On the website of the Automated Trading Championship 2006, the "Statistical Report #4" has been published (http://championship.mql4.com/2006/news/99/). This statistical report gives analytical results for 25 most profitable Expert Advisors as of the 7th of December 2006. The following characteristics of the Experts were analyzed: profit in points, efficient factor, amount of closed positions, position holding time average, percentage of profitable positions, total profit and loss, expected payoff. Its interesting: there is no expert with the efficiency factor exceeding 20%. Hope this information will be useful.
Maybe if you are interested, you could write a little program that simulates people (or "expert advisors") who trade at random. Then you can calculate the efficient factor of these random traders and construct a probability distribution of the efficient factor. Perhaps 20% is 4 standard deviations ABOVE the mean, i.e., perhaps it is damn good. Something like this: Code: for person = 1 to 1000000 step 1 for trade = 1 to 500 step 1 random entry random exit record trade result endfor calculate statistics for this person including efficient factor endfor sort all 1000000 efficient factors from smallest to largest construct probability distribution of efficient factor calculate mean and standard deviation of efficient factor plot distribution