PureBytes Links
Trading Reference Links
|
How well did a trading method work in calendar 2000?
* "Buy-and-Hold the S&P 500" gave a total return of -9.7 percent
* Several bigname money managers blew up & closed their funds
during 2000, most notably David Druz and Richard Dennis. This
also happened to CTA's that post messages right here on the
Omega-List. They blew up, lost a huge chunk of their
investment, and closed their funds.
* The Top Performer managed futures fund (Futures Magazine,
Jan 2001, p. 22) made +19.32 percent. The average performing
futures fund made -4.6 percent.
But some trading methods did quite well. The fellow who led
the Robbins Trading Contest made >800 percent in 2000, and the
second place contestant made >300 percent.
I'd like to point out that the Pretty Good Oscillator ("PGO")
trading system also performed nicely in 2000. It made +123%
during 2000, with a max drawdown of -20.8 percent. For those
who are interested, PGO's Sharpe Ratio was 2.6 in Y2000.
Buy-and-Hold's Sharpe Ratio was negative in Y2K.
PGO was released into the public domain back in 1999, so its
performance in 2000 is completely without hindsight or data
fitting. PGO is a long term trendfollowing system that uses
the exact same system-code and the exact same parameter values
to trade 20 different commodity markets out of the same account.
The markets are drawn from four commodity groups: the oils-
energies, the currencies, the bonds-notes-interestrates, and
the softs.
Here is a hyperlink to the original unveiling of the PGO
system, which occurred right here on the Omega List.
Thanks again to the nice folks at PureBytes for the archives!!
http://www.purebytes.com/archives/omega/1999/msg29305.html
Note the date: October 1999, i.e., before the year 2000.
You'll also note that there was a teeny puzzle included along
with the Omega EasyLanguage code, which was quickly solved
by Dennis Holverstott and Mark Jurik.
So, a couple weeks ago, I decided to test PGO to see how it
fared in 2000. The answer is: quite well, I think. Here
are a few summary statistics for PGO in Y2K:
.... 123.28 ....... Compound Annual Growth Rate (percent)
..... 20.80 ....... Max Drawdown (percent)
..... 5.926 ....... (CAGR / MAXDD) Ratio
..... 45.36 ....... Annual Standard Deviation (percent)
..... 2.608 ....... Sharpe Ratio
As before, I also ran the MAXIMUM SLIPPAGE test. In this test,
we rig the software so that buy orders always suffer the maximum
possible slippage: all buys occur at the High of the day.
Similarly we rig the software so that sell orders always suffer
the maximum slippage: all sells occur at the Low of the day. This
idea came from author Fred Gehm; it's a torture test designed to
see whether a system is robust against slippage.
In calendar 2000, the PGO system executed 66 trades. Even with
MAXIMUM SLIPPAGE applied, and also a commission of $25.00
deducted, 28 of those 66 trades were winners and 38 were losers,
a win percentage of 42.4%. This is a tiny bit better than the
previous test result for 1987-1999, which came in at 41.5% (on a
total of 648 trades). The average profit per contract in this
test was $1,397. Big winners in 2000 were Unleaded Gasoline,
Lumber, British Pound, and Natural Gas. Big losers in 2000 were
Cotton and 30 Year Bonds.
For completeness, I re-ran the full test using the full history,
carrying forward from 1/1/87 to 1/3/2001. These longer term
results are pleasing as well. The full equity curve can be
downloaded (it's a MS Excel spreadsheet with a graph) from
http://www.mjohnson.com/trecipes/pgoy2k.xls .
Some people like to look at summary tables, others prefer to
analyze the equity curve, so I've provided both.
... 80000.00 ....... Starting equity
202603393.00 ....... Final equity
202523393.00 ....... Net profit
...... 73.57 ....... Compound Annual Growth Rate (percent)
...... 45.13 ....... Max Drawdown (percent)
........ 239 ....... Days in longest drawdown
...... 1.630 ....... (CAGR / MAXDD) Ratio
...... 46.11 ....... Annual Standard Deviation (percent)
...... 28.83 ....... Average max annual drawdown (percent)
..... 10.548 ....... Percent of days making new equity highs
...... 1.487 ....... Sharpe Ratio
...... 2.127 ....... Semideviation Ratio
...... 6.497 ....... Return Retracement Ratio
...... 1.832 ....... Sterling Ratio
... 19870101 ....... First day of test
... 20010103 ....... Last day of test
....... 3650 ....... # days in test
...... 75.00 ....... Slippage and commission per contract
You can download the PGO system itself, from this URL
http://www.mjohnson.com/special_needs/pgoy2k.zip
Notice that this is a .ZIP archive, which contains several
files and folders. You can unpack it using, for example,
Winzip. A demo verion of Winzip may be downloaded from
http://www.winzip.com
The PGO system is presented in a variety of formats: plain
old ASCII, TradeStation 4 Easy Language Archive, TradeStation
2000 ELA, and so on. If you're an owner of Trading Recipes and
are having a hard time getting the PGO system up and running
on TR, pop me an email (mark@xxxxxxxxxxxx) and I'll try to help.
The trading system itself is shockingly rudimentary. There's
an oscillator (the so called Pretty Good Oscillator) that swings
above and below zero. When the oscillator swings above +3.0,
that's a BUY signal. When it retreats below 0.0, that's an
EXITLONG signal. When the oscillator swings below -3.0, that's a
SELLSHORT signal. And when it advances above 0.0, that's an
EXITSHORT signal. It's pretty simple.
If you've got Omega TradeStation, you don't need to take my word
for it: you can test PGO yourself and refute (or confirm) my
results. You can put in all of the bells and whistles that I left
out, such as filters, triggers, protective stops, profit targets,
trailing stops, and so forth. Go ahead, don't be scared.
Play with it, change it around, monkey with its operations.
Here are a couple of additional hyperlinks related to PGO:
http://traderclub.com/discus/messages/18/444.html?SundayJanuary1420010654pm
http://traderclub.com/discus/messages/18/1117.html?SaturdayJanuary620010819am
I hope you enjoyed reading this.
Mark Johnson mark@xxxxxxxxxxxx
|