It is so refreshing to know that none of you knows shit about reversion to the mean and how to calculate that mean properly. When I saw the thread title I was concerned that some of you might get it, but it is such a relief to know that you are all still in the dark ages of "Bollinger bands" and similar crap. It is, however, interesting how the oldest method in the book is still not known to the majority of traders. Mr. Galton would definitely have a laugh! Anyway, this thread makes me feel so much better!
I worked on mean regression stuff. The best I could come up with was a variation where I bet bigger the greater the deviation from the mean... it was a martingale idea that made money fast and had a high risk of ruin... [averaging down is said to have killed more Jews than Hitler ]. I thought that I would run it some day with the idea that I knew it was going to blow up occasionally and I would have the money to fund it again every time... what you are really risking is that you will get a run of blowups and be unable to keep on playing. It sort of opens up a trading and mathematical realm that I don't want to enter really... now if it was OPM and I was on commish... heck yeah, call it a Hedge fund...
The "ruin" could be avoided. What function did you run as your mean? Please do not tell me it was a moving average of some sort.
Maestro, you are being "mean". In stats 101 the mean is the simple average. I have to admit, I looked at the charts a thousand times and thought about how it seemed ridiculous to watch the mean itself moving and price reverting to the mean at a point where the trade was a loss... I moved on, no indicators at all nowadays, none, nada, zip... but if there were a different way to know the mean I might venture back into that area, it was more fun really...
I wasn't trying to be "mean" Mean or expected value used in the distribution formulas could be calculated using all sorts of interpolators. Moving average has a very bad reputation of being "late" and not representing the "mean" as such that the distribution of the price over it would be stable. There are higher degree polynoms that accomplish this task much better.
in for MAESTRO to tell us the correct way to calculate the mean using higher degree polynoms. forex forums are full of RTM/average down strategies/EAs/whatever. most of them suck. a few are hedged. a couple have lasted for years until they finally blew out on the day price moved [your max drawdown] + 1 pip. some hypothetical examples were run where when that happens you survive cuz you've increased your account by enough in the meantime and made regular withdrawls. but they all exist on forex where you can start with microlots and have 1 pip = $0.01 to allow averaging down. then you've got the grid system guys but thats not really RTM. anyway, RTM can't work without either 1) perfect entries or 2) averaging down. since i havent mastered #1 and i dont feel safe doing #2, im all in for non-ambiguous discussion here.