I'd like to get some opinions on what you might consider to be a rough approximation for attaining ergodicity given the model and testing I'm about to describe. FWIW, my impression is that traders use the phrase robustness to mean something like ergodicity. Basic info on the model: I just launched fully-automated trading of the model in mid-August. The model has an average holding period of just under 1 day. The model trades about 5-6x per month per instrument on average. I use it to trade a variety of futures contracts and it has tested comparably on spot FX and individual equities. Testing: I've chosen to do only the simplest of brute-force optimizations (besides, there are only 11 parameters that actually vary)Instead, I've gone down the path of trying to make the model as generalizable as possible. That is, my goal is to test the model on an ever increasing number of instruments over 1 year's worth of data in order to get some reasonable ensemble stats. Results: So far, I've tested (and in most cases trading real-time) 8 instruments over a period of one year. The # of trades in the sample is +/- 400 trades. A few of the ensemble stats beginning to emerge are as follows: Sharpe ratio: 1.75-2.25x % Wins: 53%-60% Avg Win/Loss: 2.0x Question: How many instruments, how many trades, and how much time would you test to get performance stats that you thought were representative of the ensemble of instruments? On what basis did you reach your conclusion? Thanks much in advance for your help, Lou