On the website of the Automated Trading Championship 2007, the article The Automated Trading Championship 2007 Is Over! has been published. 12 weeks of the most interesting, most exciting event in the world of automated trading - the Automated Trading Championship 2007 have passed. All Expert Advisors have been stopped. It's time to acclaim the Winners and hand them their prizes. The full text of the article was published on the website of Automated Trading Championship 2007 in the News section.
Two of the top 3 are from the Ukraine.. those guys must be smart, I heard that the Soviet Education system was top notch... The winner multiplied his account by 2.3 in 12 weeks, he had to average a little more than 7% per week to do that.. I think there are traders that can beat that without neural networks actually... it's sort of like the IBM computer using "brute force" methods beating a chess champion.. apparently trading is not quite as complex as chess. Nonetheless those guys are far enough ahead of the crowd that it would be silly for them to start hedge funds, they would be trading huge size in a couple of years and keeping all the profits at the rate they grow an account...
Interesting how the Yank in second place based his system around daily timeframes, considering the competition was for 12 weeks.
Actually, you're both wrong. The winner's neural network-based EA started with $10,000 and, after 12 weeks, ended with $130,475. That's a 13-fold, or 1,205%, return. That's averaging a little over 80% per week, conservatively assuming weekly compounding (rather than actual continual). Done over 647 trades... all those statistical-fluke theorists are going to have a field day... not. The maximum drawdown was 28.58%, early on. Leverage averaged around 10:1. Hardly excessive, considering 1) the typical frequency and duration of trades and 2) that the instrument of choice was a relatively pedestrian, well-behaved EUR/USD. Human, manual traders that can beat that? Yeah, that's the ticket.
I agree wholly with your views. Risk/returns look very nice. 10:1 leverage is probably the maximum i would take personally.
The contest organizers may want to allow bystanders to invest in one or more of the EAs. EA authors ("contestants") put up $10K of their own money, as before. Passive investors could wager any amount from $5K to $50K, and divide it up among the contestants however they wish, before the contest begins. When the passive investor wagering is closed, the contest organizers would add up the additional amount of account equity for each contestant. Let's say contestant "Jill" received a total of $3K of wagers by passive investors. Her account is now $13K at the start of the contest. The contest begins, Jill's EA trades, and everyone gets very excited: the contest organizers, the contestants (authors), AND the passive investors. At the end of the contest, Jill gets to keep (10/13) of her final account value and the passive investors that wagered on Jill get to keep (3/13) of her final account value. The contest organizers get a huge amount of additional attention, exposure, and recognition. Everybody wins. If it were to happen, I would love to know the final result of investing an equal amount in EVERY contestant. Don't even try to identify the good ones or the bad ones beforehand, just invest in them all equally. Run the contest and see what the final profit or loss happens to be. If it's profitable, the Royal Bank of Scotland (or Goldman Sachs, or Renaissance Technologies) may want to duplicate the entire effort in-house.
"We checked quite easily 2000 prospective participants and two thirds of them were not admitted. However it was almost impossible to check the remaining 603 participants manually." That is quite a number of possible cheaters. They did check the top 10 manually....
Interesting thought. Unfortunately for the likes of RBS, GS and RenTec, looks like that would be a losing proposition. The combined participants' equity after Week 9 (with 3 weeks to go) has been published by the organizers. Here's the bottom line as of Dec. 2nd: That's a 34% loss. The final combined participants' equity has not been published, as far as I know. (Although anyone with enough time on their hands could add up all 603 individual equity numbers here, say, by doing repeated copy & paste from html into Excel; or simply ballpark the average of each page, times 15, repeat 40 times, add'em up.) In any case, skimming the final standings, I'd be surprised if the combined participants' equity at the end were significantly different from $3.97M after 9 weeks, give or take a few percent.