Shifting To Another Dimension
by John Sweeney
I have spent the last several months tweaking a simple trading system to provide good profits with
minimal drawdowns. The underlying idea ("Settlement," November 1991 through February 1992) isn't
too complicated, and it is summarized in Figure 1. Last month, 1 tried to exploit the underlying system by
adding on trades to the underlying positions with a separate, simple rule: Add positions when the trending
price retraces to the simple average following it (Figure 2).
This logic failed, though it was spectacularly successful in some test periods. Its maximum adverse
excursion (MAE) distributions showed no distinction between winning and losing trades, even in the
winning test periods, which forced me to reject the strategy for this system.
TRENDS TAKE TIME
It's tough to ignore that the underlying system completely missed several good trends in different testing
periods. I don't propose to "fix" that, because we'd then be back at ground zero with a whole new system
to validate. Instead, I point out only that one of the ways we visually identify a trend is the consistency of
price change over time. In other words, one of the ways we know we're in a good trade is that we're still
in it a few days after we get started. The good ones keep going and the bad ones are usually stopped out
immediately or in a few days.
After all, within the trade's horizon it can only go for us, against us or stay roughly neutral. The 30-40%
success rates so common in trading systems are a reflection of that. Our task, then, is to stay with the
good ones and reject the bad ones.
Just to give you an idea of how this looks, look at Figure 3. This relatively straight line is partly a result
of the simple fact that it takes time for prices to change. I've found that the slope of the line changes with
the trade able and its units of denomination.