Wednesday, February 26, 2014

Putting Predictability on Your Side in Trading

Abnormal Returns recently highlighted an insightful quote that suggested that market discussions should be evidence-based when it came to the issue of predictability.  

My sojourns through quantitative approaches to markets have taught me that predictability is itself a market variable.  Some market periods and some market time frames are more predictable than others.  Such predictability waxes and wanes over time.  The greatest statistical edge may be on a short time frame at one point; later, it will be over a longer horizon.

The problem with many trading approaches is that they are like blind men in the elephant parable.  They touch one part of the market elephant and assume it's the whole.  In the ideal world, one would be a daytrader when the day timeframe held the greatest edge, a momentum trader when there was evidence of momentum persistence, an investor when there were unique opportunities over long holding periods, etc.  Like the blind men, when we limit ourselves to single trading timeframes, assets, and holding period, we impose our assumptions and distortions and fail to see the whole.

One can study predictability elegantly by constructing rigorous trading systems for particular assets, time frames, lookback periods, etc.  (See Adaptrade for a particularly useful system-building tool).  What you'll find is that sometimes it is impossible to construct a statistically significant model without running hundreds of tests with dozens of variables and thereby overfitting the data.  (Check Marcos Lopez de Prado's work on backtesting to better understand the issue of overfitting).  For other markets or time frames, one can identify robust systems that didn't require undue complexity in construction.

Treating predictability as a variable changes a lot of things:  what you trade, when you trade, how you trade--and how you prepare for trading.  If you think markets offer constant opportunity over all assets and occasions, you'll constantly trade.  If you perceive that opportunity is distributed irregularly and asymmetrically, then the most important part of trading is identifying opportunities when predictability is on your side.

Further Reading:  Every Trader is a Trading System