|
PureBytes Links
Trading Reference Links
|
Joseph Schedlbauer asked:
>Would it not be true that the MC simulation, due to the high number of trade
>sequences it generates, would come up with many sequences of trades which
>contain unrealistically long sequences of losers, and therefore overstate
>the system's max drawdown?
The number of losers is always the same. The number of losers in a
row can be different, and can occur at different times. The point
here is that successive trades are not necessarily dependent on
the previous trade. If you have a system that generates a certain
percentage of losers, there is a probability that over a period of
time, you will get a lot more losers in a row than the period under
which you originally tested. The MC simulation attempts to show you
what the effects of that would be.
For example, the digits in the number pi are deterministic (you
can generate them by a formula), but evenly randomly distributed.
However, you find patterns in any random sequence. If you look long
enough you will find a sequence of, say, 1,000 digits in a row less
than or equal to 4. Likewise if you trade a strategy long enough
you will find an unusually long clustering of losers (or winners).
>Wouldn't it be better to look at the system's losing percentage and
>average losing trade and use statistics to estimate the probability
>of a particular level of drawdown?
MC simulations *do* generate statistics. That's the whole point.
If you want to find the position sizing strategy that maximizes
profits and minimizes drawdowns, you can use just the one sequence
of trades in original order, but then you'll have no clue about what
ELSE the system is capable of throwing at you given the same winning
percentage, win size, loss size, and risk.
-Alex
|