What I am doing is using simplistic coin toss simulations to investigate the 'real' behaviour of no win, breakeven systems (the null hypothesis) and how some of the evaluation metrics pan out on that data (which has known W/L and expectancy metrics).
In short I am investigating the strengths/weaknesses of some of the metrics and also the alternative to doing all of the evaluation at the backend (equity curve analysis). The way that the rootcause metrics work is independent of money management. For MM I am deferring to optimalF, for the moment, and Vince's work in general. OptimalF shows us the point where we can run,if we want to maximise returns, and also the risk that goes with that - many of us don't want to assume that risk but it is good to know where we are on the sliding scale. I have an Excel sheet, that demonstrates the starting point for root cause analysis, almost ready to roll - it is a lot easier to discuss/explain/learn when we have some visual/tactile aids. It is almost ready to post. Since I am working 'live' I am thinking about posting the draft as it progresses - like the chapter of an online book - people can follow as I correct errors, extend the post, cut and paste paragraphs etc. Having some trouble uploading to the UKB at the moment - as soon as I sort that out I will upload and you will have some copy to bounce off to get your ball rolling. brian_z --- In [email protected], "gerryjoz" <[EMAIL PROTECTED]> wrote: > > Hi Brian, > > a comment on the coin toss equity curve note you penned. I am guessing > if you play with position size in your simulation, you effectively > have contemporaneous curves which you would then sum. Then the curves > won't fan out so much. > More accurately, in some periods you toss a coin and others you don't, > you are either in the market or not on a postion size for separate > streams. Better still, you have n positions with a final expectancy > for each with a empirical distribution of the number of bars over > which you hold the stock, and then you add the result. > > In any case, you finish with an argument for smaller position sizes > and positive expectancy. > Still thinking about your earlier post. > > regards > Gerry >
