Biased Algorithmic Trading – How to Trade UP & Down Markets
Many years ago I figured out that algorithmic trading for the most part subscribes to the age of paradigm that “nothing works all the time but some things work some of the time” This is very true fro trading approaches in most markets.
One concept that I do like to employ is the concept of biased system trading. What is biased systems Trading? Biased system trading means that a trading approach (in my case algorithms) can be tilted to towards favoring either up trends or down trends. The most obvious example of this is the mutual fund industry that buys long only. They do well when the equity markets rise and poorly when they fall. For trading the same concepts exists but on a smaller scale. Why Bias at all? The answer is behavior. Human beings are generally predictable and so are their behaviors. How a market behaves in an uptrend is different than how a market behaves in a down trend. By designing a trading approach for different markets, traders can achieve more success AND !! if a biased algo is starting to reverse its profit trend (we call this Algo trend) then its a clue as to a new trend developing.
I run 2 types of Algos, Trend and Chop market with signals designed to profit when the market is trending or chopping. I also like to bias these models market towards one trend or another. The benefit of of this approach is that the markets tend to move in waves, both up and down and as such there will always be something working. Below are to graphics showing two US 30 year algos, a bull biased algo on top and a bear on the bottom.
In the top, we see how a bear system performed well in a downward environment while the bottom shows how the bull performed when prices rose. Click for full screen images.
Author

Dave Poxon
The Index Strategist
During the past 25 years, I have worked at some of the largest investment firms in Canada including BMO Nesbitt Burns, Canaccord Genuity, Refco Futures Canada, performing quantitative research and trading.

















