Prediction

[lazerradial2003]

A few months ago I wrote this:

http://www.moneywhatmoney.co.uk/index.php/Bens-Articles/Why-We-Can-t-Predict.html (please read on this isn't going to be an advertisement for my site!)

It was heavily influence by Fooled by Randomness and The Black Swan, mainly because they were the books I was reading at the time but was broadly based on ideas I'd been toying with for a while.

I was wondering what peoples views are on this sort of analysis of prediction in terms of trading? I find this view beneficial when making trading decisions but several people I've spoken to found it incomprehensible that I found that view of prediction compatible with the idea that money can be consistently made through trading.

I'd be very interested in anyones view on this!

 

[lazerradial2003]

Anyone?

 

[dcraig1]

IMHO there are lots of problems with this stuff. For starters use of randomly generated time series assumes market participants behavior is random (ie independent of past events) and not influenced by previous collective behavior that determined previous price points. That is a very strong assumption to make and really flies in the face of common sense. Just because a randomly generated chart "looks like" the chart of an actual financial instrument means very little. That is very subjective stuff without some objective measure, and I don't mean drawing trend lines on both.

I don't think much of the billiard ball analogy either with it's loose use of terms like "accurately determine postion". It is far too vague. You've got to specify how accurate the prediction needs to be to be useful. I suspect that for the example, the uncertainty would be very large and prediction meaningless - but that matters not a zot when dealing with an entirely different problem of predicting market behavior to some degree of usefullness. Just because some maths can be applied to both does not show that the problems are equivalent or even if the maths used is the most appropriate in both cases.

I think the whole thing is back to front. Applied maths and science often proceeds best from an intuitive insight into the subject under investigation. Your articles get it back to front by starting with a mathematical presuppostion and trying to enforce it on the subject.

 

[lazerradial2003]

Thanks for the response, lots to think about there. Hope the rest of this makes sense, is quite late so might have to do some editing in the morning!

The article with the randomly generated time series was only supposed to be making a comparison between two sets of data which are similar in that they are both the results of a pseudo random process; ie one which I do not know the rules to.

Is the use of pseudo random data as a control what you mean by the mathematical presupposition?

I think the billiard ball problem is a useful illustration of what I was trying to put in more social terms earlier in the article of what happens when lots of small uncertainties are multiplied together and how quickly the inaccuracies become a significant proportion of the final measurement. I suppose the analogy breaks down because in the billiard ball example a useful prediction will tend to be an exact, correct measurement every time whereas in the markets a useful prediction will be one which the cumulative return of acting on is greater than the cumulative loss which doesn't really transfer to the idea of a repeatable scientific experiment.

Hope that all makes sense, very useful feedback!

 

[lazerradial2003]

Any more comments on this? Hope no-one minds me reviving an old thread, I'm working on some new things based around it so looking for all the feedback I can get.

Cheers

 

[rols]

Any more comments on this? Hope no-one minds me reviving an old thread, I'm working on some new things based around it so looking for all the feedback I can get.

Cheers

I assure you the market is NOT random.

It is a highly efficient machine designed to take away your money as quickly as possible.

 

[lazerradial2003]

Don't get me wrong, I agree that the market is not random, I was just wondering on what grounds different people suggest that it isn't and what people define as randomness?