Dallas TX/USA
Posts: 154 since Jun 2009
Thanks: 33 given,
283
received

I have no opinion of this particular method, but this part is important in a general sense. You can't figure it out any more than you can figure out why a coin will come up tails 6 times in a row sometimes. In some sense, that's what random means: you can't figure it out.
Whenever theoretical probability and reality come face to face, reality kicks theory in the nuts and runs away laughing. If the theoretical number is correct, then over time you can expect to converge on it with diminishing variance. In any short run, though, you can have incredibly painful "anomalies" (just like flipping a coin). And that's with any method, free, paidfor, or otherwise.
So, when you have a trading idea that you expect to work with high probability, you simply can't be swayed by small strings of results. Even with the coin toss, after a couple hundred trials, you can have 60% tails. It doesn't mean the coin is bad.
By the same token, you can look at a short run of results and it will look phenomenally good (like if you get heads 9 times out of 10). This is how you can advertise a worthless indicator without much trouble... you just show people the good runs and let them assume the rest. But it's also one of the reasons why supposedly "smart" backtesters fall down. If you consider the relatively small amount of data most retailers have available to backtest with, the average underdog backtesting trader is screwed before he/she starts. And they usually blame curve fitting! LOL
(not that curve fitting isn't another big problem)
