|
PureBytes Links
Trading Reference Links
|
Thanks to all who responded, it is clear to me now that I must read up
on this topic before blindly applying the software. Thanks,
Gabriel
-----Original Message-----
From: DH [mailto:catapult@xxxxxxxxxxxxxxxxxx]
Sent: Thursday, November 20, 2003 3:40 PM
To: Omega List
Subject: Re: Monte Carlo Simulation of DrawDowns
> One system I tested had a maximum actual drawdown of
> 83,000 over a ten year period. But when I look at the probability of a
> drawdown exceeding 83,000 over a one year period it is 63%.
Gabriel, I'm not familiar with your MC simulator but one wonders how
they come up with 1-year probabilities from 10-year data. If they're
simply doing the reverse of your calc, a single trade could radically
change the 1-year results. You're working with very small numbers when
you subract (almost 1) from 1.
99.9999% in 10 years --> 60% in 1 year
99.999% in 10 years --> 50% in 1 year
99.99% in 10 years --> 37% in 1 year
My own simplistic MC program just calculates probabilities over whatever
the test period is and doesn't try to extrapolate that into longer or
shorter timeframes. I suppose it could do that but, personally, I'd
rather just look at the raw 10-year number.
On the other hand, it certainly seems reasonable that there is a near
100% chance that there will be a drawdown sometime in the future that's
as big as the one in the past. That's especially true if your system was
optimized to prevent drawdowns on past data. What's more interesting to
me is the other end of the spectrum -- the 10% or 1% numbers rather than
the 99% numbers. What account size would be needed so you would only
have a 10% or 1% probability of going broke over the next 10 years?
--
Dennis
|