Re: [OPE] Digital investor worlds

From: Ian Wright <>
Date: Tue Jan 18 2011 - 12:34:49 EST

Hi Paul

> Lay mans question: With respect to flash crashes etc ... what I don't
> understand is IF 'chaos' theory mathematicians (who believe mathematics is
> inadequate, not that chaos exists) tell us that  quite unpredictable
> behaviour eventually occurs when 'predicting' a next set of prices and then
> compounding that by reiteration, why can't we expect the algorithms
> eventually to express 'chaotic' behaviour as they race ahead 'predicting'
> the future (actually as a result of an inadequate number of decimal points
> for all data in their systems) in the first place? Either algorithm sales
> people are hiding their mathematical inadequacies or everyone is keeping
> quiet about the problem to keep playing a game that must always break down
> if it 'over calculates'.

I do not know much about algorithmic trading. So the following is speculation.

Speculation is a Keynesian beauty contest. So the algorithms are not
only in an arms race with each other, but need to recursively try to
predict what the other algorithm is doing. In these circumstances any
medium to long-term prediction is impossible. I expect the successful
algorithms identify arbitrage opportunities on very short time scales.
Their predictive power will likely only manifest as small percentages
better than chance.

ope mailing list
Received on Tue Jan 18 12:37:11 2011

This archive was generated by hypermail 2.1.8 : Mon Jan 31 2011 - 00:00:02 EST