A.I.M. Users Bulletin Board

[Conrad]

Clive, I propose we appoint you Professor of the AIM University!

When I read your comments I have a faint feeling I understand some of it but then I realize that I was mistaken . Other than saying rapidly: "Yea, it sounds good" it goes past me at great speed.

I will make an attempt to respond to the parts I do get and relate it to the test we did:

The results you achieve even when comparing two exact same machines (methods) with the exact same settings/parameters will potentially significantly differ given even just a minimal change in starting price value. The same applies for even just small differences between parameters/settings.

When you mention "machines" here they are theoretical systems and should respond identically. . . in real life production machines are never identical as they operate with in tolerance ranges and in that way they their functioning can diverge even with identical input. With different inputs the difference would be expected to grow. With small differences in the parameter settings the development of large differences would be expected (the machines are no longer identical). Also when two people run an identical “machine” they would likely execute trades differently and if not doing so intentionally different brokers would execute trade orders differently and charge different fees. It would be a miracle if two "machines" get the same result in finite time. With different "machines" one could make money while the other losses the bundle!

If I understand correctly Vortex is similar to AIM but uses a slightly different method of Portfolio Control updates after each trade and has separate buy and sell aggression factors, which scales up or down trades sizes according to recent events (predictions). Which is somewhat similar to how LD-AIM scales trade sizes but not in such a dynamic manner. Similarly using a sell aggression value that produces a zero trade size is akin to pulling a Vealie.

Actually The PC update is very different. I have modified the AIM PC update for buying specifically to eliminate the Residual Buy Advice. . . used to call that "The Lichello Flaw" so that the Buy Advice = 0 after each Buy. This was achieved by replacing the 0,5* Buy in the AIM update function with Fb*Buy and setting the Buy function as

Buy = M*(PC-SV)

and solving for the value of Fb to make the Buy Advise = 0 after the trade was executed and PC was updated. The solution resulted in

M=1/(1-Fb)

and allowed in fact the Buy Aggression Factor Fb to be any number < 1. . .(using Fb > 1 makes a Sell from a Buy so is not useful)

and thereby the Aggression Factor Fb was born.

I used the same function for selling with Fs as Sell Aggression factor.

The default setting is Fb=Fs=0

Another principal difference is that after a Sell the Vortex PC is updated also to give Trade Advice =0 after the trade is executed.

In this way after any trade executed as advised the PC(new)=SV(new)

In regards to setting the Sell agression so low that the Sell Advice = 0, you suggest that this is like a Vealie. In Vortex I call this "Let the Profit Run". Is dat exactly as the way a Vealie works?

With the default settings Fb=Fs=0 Vortex trades more aggressively than AIM on the initial selling and buying as well as on the buying because the AIM SAFE*SV "trade damping factor" makes the AIM trades smaller at an identical start condition.

However the different PC-updates has demonstrated to make AIM and Vortex very, very different "machines" that react to price changes in a very different way, unless the overall response for the two systems are artificially tuned to give the same output at the end of the run, but these settings would not work for a different price structure.

For this reason any further testing is in my opinion not useful as there is no logical reason for making AIM and Vortex respond in the same way.

I shall complete the analysis of the test on the basis of the AIM-Vortex Trade Charts I made later today(on May 15). I got to hit the hay now, it being 03:15

Zzzzzz.