Just read an interesting article in Markets Media about machine learning in trading. Our chief systems designer, Murray Ruggiero, wrote a book about this a while back regarding using neural networks---computer programs that learn and make predictions-- to trade. To me, the most interestsing use of this technology is when to use trend following strategies and when to use mean reversion strategies. Trend following strategies work best in trending markets while mean reversion strategies work best in choppy markets. It is easy to figure out what type of market we have been in, but to be able to accurately forecast what type of market is likely in the future would allow the trader to find an optimal allocation between trend and mean reversion strategies.