[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: is profit factor the best measure



PureBytes Links

Trading Reference Links

Very interesting.
What is W^

Thanks
Peter


-----Original Message-----
From:	Kevin Morgan [SMTP:kmorgan@xxxxxxxxxx]
Sent:	Sunday, August 23, 1998 4:16 AM
To:	RealTraders Discussion Group
Subject:	Re: is profit factor the best measure

A measure that I really like is Ralph Vince's "pessimistic return
ratio", PRR.  To quote from "Portfolio Management Formula":

  "The PRR is essentially the profit factor measure, only adjusted to
   give greater weight to more reliable passes (passes with more
   trades).  The PRR is the profit factor with one fewer square root
   winner and one more square root loser.  Mathmatically:

PRR = (((W-W^(1/2)))/T)*AW) / (((L+(L^(1/2)))/T)*AL)

where
 
   W = # of winning trades
   L = # of losing trades
   T = total number of trades
   AW = average winning trade amount
   AL = average losing trade amount

  "In over words, the PRR is the profit factor weighted down inversely
   by the number of trades.  PRR values greater than 2.00 are
   indicative of a very good system.  Over 2.50 is excellent.
   The PRR measure is perhaps the single best measure of system
   performance..."

Warning (perhaps obvious): a curve fitted system that won't perform
in real-time can still have an excellent PRR!!

Since there's been a few comments about the importance of maximum
drawdown.  Just to inflame things a little: in this same book Vince
has an entire section titled "Drawdown is meaningless, biggest losing
trade isn't".  His basic reasoning is that assumption there is
a maximum drawdown is an illusion.  This is because sequences of
trades are really independent trials, and the sequence in a given
test run that results in a drawdown is arbitrary.  There is no limit
to how bad things could get, it just becomes less and less likely,
based on the probabilities of the system.  What is (somewhat) 
controllable is largest loss, and system frequencies of wins/losses
and relative sizes.  Fixed fractional trading is all wrapped around
largest loss.

On equity curve, I agree that a smooth equity curve is desirable, but
only if we are talking about fixed contract size trading.  Making alot
of money requires compounding of trade sizes, and with that comes
increasing increasing equity curve variation.  It's a requirement to
get "big spikes".

-k