Now if you create a average system with the same winner mean of 150 and mean loser of 100, it will have a std. dev. of 100% of winners and losers. You can see from this post it's costing you a couple of % in consistency. Check the % profitable periods versus the table.
And if you're trading like a pro (cutting losses short and letting winners run), then the same mean win and loss gets pretty lopsided. Notice how much the expected outcome and 50% profit factor change just by using discipline. If you look this up in the table for the 2.04 profit factor at the 50% level you see it provides about a 2% improvement in consistency over the normal distribution. Hope this provides a little more detail on the use and limitations of the table.
12 month contract. I hit my highwater mark to get paid early this year. Since about May I've just been hanging out and spending maybe 1 day a month at work to review the automated strategies. I stepped on some toes also, so it's better for me to do my own thing. I can't believe the DAX is still so easy to trade. I'm thinking about raising some money to maximize my strategy in the DAX market. If they're going to leave so much on the table I'm a pig enough to go after it.
I'd like to ask you a question on a slightly tangential topic, the testing of identifyers. Say you are building a system (ie buy open + .2*AvTrueRange) and are compiling patterns to filter trades. You test three identifitiers: (these are just examples) open[1] > close[1] high[1] < high[2] low[2] > low[3] None of them show any significance. Then you test every pair of two identifiers jointly and none of them show any significance either. Then you test all three of them together and find that they only generate 1-2 trades per year, yet all the trades are statistically significant and highly profitable (thus passing the edge test). Would you consider this to be a tradeable pattern, or would you avoid such patterns that work as a combination of many identifiers that fail when tested separately? -bulat
Acary>. Excellent stuff especially for a systematic trader like me. I especially like your method of doing correlation analysis of models as well as products to determine weighting of capital. Personally I do look at correlation among securities but I never used the correlations to weight the systems or products to achieve a smoother curve. Great work. Best of luck in future research and implementation.
alan - a few months ago you mentioned some trading plan development that you were excited about ... are your recent posts part of that evolving plan? either way, thx for your posts and welcome back to the neighborhood take care - omni
I have a question. Say you trade the following basket with the appropriate correlation factor. How can the correlation of the entire basket of currencies be judged? I am not sure how to go about doing this. Thanks.
The problem is that with "1-2 trades per year", you would not be able to judge the profitable trade sample to be statistically significant in any way, unless your test covers at least 3-4 decades.
Maybe it could be possible to devise a system that picks up future winners after positive earnings announcements, so that it would be possible to find different trade scenarios every month. The same could be said for finding losers after negative earnings in order to short and therefore balance your portfolio. Just an idea....