Alpha-Beta Trend-Following Revisited: Summary of Method

Stocks & Commodities V. 4:9 (334-335): Alpha-beta trend-following revisited by Anthony W. Warren, Ph.D.
Alpha-beta trend-following revisited

by Anthony W. Warren, Ph.D.
I n the June and December 1985 issues of Technical Analysis of Stocks & Commodities magazine, I
presented a new method for trend-following based on the use of alpha-beta filtering for trade timing and
buy, sell, hold market decisions. This method is similar to the use of moving averages for trade timing,
but has two primary advantages over moving averages:
1) The method utilizes an uncertainty band about current prices to eliminate many of the false alarm
trading signals given by conventional moving averages.
2) The method identifies periods of time to be out of the market on the stock or commodity being
analyzed, i.e., whenever a long-term trading filter moves within the uncertainty band or trend channel.
In this article I re-examine the use of this method as applied to a stock index over a several year period.
This analysis shows that alpha-beta trend-following is an excellent system for identifying buy and sell
points for trade timing without overtrading. Only the major market moves are identified, and the system
avoids non-trending markets characterized by periods of oscillation within a narrow trading range.
Summary of method
The alpha-beta trend-following method is based on plotting four time series:
• the current price time-history which is typically the closing price for daily or weekly sampled data,
• the upper and lower trend channels which define the uncertainty region for trade decisions, and
• the long-term lagged trading filter for market decisions.
Separate alpha-beta filters are used to compute the trend channels and the trading filter. In this regard, the
method is similar to that of using crossing moving averages, where a short memory filter is used to
identify current market action and a long memory filter is used to identify the long-term trend. However,
trade decisions with this method are based on the relationship of the trading filter to the trend channel
lines. The following trading rules apply:
1) If the trading filter lies below the lower trend channel, then an uptrend is in progress and one should
buy or be long.
2) If the trading filter lies between the trend channels, then no trend is evident and one should be out of
the market.
3) If the trading filter lies above the upper trend channel, then a downtrend is in progress and one should
sell or be short.
Figure 1 shows an example of the trend-following method applied to weekly closing prices of the Fidelity
Overseas Fund through October 1986. The trading filter was chosen with a 12-week lag in order to exit
quickly when a downtrend becomes established. (This is equivalent to a 25-week moving average. See
the December 1985 article for details.) Fidelity Overseas topped out in August but the trading filter did
Article Text Copyright (c) Technical Analysis Inc. 1

"Don't You Dare Mess With The Buzzwords, Ms. Ellinghaus. They Lose Something In Translation
To Plain English."
Figure 1
not penetrate the lower trend channel and give a sell signal until October 10. This is the first time that
Fidelity Overseas has registered a sell signal in the two years that the fund has been on the market!
Standard & Poor's case history

In order to get a feeling for the alpha-beta trend-following method, I analyzed the trading signals for the
Standard & Poor's (S&P) 400 stock index over a several year period. This index is similar in performance
to a moderate growth mutual fund. Consequently, a 15-week lag was used for the long-term trading filter.
(With an aggressive mutual fund or stock with high Beta volatility a smaller lag is recommended, i.e.,
12-13 weeks.)
Figure 2 shows the performance of this index from August 1984 to December 1985. The trend-following
method signaled a buy in January 1985 following a strong upward breakout in stock prices. Then between
July and October 1985 the market hit an intermediate peak and declined sharply. This resulted in a sell
signal early in September. Then, about a month later the market began a very strong upward move,
resulting in a buy signal in mid-November.
Unfortunately, this type of whipsaw tends to produce a psychological block reacting appropriately. I
dutifully obeyed the sell signal in September, but failed to reinvest in November, having decided that a
non-trending or bear market was in progress. Nonetheless, I cannot blame the trend-following method for
this stale, since the decision to sell was a prudent response to the existing chart pattern. (Eventually, I did
get back in the market, but only when the force of the bull move was apparent to one and all.)
Figure 3 shows the resulting performance of the S&P index from September 1985 to mid-July 1986. It
should be observed that even with all the backing and filling of the weekly index, the trading filter did
not come close to penetrating the lower trend channel until mid-July, i.e., the trend-following method
signaled an uptrend in progress from November until July. However by mid-July the market had clearly
hit an important peak and the time to sell appeared to be close. Figure 4 shows that the market continued
to deteriorate and a sell signal was given in early August. The market then went on to exhibit a classic
double-top, and even the bulls then knew the game was up. The market has continued to deteriorate to the
present date (mid-October). However, the timely sell signal provided by the trend-following method
allowed me to escape with a nice profit, beating the thundering herd by a month or so.
Postscript
A generic BASIC version of the alpha-beta trend-following method was published in the December 1985
article in S&C. For those who prefer a working program, the magazine will be marketing a technical
analysis code for Apple II computers which contains this routine as an option. For details on features,
availability, etc., contact S&C.
Anthony W. Warren, Ph.D. is an applied mathematics consultant for Boeing Aerospace Company. Dr.
Warren's background and security interests make him particularly well suited to develop mathematically
sound computer applications.
Article Text Copyright (c) Technical Analysis Inc. 2

Figure 2
Figure 3
Figure 4

Alpha-Beta Trend-Following Revisited: Summary of Method

Uploaded by

Copyright:

Available Formats

Alpha-Beta Trend-Following Revisited: Summary of Method

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Alpha-Beta Trend-Following Revisited: Summary of Method

Uploaded by

Copyright:

Available Formats

Stocks & Commodities V. 4:9 (334-335): Alpha-beta trend-following revisited by Anthony W. Warren, Ph.D.

Alpha-beta trend-following revisited

Article Text Copyright (c) Technical Analysis Inc. 1

Standard & Poor's case history

Article Text Copyright (c) Technical Analysis Inc. 2

You might also like