$10,000,000 after only 60 trades. Although these are exaggerated numbers, they
illustrate the point that it pays to take it easy and do the math before you jump into
a trade. And it is exactly this type of math you also need to do while researching
your systems. If for nothing else, wouldn’t it be cool to know how much richer you
are than your cubicle buddy, while five cents still is a significant amount of money
to him?
The key point I’m trying to get across here is that there is a vast difference
between a good system and a profitable system. The most profitable system
doesn’t have to be the best system or even a good system at all, with the best
entry and exit points, and traded at the lowest commission. Those things help,
but what is more important is that a good system is a system that works, on aver-
age and over time, equally as well on as many markets and market conditions as
possible. The trades produced by a very good system don’t deviate from their
average trade as much as they do for a not-so-good system. A good system can
always be turned into a more or less profitable system by being applied to the
right markets and by altering the number of shares or contracts traded in each
trade. A system that is only profitable on one or just a handful of markets can’t
be made more or less good by being applied to more (losing) markets, no mat-
ter how aggressively we’re trading it.
The better the system, the more likely it is to hold up in the future, when trad-
ed on real-time, never before seen data, no matter in which market or under which
market condition it is traded. The same does not hold true for a system that might
show a profit here and now, but only in one or a handful of markets. To find out if
a system is good or not, we need to measure its performance, either in percentage
terms trading one share only, or in dollar terms always investing the same amount.
Unfortunately, none of the market analysis and trading software packages of today
allow you to do this right off the bat.
Because of this, I prefer to work with TradeStation, which is the only off-the-
shelf program I know that allows you to write your own code to compensate for its
shortcomings. When I am working with a system for the Trading System Lab
pages in Active Tradermagazine, I usually work in a two-step process.
For step 1, I attach the following code to the system that I’m working on,
with the normalized variable set to true. In this way, the system will always buy as
many shares as it can for $100,000, all in accordance with what we have learned
about buying or selling more shares according to the price of the stock. With this
piece of code in place, I can examine the results for each individual market using
TradeStation’s performance summary, which can look something like Figure 1.1.
At this stage of the research process, I’m basically just interested in getting a feel
for how many of the markets are profitable, and to get a feel for the profit factor,
the value of the average trade, and the number of profitable trades.
Step 2, which incorporates exporting the percentage-based changes into a
text file for further analysis in Excel, will be discussed more thoroughly later.10 PART 1 How to Evaluate a System