Hi Tom,
Your comment that AIM is a proportional algorithm does not automatically mean that this increases the size of the Buys(or the Sells). It does however provide a convenient mechanism to provide the buy-manipulation that I intended with the exponential buy-rate increase. Consider this:
1 I set my buy aggression lower than NORMAL(Whatever that means!) This reduces the buy intensity(Keeps more cash longer)
2 I set my buy Resistance not very high. I want to take advantage of investing some money in case the price rises again.
3 As the price dips I can delay the buying by raising the Resistance(Price/Behaviour Driven Response). This would increase the Buy Size as the Buy Advice is larger due to proportionality.
4 As the dip continues I can increase the buy-aggression and this will increase the buy rate still more.
This essentially achieves that for deep dips in the first instance the cash is preserved early in the Dip Mode and that more cash is invested in s delayed fashion...exactly as I intended.
But Mark can do the same thing with his AI by using the Beta Multiplier with a value less than 1 and then raising the Resistance and increasing the Beta as the stock dips. So, with these possibilities, I have to agree with Mark that the performance of the two AIMs will entirely depend on the choices for setting the parameters, and if we well set the parameters so that both AIMs perform equally well, or that one performs better than the other.
I think I have to realise that in this sense in a Contest the best investor will win.
In a test on two particular systems it would be necessary to find out if for real stocks it is possible to derive at a set of parameters that will consistently outperform the other system.
Conrad
AI
