Tuesday, January 01, 2008

Trading Consistency

A very interesting blogpost about trading consistency. The bottom line is that you need to take as many trades as possible within your system guidelines to be consistently profitable as long as your expected profit per trade (expectancy) is positive. Which makes perfect sense statistically. Even my simulated trading of the Australian futures has a "profit factor" (total dollars of winning trades/total dollars of losing trades) of only 1.84 currently. I can find an average of 1-3 trades per day probably. But still around 95% of months would be expected to be profitable based on the somewhat simplified analysis in the article. But my real money NASDAQ/SPX trading is now only at a "profit factor" of 1.11. Not much more than half of months would be expected to be profitable. Which is what I saw this year roughly (7/12).

By contrast, Dinosaur Trader has a profit factor based on his daily results of 3.01. But still he had two losing months out of nine (78% consistency). That is a very negative outlier. According to the article, with 20 trading days per month consistency should be in the 97-98% range. I notice that the win percent in their simulations is very high compared to DT's or my own.

BTW DT's win percent is similar to mine: 64% but his win/loss ratio (how much his typical win makes relative to his typical loss) is 1.70. His trade by trade data might be quite different. On average he made $604 per day with a standard deviation of $2708. The t-statistic to test whether this mean is greater than zero is 2.81 which is highly statistically significant. In other words, it's not luck.

1 comment:

Richard Todd said...

If you do the analysis with a variety of win rates, the difference is negligible after a few trades. I think it's more likely the tiny sample (9 months) that makes dino's results to date such an outlier. This should make him even more confident that he will bounce back in the coming months. A profit factor in the 3s is quite good.