A.I.M. Users Bulletin Board

[Tim Reese]

Proof of concept somewhere? Statistical inputs?

Greetings to Tom and other AIMers - I'm intrigued by the idea that you can automatically harvest some of the gains from a varying input (ie stock prices). AIM seems to me more like a method of reducing risk (volatility) than of increasing returns; thus you should always get a lower mean growth but a smaller sd for AIM managed stocks. Has anyone done some more rigorous testing of the AIM engine? Most of the testing here and on the AIM sites seems very ad hoc ... one should be able to make some formulaeic predictions of performance for various stochastic inputs (ie do more than backtesting with real data).

As I see it, you can either fit the engine to the inputs (ie turn the AIM knobs) or pick inputs that exploit the engine best (ie choose AIMish stocks). It'd be more useful if there were some kind of mathematical formalism that relates the inputs, engine and goals. I'm up for this sort of an investigation, but I don't want to duplicate existing research.

Also, why hasn't retrospective analysis been more popular with the investing audience? It seems like the TA enthusiasts have zoomed right past simple methods of B&H enhancement to short-term and market timing strategies.

Thanks for replying to my newby questions. Three cheers for the R,W&B! Tim