|
PureBytes Links
Trading Reference Links
|
Al:
I certainly wouldn't dismiss the hypothetical system out of hand. If the
drawdown was caused by one or two large losses, yes, I would dismiss the
system (unless, as Lionel mentioned, it could be improved with stops). On
the other hand, if the drawdown was caused by a string of many small losses,
I would:
1. check to see if there was a mathematical dependency between wins
and losses. If so, it could be exploited by reducing trade size when the
likelihood of a string of losses arose; or
2. with a 60% win rate *while* experiencing strings of many small losses,
I would be encouraged to work my behind off to identify the cause, and
incorporate a filter to eliminate the bad trades. Eliminating these trades
would make it a killer system with a very high win rate.
Also, recognize that if there is no mathematical dependency, there can be no
assurance that the $19,000 drawdown will ever repeat or that it represented
the greatest possible drawdown. One simply cannot rely on the $19,000
figure.
Regards.
----- Original Message -----
From: Al Taglavore <altag@xxxxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: August 30, 1999 16:35
Subject: Re: Is drawdown meaningless?
> Glen, if you had a methodolgy that resulted in a 60 % win factor that
> produced $12,000 worth of net profits but had a draw down of $19,000 and
> you had a trading account of $20,000, would you want to trade this
> methodolgy and risk $19,000 (or most of your account) to make $12000?
>
> I interpret draw down as the 'risk' measure in "risk and reward". Also,
> if your system tests in test data showed a 60% win factor, that is not
> indicative that the same percentage will apply to out of test data or the
> real world.
>
> I do think you have started a very interesting thread, and I look forward
> to reading other opinions of this question.
>
> Al Taglavore
> ----------
> > From: Glen Wallace <gcwallace@xxxxxxxx>
> > To: MetaStock listserver <metastock@xxxxxxxxxxxxx>
> > Subject: Is drawdown meaningless?
> > Date: Monday, August 30, 1999 5:14 PM
> >
> > I first came across this idea several months ago, but at first I did not
> > fully understand. The fog is now lifting, and I want to toss this out,
> > first, to get other people's thoughts and reactions, and second, to
> > collect my own thoughts.
> >
> > Simply stated, drawdown is the largest equity dip caused by a sequence
> > of losing trades, and many use it to assess whether a system is
> > tradeable.
> >
> > Assuming there is no mathematical dependency between wins and losses
> > (ie. a win does not beget a win, or a win does not beget a loss) in a
> > system, the fact that your system test just happened to find, say, five
> > consecutive losses is irrelevant. Consider a series of coin tosses. Even
> > though there is a 50% probability of heads, a sequence of five
> > consecutive tails is fairly common, but shouldn't influence a person to
> > not bet on heads. In fact, a sequence of 10 consecutive tails is quite
> > possible, yet the probability of heads on the next toss is still 50%. An
> > analysis of this 10-sequence drawdown might mislead an investor into
> > believing the coin-toss trading system is too risky. In contrast, let's
> > say that the series of coin tosses yielded no consecutive losses,
> > thereby giving the investor the false impression of a small risk. In
> > either case, the calculated drawdown is meaningless.
> >
> > If there is no dependency between wins and losses, can we expect past
> > drawdowns to repeat? Can we expect future drawdowns to be bound by past
> > extremes? No; there is no limit to the possible number of consecutive
> > losing trades. Drawdown is not meaningful, other than to give some false
> > peace of mind.
> >
> > Since there is no limit to the number of possible consecutive losses and
> > it is uncontrollable, the better yardstick is the size of the largest
> > losing trade. To an extent, this can be controlled. And with proper
> > money management techniques, you can structure your trade sizes so
> > as to survive the inevitable string of losses.
|