Replies to post #7099 on A.I.M. Users Bulletin Board (AIMUSERS)

[Qarel]

Hi LC (and Hi aptus)

I am just pondering this, and more for my amusement than from any real wish to change the AIM algorithm. I already did the same test and found how raising PC after a sell instead of after a buy had a severe impact when you consider the Lichello cycle. But for obvious reasons this isn't very informative: the Lichello cycle is almost custom built to make the trading side of AIM shine. (Real trading systems like Vortex or X_Dev do even better.) But for my Nasdaq100 sheet this variation actually was beneficial, so things remain undecided. Perhaps someone with a huge database and a good backtest engine could do some number crunching. (Yes, that's you, aptus. I hope you are interested!)

Regards,

Karel

[Qarel]

$1,000,000? Make that $16,312!

The effectiveness of AIM in the Lichello cycle is partly due to the cycle itself. I rewrote the cycle so that it would go from 10 through 4 to 10 in 180 data points, with the help of a cosine. That makes two cycles in 360 points, or 15 cycles, as in Lichello's example, in 2700 data points. (Thank you, Dan Bricklin, for spreadsheets!) So what does AIM do with those 15 cycles? It dies in the 10th cycle with a last buy at $7.73. After that, no more action, and the final score is $16,312.

BTW, the (tentative) add-to-PC-after-a-sale version of AIM dies in the 15th cycle after a buy at $7.83. The hike back to $10 doesn't trigger a sale. The final score here is $21,813.

Of course this doesn't disqualify AIM. But in my opinion it does disqualify the Lichello Cycle, in case that was still necessary.

Regards,

Karel