Tuesday, February 5, 2013

Mean Reversion Trading with the Adaptive Cycle Toolkit (ACT)



By: Michael Burgess, creator of ACT

In order to succeed at trading, use a simple approach with a statistical edge. It should be simple, or the trader becomes confused in the heat of trading. It should have a statistical edge, or the trader is simply gambling.

One powerful trading technique uses statistical principles of reversion to the mean. This is a sophisticated way of saying that a price that moves too far from its average tends to return to its average. Variations of this technique have been widely written about by such noted traders and authors as John Ehlers and Larry Connors (ACT is not associated with or endorsed by John Ehlers, Larry Connors, or any of their entities). This reversion to the mean is often the most powerful and probable in the direction of the dominant trend. Buy valleys in an uptrend, sell peaks in a downtrend.

Sounds simple, right? Ok, so how do you dynamically measure trends while avoiding market whipsaws, and how do you identify valleys and peaks quickly enough to capitalize on them?

The powerful methods included in ACT allow us to answer these questions. First, let’s explore how ACT can help, and then we will develop a specific mean reversion technique. The approach is simple, and consists of only three plots: a price chart, and two indicator plots.

ACT achieves these results because its methods are dynamic. As traders, we know that markets change -- they trend, cycle, and drift. ACT applies advanced signal processing concepts to rapidly identify this change, and allows a trader to adapt to the market. ACT is based on the research of renowned quantitative analyst, John Ehlers, as described in Rocket Science for Traders and Cybernetic Analysis for Stocks and Futures.

Applying methods available in ACT, a trader may:
  • Identify trends quickly
  • Know when the market is cycling
  • Know when the market is noisy and perhaps stand aside
  • Pinpoint extreme prices with the Fisher transform
A Simple, Adaptive Mean Reversion Approach

The approach consists of the following steps:
  1. A plot of the security for the time interval to be traded (e.g. intraday, daily, etc).
  2. An inner window with a plot of the ACT MAMA/FAMA nonlinear averages. MAMA is the faster average; FAMA is the slower. These averages rapidly adapt to abrupt moves, but often remain apart during market drift, thus reducing whipsaws.
    The custom algorithm for adaptively calculating these averages is based on mathematics described in the Cybernetics book. To our knowledge, this formulation of MAMA/FAMA is only found in ACT, and not in any other discussions on the web or in print. Earlier methods rely on less responsive techniques to recover cycle measurements. These averages are discussed in more detail in the ACT manuals available with the add-on.
  3. A second inner window with special 3 bar statistically normalized Laguerre Stochastic. The Laguerre Stochastic applies an unusual mathematical approach to very short data lengths, and is discussed in the Cybernetics book and the ACT manuals. The result is a very smooth, responsive oscillator.
    Concepts related to the Fisher Transform are then applied to normalize the oscillator within +/-3 standard deviations. The Fisher Transform is discussed in the Cybernetics book and the ACT manuals.
  4. The trading approach is simple. If MAMA is above FAMA, look for oscillator values below -2.5 standard deviations as potential entry points to “buy” If MAMA is below FAMA, look for oscillator values above +2.5 standard deviations as potential entry points to “sell short.”
  5. Maintaining a trade size that keeps each trade’s exposure to about 1% or less of account size should the trade hit the trader’s “stop loss” point is often advisable.
  6. That’s it. The images below show the plot of IBM from 2000 to 2007 with the indicator plots so you can visualize the process. The approach is intuitive, and should be almost immediately apparent.
  7. MetaStock code for this approach using the ACT tools will be included with each ACT purchase.
Conclusion

Thank you for spending some time with us today.

ACT is based on the realities of market dynamics, and the mathematics behind it are solid. You can apply its features to measure trends, cycles, market noise, and adapt to market conditions. The features and practical applications are discussed in comprehensive, yet understandable manuals.

You can apply its features to create or support your own approach. If you want a quick start, take a look at the mean reversion approach described above. We think you will like it.

It is now available to our domestic and international MetaStock users in a convenient, downloadable format!


Chart 1 - IBM March 2002 - November 2003
Chart 2 - IBM August 2005 - March 2007
Post a Comment