A.I.M. Users Bulletin Board

[Conrad]

Mark,

Thanks for detailing what you mean exactly with back testing. Although I am not familliar with the specific method you desribe, I get the drift of it. I would suspect that this type of testing would also only be meaningfull is the Model you build is used for a stock that has approximatelysimilar behaviour as the Model. With this I mean that if the Model works well with 25% amplitude cycles then I would expect it not to do well with 7% amplitude cycles(other things being equal).

My thinking is that with runing a number of test with stocks that vary quite differently(not only with amplitude but also
with other differences) and optimizig with overfitting then I will arrive at global parameter settings that would be "best" for stocks that behave more orless the same way. More complex testing is for me not possible at this time as I have no detailed know-how on that type of testing.

If you are interested to do the type of back testing you speak of with my AIM then you are welcome to do so. I could then give you the combination of pc-correctors I have setteld on for now. Thanks to an Vortex-AIM Excel spreadsheet that was provided to me by Lou Dina I was able to do some fast testing with some variations of my basic system.

Regards,
Conrad