After fuckin around in options for 10 years I am dipping my toe into trading the ES (I started a thread about a week and half ago re my first foray) Question - I am not a statistics major and I would like to find out some basic confidence levels. After 100 trades what confidence level does the results bode for future trades? (I hope you know what I am getting at) Or, what # samples correspond to common confidence levels? 20? 50? 100? 200? 300? 500? 1000? IOW, if I average X per trade in a 100 trade run, how confident should I be in that average continuing over say 2000 trades? Any help greatly appreciated. Still trading options but sometimes not liquid enough.

From a statistical POV, you don't want to be looking at averages that way. Think about it this way; suppose you had 99 -1% trades and one 500% trade. What is your confidence that the average of that run (~4%/trade) would reflect your next 2000 trades? Hopefully, not very much. Start to look at distributions of your trades and learn about standard error and confidence intervals.

The trades will be short term and 99% of the returns will be -10 to + 10 points, so I don't think a 500% outlier will happen.

Here is a more concrete example from last week: 32 consecutive round trip trades total gain was 6.0 points or an average of .1875 points per trade the standard deviation was .31 (the total of the squared gains or losses was 101.34, the square root of which is 10.06. Divide that by sample size of 32 = .31) What I want to get at is if I do a 100 trade sample size, how do I estimate the confidence level of future returns using average gain or loss and standard deviation?

I find a variety of statistical tools at this web page: http://www.stat.tamu.edu/~jhardin/applets/index.html http://www.stat.tamu.edu/~jhardin/applets/signed/intronormal.html For mean = 0.1875, sigma = 0.31, 32 data points and assuming normal distribution then at 80 % confidence level the confidence intervals are 0.117 and 0.258. For mean = 0.1875, sigma = 0.31, 100 data points and assuming normal distribution then at 80 % confidence level the confidence intervals are 0.148 and 0.227. === Is this the type of analysis that you are looking for? Statistics describe the past. Nothing can describe the behavior of nonexistent prices in the nonexistent future. Anything can happen.

Awesome links! I will do 100-200 trades during the next week and plug the results into the confidence intervals applet and see what happens.

Also, you have to keep in mind that markets change. Using confidence intervals assumes the statistical system remains constant. Unfortunately, markets are constantly changing. Right now, we're experiencing a highly volatile bear market, which may make your system do better or worse. That's not saying to ignore your stats now, but don't think a sample right now will help you model later with good accuracy.

You are correct and I am attempting to take the volatility into my algorithm. The trades are quite short term. My positions can be measured in minutes (or less).