|
PureBytes Links
Trading Reference Links
|
Gary:
I think we are essentially saying the same thing, but here are a few
comments.
> If there is no dependency, then technical analysis is just wishful
> thinking. If the basis of TA is correct - that patterns do exist in the
> market, then past actions do affect future outcomes. If there is no
> dependency, then you can not devise a successful system for entering
> and exiting the market.
I was referring to a dependency between wins and losses (not the dependency
we try to find between price and an indiactor). We have probably all
designed a system that trades in the wrong direction, for example, when a
strong trend starts. If these strings of losses are consistent, you may
have a high assurance that if you have one loss in these circumstances, it
will be followed by several more losses. If you have this situation, then
drawdown is a reflection of this statistical fact. If no there is no
dependency, then drawdown is more likely a coincident string of
random losses.
> To use the largest losing trade instead of the largest drawdown would put
> you at risk of ruin with one large loss followed by one other loss. Of
> course, using the smaller of the two assumes the other won't happen. I
> think adding the two to determine the minimum capital requirement would
> be more realistic ....
Using the largest losing trade to represent the maximum possible series of
losses would indeed be devastating, because you will suffer a string of at
least average losing trades, and quite possibly a string of largest losing
trades. It is your money management system that tells you how much to risk
on each trade, in light of the probable number of losses. My point was that
using drawdown to determine risk or trade size is just a weak approximation
of potential losses, and makes for a very risky system. Since one cannot
nail down a definite number of consecutive losses he will suffer as implied
by his drawdown calculation (or, much less, control this), he should look
instead to the largest losing trade (which can, to an extent, be
controlled). Combine this with a statistical distribution to anticipate the
losses (this is what a good money management system does or is broadly based
on) and you have a sound system for determining risk and trade size.
Regards.
----- Original Message -----
From: <Randall_Gary@xxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: August 31, 1999 07:24
Subject: Re:Is drawdown meaningless?
> >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.
>
> If there is no dependency, then technical analysis is just wishful
> thinking. If the basis of TA is correct - that patterns do exist in the
> market, then past actions do affect future outcomes. If there is no
> dependency, then you can not devise a successful system for entering
> and exiting the market. You might, however, develop a system that
> manages trades by limiting losses and letting profits run. However, for
> this to be successful, such a system DEPENDS on large runups to
> cancel the limited rundowns. If the markets were truly random, resulting
> in a 50/50 outcome, every trading system would eventually fail. The
> only way to win would be to quit at the right time.
>
> No gambling casino, insurance company, or trader can survive without a
> statistical edge. They also require enough capital to keep from going
> bust. Whether that happens with one trade or thirty is irrelevant; knowing
> the total capital requirement is the key to success. Since we can't
> accurately predict every event, we must accept the fact that our analysis
> is totally dependent on statistical theories. We can take heart from the
> knowledge that there are consistently successful traders who employ a
> variety of systems.
>
> To use the largest losing trade instead of the largest drawdown would put
> you at risk of ruin with one large loss followed by one other loss. Of
> course, using the smaller of the two assumes the other won't happen. I
> think adding the two to determine the minimum capital requirement would
> be more realistic, and many traders suggest multiplying the drawdown by
> two, three, or more just to cover life's vagaries. In any event, having
> enough capital to survive whatever can happen is the key to allowing a
> successful system to work. Gamble with that fact of life and you've
> thrown away any statistical edge.
>
> Gary Randall -- Brunswick, Maine
|