Message Font: Serif | Sans-Serif

No. of Recommendations: 2
Tharps courses in part deal with psychological issues that impair effective trading
Right, that makes me wonder too... The question is if this is a sign that his psychological lessons to "take control of your life" aren't effective. I think, statistically, a single sample is not relevant to make a judgment...

I hope you aren't insinuating that just because its software then the results are valid.

No, I just mean they used a kind of back-test software ( Athena) that takes its data from other programs such as Trade Station or Metastock: They had only to build the positioning algorhtms. It's simpler then having to build all by themselves, and, less prone to errors.

99% are curve fit

I cannot see how curve fitting would be relevant for this test.
As Tharp says, the very same signals are the input for all the positioning strategies. Curve fitting is relevant as entry/exit signals, but once those signals are set, the difference in results has to be due to position sizing.

Anyway, I've just downloaded Sux spreadsheet with daily data of RS-IBD. So I'm thinking in how to use this excellent database to perform this test using RS-IBD screen.

I have to think about a way to define the size of the position as a percentage of the total investment. So here it's best to think in terms of relative volatility. I mean, the volatility of a position in terms of the total volatility of the portfolio.
I think the best way is to define the volatility in every rank , first, in terms of volatility per dollar invested (VRi). then take the inverse of this figure

IVRi=1/VRi.

We define TIV as:

TIV= sum{IVR1...IVRn}
where n is the number of positions you take on the screen.

Finally, if C is the capital of your trading account, then, the position for every rank Pi will be:

Pi= C*IRi/TIV

The other question is HOW should we define volatility.
I mean what's the right time frame. short term or long term?. Definitely, using Sux database it has to be short term, but maybe the best way is to look at longer term volatility. Is it a better way to get the actual implied volatility figure used to get option prices ?

Ideas are welcome

FSC