(Off-topic)
Rainier -
My letter that Julianne mentioned was in this last issue of TAS&C, and the letter was in response to an April article in the same magazine.
The editor was comparing the results you would get by trading a 'high volatility' system vs. a 'low volatility' system. In a nutshell, here's what he did:
1. He postulated a trading 'system' which generates, at RANDOM, trade returns of -50% to +50%, with an average trade return of 0% (zero profit, imagine that!). He called this the 'high volatility' system.
2. He then postulated a 'low volatility system' which generates half the returns of the high system.
3. He then took a number of random trades in the high volatility system, and calculated an overall return from those trades:
An 8-trade example from the high volatility system could be:
Returns of +36%, -13%, -44%, -35%, -2%, -9%, +28%, +38%.
The overall return in this example would be:
1.36 x .87 x .56 x .65 x .98 x .91 x 1.28 x 1.38 = 0.68, or -32%
By halving each return, the low volatility system would give:
Returns of +18%, -6.5%, -22%, -17.5%, -1%, -4.5%, +14%, +19%
The overall return in this example would be:
1.18 x .935 x .78 x .825 x .99 x .955 x 1.14 x 1.19 = .91, or -9%
In this example, the lower volatility system wins, since it lost less money.
He ran many iterations of this exercise, and found that the lower volatility system won a large portion of the time... indeed, when I ran 200 iterations in my Excel model just now, I found that the low volatility system won 73% of the time.
Now, I took exception to the article, because the set of INPUT assumptions were pretty bogus, in my estimation. After all, would you trade with a system that netted you an average profit of ZERO? Would you trade a system that allowed you to lose as much as 50% on a trade? If you did, you might not be a trader for very long! :-)
In my letter, I hypothesized an alternative 'high volatility system', which employs better money management (i.e., stop trades to limit losses). My 'system' allows a win of up to 25%, a loss of as much as 15%, and averages 5% profit per trade. A more reasonable system, in terms of the real world.
In my example, the low volatility system, again, is the same as the high volatility system, but with each return halved.
Running these two systems against each other gives COMPLETELY different results... I just ran my two systems against each other 200 times, and the HIGH volatility system won 164 times...
As you can see, the results you get from such an analysis are VERY dependent on the assumed inputs. I think my input assumptions are more reasonable for a 'rational' trader (to borrow a term from you!).
I told the editor I thought his article was misleading, and that he should have performed a sensitivity analysis to see how his model responded to VARIOUS systems, instead of just the one he used. He told me that while he didn't think his article was misleading, he did 'commend' me for testing the model using different inputs... then he invited me to share the work in an article.
As you can tell, this was much LESS an exercise regarding TECHNICAL ANALYSIS, and much MORE an exercise in MATH/STATISTICS, right?
Any thoughts?
Loren |
