|
PureBytes Links
Trading Reference Links
|
Scott wrote:
> As far as exceeding historical DD, I expect that to happen. But it sucks
> when it happens right at the get go. If I really believe in my system, I
> give it at least 2X historical DD and preferably 3X.
That is the conventional wisdom but, IMHO, DD's can be MUCH worse than
that, at least for the S&P. There has been a lot of discussion about
this stuff on the TradeLab list (highly recommended for those interested
in moving beyond "gambler" indicators).
The first thing you need to do is adjust your historical P/L numbers and
drawdowns for the volatility at the time. Just 2 years ago, the
volatility of the S&P was only 1/3 what it is now. If your maxDD
happened 2 years ago, you should multiply the number by 3 to express the
DD in "today's dollars".
The next step is to feed your volatility adjusted P/L numbers for each
trade into a Monte Carlo simulation. Scramble the trades into a
different order and recalculate the maxDD. Repeat 10,000 times or so.
Plot a histogram of the drawdowns and pick a number that matches your
risk tolerance... ie .1% or 1% or 10% chance of going broke.
Applying the above to an S&P system with a TS MaxDD of 100 big handles,
I decided that, at current volatility levels, 500 handles is about the
right account size for a very risk adverse trader. A trader willing to
accept a 10% risk of going broke could trade it with half that. It's
still a good system. TS just WAY understates the risk.
--
Dennis
|