[R-SIG-Finance] Burns on Cramer

[Patrick Burns]

BBands wrote:

>On 10/9/07, Patrick Burns <patrick at burns-stat.com> wrote:
>
>  
>
>>I think goals are of secondary rather than primary importance.
>>Reading my horoscope in the newspaper may satisfy my goal
>>of knowing how to run my life, but I think a more important
>>question is whether that is the action of a rational person or not.
>>    
>>
>
>Ah, so we come to a divide. I suspect that in general we'll find
>practitioners on one side and analysts on the other. In any case, I
>got a lot out of a day spent thinking about it. Thanks for your
>papers, always interesting.
>
>As for your poker analogy, since the game most often comes down to a
>few, or less, key hands, I'd guess that overall results have better
>information than how the player performs on an average hand.
>  
>

I'm certainly no expert in poker so you may have it right for
poker, but I would rather entrust my pension to a fund manager
that has shown consistent positive skill rather than one showing
erattic skill.

>By-the-by, we've been using bootstrap distributions to evaluate the
>probability of success for certain trades for a long time. Although we
>are happy with this approach, the issue to me is getting enough
>different kinds of environments in to the bootstrap sample. To wit, a
>sample composed of daily US stock market data from 2002 and 2003,
>yields rather different results than a sample composed of data from
>2005 and 2006. We are currently using trailing two years, but there
>must be a better method of sampling.
>  
>

The market changes over time.  You might note that I emphasize
the time period when speaking of exhibiting skill.  It seems to me
that there is no alternative but to do the testing over several periods.
I would, of course, be keen to hear other solutions.

Pat

>All the best,
>
>    jab
>  
>