Long Term Secrets To Short-Term Trading

Figure 2.1 Test of a timing system.
Our first test was on Soybean prices from 4/29/75 through 1/1/87, and it looked at all possible 
combinations of short-term averages from 5 to 50 days against the longer term or second average from 10 to 
60 days.

---

25 
The best result, during this time period shown here used a 5-day average against a 2 5-day average. This 
time-based "system" made $40,075 with 54 profitable trades out of a total of 153. Wow, have we 
discovered a money machine', 
Figure 2.2 shows what would have taken place had we traded this system from 1/1/87 through 
4/23/98. The results are not promising. Whereas our accuracy at 31 percent winners on 163 trades has 
improved, we actually lost money, $9,100 to be specific and along the way suffered a drawdown (how 
much the system went against you before getting back in the black) of $2 8,612. Putting up $2 8,612 to lose 
$9, 100 is hardly a good wager, The average profit per trade was $-55. What happened to the original 
cyclical or time influence? Beats me! 
Reversing the process, I checked to see what two moving averages worked best in the second time 
period from 1/1/87 into April 23, 1998 (see Figure 2.3). The best combination was a 25-day moving 
average against a 30-day. This made $34,900 with a nice 59 percent accuracy. This system made S234 per 
trade and had a drawdown of $13,962. This too, is not a good bet. 
Applying the best case results back on the earlier data, out of sample, produced a loss of $28,725, as 
shown in Figure 2.4. Forward, backward., the time or length or cycle of the moving average that worked in 
one time period does not work in another time period. 
"Perhaps," you query, "The problem is not that time does not work but that Soybeans do not trend 
enough." 
What appears next is the best case study of a moving average crossover system on the British Pound, a 
very trending market. From 1975 to 1987, the best such crossover system was a 5-day average versus a 
45-day, making a most impressive $135 443