June 26th, 2006, 2:42 pm
I don't do this, so I feel free to speculate. 1. I believe there is a small amount of forecastable direction. Whetherit's something you could make any money on is another matter. 2. In active equities, there are 100,000 - 200,000 ticksper day, so obviously it's going to be your computer vs. another guys.The first step would be to try to write some rules that predict themove off the order book imbalances. To do that, you need somereally good, accurate, timely data. 3. There are a zillion smart people who have already tried this. 99% of themfailed to create anything significant. The remaining 1% havespent big bucks and are now in the nth iteration of their trading system.Their system keys off a whole bunch of explanatory variables besidesthe order book. This is your competition. To make any money, your systemhas to be better than the best of these. regards,