call to see if some of these people are real before sending them money? Are the accounts covered by some kind of insurance in the event the brokerage house goes out of business?
Visit the US Securities Exchange Commission's Web site for information on what to look for in brokerage services and what questions to ask. The SEC Internet address is https://www.sec.gov.--EditorKAMBIZ HEIDARIAN Roswell, GA
ERRATA
In our June Books for Traders, we printed incorrect contact information for publisher TradeWins. For the book How to Become a Real-Time Commodity Futures Trader From Home, the listing should have read:We regret the error.--EditorTradeWins Publishing PO Box 1010 Wilkes Barre, PA 18703-1010 phone 800 710-8552, 717 822-8899 fax 717 822-8226
