Nazz, the big losses appear to be a case where the model didn't fit the situation, e.g. there was a big gap up or gap down followed by a quick reversal not immediately caught by the model. In actual trading, I would be able to catch those, but I'm doing the best I can to lay down rules that don't require a lot of discretion. When I have to make judgment calls is where I seem to get into trouble. I'll take a look at the impact of time in position though. Thanks for your comments.
The stats look pretty decent, although it should be noted that the last 3 years have marked a bull market. Did you backtest the 2000-2003 period?
ROA ( net profit / max intraday DD ) is really impressive. Profit factor is not incredible but OK. It seems like this system is pretty good as you have performed out of sample tests. Perhaps, you just have to control if it is not subject of the usual pitfalls : limit order queues and execution, liquidity...
No, but the profit was evenly distributed between long and short positions, so I would think it would do as well in a bear as a bull. Does that make sense?
I guess the profit factor is the thing that concerns me the most. I don't like to see lots of losing days either. I'd be a lot happier if the profit factor was more like 2, but is that a realistic expectation for index futures? I guess that's the heart of my issue: what should I expect for a futures trading system? I don't have anything to compare to other than other systems that performed worse.