Trading
Systems 201
If you haven't done so already, please read the
Trading Systems 101 article here.
Just
building a trading system that has a nice edge, and a great
reward/risk ratio is not enough. There's still so much more to
do in order to keep making great gains year in, year out.
Creating
a Suite of trading systems
A
starting amount like $100,000 will have little problem trading a
good system that trades only liquid stocks. However, after
several million dollars, it can be tough to get good fills on
very short-term trading systems. I know...what a horrible
problem to have.
Not
only that, but when you've made good money over many years of
trading, you tend to want to experience less drawdowns (I'm far
less willing to experience a drawdown now that I'm older...but
that's really a personal choice).
For
that reason, it's necessary to spread money around several
different trading methods. For example, I can allocate my money
between four trading system:
Profit
Taker - 50%
Profit
Taker 2 - 25%
Breakout
- 25%
Breakdown
- 50%
Profit
Taker is always trading even if it's a bear market. This
short-term system doesn't care. PT2 and Breakout are only in
when our Smart Money indicator is LONG, and Breakdown is only
trading when the Smart Money indicator is short.
100%
of my trading capital is ready to invest at all times, and the
trading methods use different criteria, so they're not stepping
on each other's toes so to speak.
When
I split the money between systems, I'm not simply trading just a
few stocks from each system. No, I'm spreading between 10 stocks
in each system. By doing so, I will move the market much less
than if I put all my money into one system.
When
the trend is LONG, I will spread money around 30 positions. When
the trend for the market is SHORT, I will spread between 20
positions for this particular suite of systems. Here are the
results from 1995-2006:
Yearly
Performance Summary
| Year |
Days |
Closed
Balance |
End
Total Equity |
Total
Equity Gain |
Gain
% |
#
Trades |
 |
|
|
|
|
|
|
|
| 1995 |
365 |
$135,717.09 |
$143,178.59 |
$39,827.65 |
38.5% |
192 |
| 1996 |
366 |
$172,893.71 |
$169,844.62 |
$26,666.03 |
18.6% |
243 |
| 1997 |
365 |
$227,694.38 |
$222,587.93 |
$52,743.32 |
31.1% |
229 |
| 1998 |
365 |
$405,343.93 |
$447,943.78 |
$225,355.85 |
101.2% |
261 |
| 1999 |
365 |
$1,019,471.66 |
$1,016,365.81 |
$568,422.03 |
126.9% |
313 |
| 2000 |
366 |
$2,252,943.16 |
$2,202,048.07 |
$1,185,682.26 |
116.7% |
274 |
| 2001 |
365 |
$3,649,384.05 |
$3,816,979.11 |
$1,614,931.04 |
73.3% |
270 |
| 2002 |
365 |
$6,582,822.74 |
$6,608,223.55 |
$2,791,244.44 |
73.1% |
294 |
| 2003 |
365 |
$9,535,367.34 |
$10,314,563.30 |
$3,706,339.75 |
56.1% |
400 |
| 2004 |
366 |
$13,288,559.90 |
$13,279,726.81 |
$2,965,163.51 |
28.7% |
417 |
| 2005 |
365 |
$15,263,380.69 |
$16,242,208.00 |
$2,962,481.19 |
22.3% |
436 |
| 2006 |
361 |
$27,534,735.32 |
$27,534,735.32 |
$11,292,527.32 |
69.5% |
467 |
With a CAGR of
54.1%, a MAR of 2.76, Ulcer Index of 5.46 and Longest drawdown
of 9.7 months, it's a very robust collection of systems.
System
Diversification
While splitting money between
different systems is great, just how related is each system to
each other?
For example, what
if I had a few systems that bought breakouts in different
time-frames? So system 1 might buy a 10-day breakout, and system
2 might buy a 20-day breakout. Is this really a good strategy?
My experience says no.
I like to use different styles for
each system. That makes the correlation much lower. For example,
the Smart Money uses data that has nothing to do with price. The
Breakout system buys breakouts (although it's more complicated
than just an n-day breakout). PT2 buys ahead of a breakout.
Breakdown shorts breakdowns. Base Hit buys on a dip.
By throwing these different styles
into the mix, you can often have a system zigging while the
other zags. This makes for a much smoother equity curve.
Side note -
I believe this is while Richard Dennis had a tough time getting
investors to stay with him later in his career. He made huge
amounts of money, but the drawdowns were large, and he did not
diversify between system styles I believe (can't say for
certain, but Curtis Faith, one of the original Turtle Traders
was only taught two breakout systems when he was trained by
Richard and Bill).
Knowing When to
Fold 'Em
One important question most system
designers fail to ask themselves is just when to stop using a
system? After creating systems and using them in the real world,
I know from experience that there are times when a system just
isn't working.
For example, I might create a
system with more than a dozen years of back testing and
thousands of trades. Without warning, the system might start
trading sideways, or enter into a normal drawdown and then never
recover. A year later, the system acts like normal and goes
straight up for several years.
These things are cyclical I've
found. I remember several years back when traders were talking
about the death of trend following. There was a drawdown for a
couple years, but then those systems started putting in stellar
gains. Currently, trend following in futures is behaving poorly
again. Nothing new here.
What I have done to combat this is
pretty simple really. I keep track of the individual equity
curves of each system. If one fails to meet my criteria (which
could be as simple as a long-term moving average of the system's
equity), I take it out of the mix. When it starts working again,
I "re-hire" it so to speak and put it back into the mix.
Advanced
Statistics
When
I first read about some of the traders out there making a
killing in the stock and futures markets, I had to wonder how
much of their success was due to being in the right place at the
right time. For example, what if some of the famous trend
traders like Richard Dennis or Ed Seykota didn't start trading
during the huge commodity bull market?
A
random look
The
answer to this question can be seen if we take all our trades,
and randomly scramble them up to produce several thousand
different equity curves. Some equity curves might start off with
50 winning trades in a row, or some could pick all losers.
Creating
these random equity curves is called a Monte Carlo simulation.
In
the above example, 100 lines were drawn onto the graph. The blue
line was the actual equity curve, while the grey lines were
randomly scrambled equity lines based on the trades generated.
When
I run a Monte Carlo simulation, I actually run 4000 different
equity curves. Such a huge number a samples can allow me to make
new observations. For example, I can now find what the annual
growth rate of the top 90% of equity curves was. If 3600 of
those equity curves made more than 40%, I have a 90% confidence
level that the CAGR will be 40% or greater per year. The future
is never known with absolute certainty, but this kind of
simulation really dives into the efficacy of a trading system.
You
can run a Monte Carlo simulation on just about any statistic you
want: MAR, max drawdown, longest drawdown, etc.
As
you can see, there's quite a bit of research and development
that goes into not only making trading systems, but actually
trading them. This is how professional hedge funds manage
trading systems. The more you treat trading as a business, the
more successful you will become.
Happy
Trading!
Previous
Article: Trading Systems
101
Click here for instant access to your free
trial
The results listed herein are based on hypothetical trades.
Plainly speaking, these trades were not actually executed. Hypothetical or
simulated performance results have certain inherent limitations. Unlike an
actual performance record, simulated results do not represent actual trading.
Also, since the trades have not actually been executed, the results may have
under (or over) compensated for the impact, if any, of certain market factors
such as lack of liquidity. You may have done better or worse than the results
portrayed.
