|
PureBytes Links
Trading Reference Links
|
Lesson: Picking markets to include or exclude in a broad diversified trend
following portfolio based on past performance can be as hazardous as market
timing in equities.
One of the characteristics of diversified long term trend following is the
80/20 rule. That is, 80% of your profits comes from just 20% of your trades
(rule of thumb). Note that the 80/20 rule also exists in equities trading.
In a buy and hold position on a stock index, 80% of your profits comes from
20% of the days in the market. The reason market timing is tough is that if
you miss even a few of those 20% days when the market makes big up moves,
your return gets hurt dramatically. Similarly, in the futures markets, if
you are trading long term trend following (analogous to buy and hold in
equities in that both methods are designed to capture the inherent return in
the asset class) and your portfolio happens to exclude the particular market
that makes the monster move, your return suffers badly. I thought you might
be interested in seeing some real numbers.* What really makes the numbers
below amazing is that Cocoa had the absolute worst historical track record
for the 18 years ('80-'97) I tested this system before trading real money in
1998.
It lost money on balance during that period. Gold was not a big winner
during that period either. I was sorely tempted to exclude Cocoa from the
mix when I started trading real money.
* I started trading real money with this system in Sep 98. For positions
that were existing going into that time, hypothetical system results are
reported. In all cases, slippage and commissions are accounted for. These
numbers reflect an initial account equity of $32K. Midam contracts are used
for JY,CD, DM, and LC.
Mkt Dir From Net (as of 7/1/99)
Can$ Long 4/16/99 ($370)
DMark Short 3/4/99 $3351
Corn Short 6/23/97 $909
Sugar Short 1/15/98 $2938
Cocoa Short 2/9/98 $12,188
Cotton Short 1/15/98 $392
OJ Short 1/7/99 ($894)
Cattle Short 12/3/97 $4272
Gold Short 1/1/97 $9389
Copper Short 9/16/97 ($2098)
JYen Long 10/8/98 $0
Crude Long 4/16/99 $0
Just 2 trades (Cocoa, Gold) account for $21,577 out of $30,077 net
trade profits. In this small sample, 17% of the trades accounted for 72% of
the profit. How about that. The 80/20 (approx) rule lives on.
Some notes:
Although the system has been positioned in each market continuously in the
direction indicated, the number of contracts 'on' during the indicated
period has not been constant. There are 2 basic components to the system.
The first generates a directional strength value between -1.0 and +1.0. A a
DS value of -1.0 means short at full strength. A value of -0.4 means short
at 40% of full strength. This DS value is fed into a position sizing
algorithm that calculates the number of contracts to have on depending on
the DS value, current volitility, contract size, and current equity. So
while the DS value may be continuously negative (indicating a short
position) for 2
years, like it has been for Gold, the number of contracts 'on' at any one
time may vary between several and none. However, in my accounting, this is
just one extended short trade in Gold. This is *long* term trend following
after all. Also, this is why there is zero net profit for JY and
CL...althought the system has been positioned long for
some time on each market, the DS value combined with the contract size and
current volatility is such that the position sizing algorithm outputs 0
contracts to have on in the long direction.
The system trades Bonds, but only in the long direction. (No need to have a
switching algorithm with Bonds or Equity indexes since the inherent return
is captured by a constant long position). IOW, the system has been long
bonds forever. Like the other markets, the number of contracts to have on at
any time depends on the contract size, current volitility, and account
equity. FYI, Bond profit is $2843 since 1/1/97.
|