|
PureBytes Links
Trading Reference Links
|
>
>Phil, could you expand on this just a bit? How can the single contract
>simulation be misleading and I assume you're implying that multiple systems
>under these conditions would be more accurate due to a sum of the errors vs
>error of the sums effect? I care, I care! :)
>
>Kent
this is something totally different!
Assume you want to use about the same leverage - or at least a known amount
of leverage - on each trade. For example, say you want to risk a relatively
constant portion of your account on each trade. Or experience relatively
constant equity swings for a given account size. Or just assume you want
the drawdown numbers in the systemn test to actually relate to something.
Wouldn't that be nice.
It's not an unreasonable goal. But a single-contract simulation won't
necessarily get you there. The problem is illustrated beautifully in a
long-term market like the SP - here's a painful example:
Say your system just says something like
if whatever then buy close;
if barssinceentry>=7 then exitlong;
Say on 8/23/82 you had a signal and you bought 1 contract - the price was
118. The average daily truerange was about 2 points. Seven days later you
sold it at 124. How much money did you make? According to TS you made 6
points * $250 per point = $1500. Nice trade!
Now say it's 8/26/98 and your system tells you to buy one contract. But now
the average daily true range is about 20 points. You buy at 1087 but this
time the stoploss gets hit, at 976. Now you've lost 111 points * $250/point
= $27,275.
Too bad you have to go out and get a day job - all because you didn't
normalize your system to be independent of price and volatility.
Sorry for the long-winded explanation. The problem can be dealt with by
something like the following:
if whatever then buy 3000/(Xaverage(TrueRange,50)*Bigpointvalue) contracts
close;
Here your equity will swing exactly $3000 per truerange. Regardless of
whether the truerange is 2 points or 20. Btw credit goes to Dennis H. for
this solution - I didn't think of it but am happy to use it!
|