Blog Profile: CSS Analytics, Adaptive Systematic Trading
June 2, 2010 Leave a comment
The premier place when if you want to be learning about systematic trading. Its interesting to note that the author David Varadi was formerly a research analyst who used to condemn technical analysis, but made a complete u-turn when he understood the basis of technical analysis.
One of his regular rants include his discussion on creating adaptive indicators. Basically it means that no matter what your indicators you must include them they cannot be static and they must represent the current state of the market. The most important part of such an algorithm should be:
There are a lot of factors to consider when creating a learning algorithm, but one of the most important is that it is “bias free.” What I mean is that the algorithm should not have any preconceived trading style such as a bias toward trend following or mean reversion. It should also consider all possible combinations of a given strategy within reason. To be ”bias free” the ” adaptive time machine” should be just as likely to go long or short given strategy under the right circumstances. In this example we will focus on a short term strategy algorithm to keep things simple, but our testing reveals that it is effective using intermediate and long-term strategies as well.
In another article he using statistical filters to choose which strategies to implement and compares it single strategies such as buy and hold, mean reversion and follow through. The results are astounding as his adaptive algorithms even at 50% confidence interval outperforms any singular strategy by a country mile.
Elsewhere, he also discusses the use of relative strength strategy and its basics. In addition, he also performed analysis on the individual ETFs which he uses in his relative strength strategy to make use of the sector strength. In his analysis he performs tests to check for multi collinearity between the ETFs and finally comes to a conclusion of a set of ETFs that have as little correlation to each other as each other.
Separately, he also using his adaptive techniques on breadth indicators. He makes improvements on traditional breadth indicators as he notes that:
I had a lightbulb moment several weeks ago in which I theorized that certain stocks within the index were more likely to be the “drivers” of index returns than others. Different variables such as volume activity and others may be able to help separate which stocks were useful for predicting index returns versus those that were not.
CAUTION THIS IS A VERY TECHNICAL BLOG