Evolution of Trading SystemsAlmost one year has passed since Automated Trading Championship 2006 started. The new contest of automated trading systems, Automated Trading championship 2007, will start very soon. How have the views of developers changed over the last year?
The Championship of the last year was symbolic, unique and useful. It brought together a good many various trading systems having different algorithms, characteristics and realization levels. Thomas Bopp says that "it was a great event because everybody knew in the end at what level his or her Expert Advisor stood. When someone develops an Expert Advisor, he or she doesn’t know whether it will be a good one or not. The Championship helped see it". In Thomas' opinion, the contest also helped developers of effective systems to go through paces and to make a statement. Automated Trading Championship 2006 helped to see problems Expert Advisors' developers face regularly, to understand what mistakes they make, what they omit when developing Expert Advisors. We saw giddy ups and disastrous failures in the last-year contest. Expert Advisors burst into the Top Ten like the wind and left the first page as quickly in order never to appear on it again. The most shining example is, of course, Sashken's Expert Advisor. He gained huge profits due to an error in the code and stopped only by Margin Call. There were such Participants whose profitable strategies also traded 0.1 or 1.0 lot due to some sad mistakes. They all did not take prize places nor they shared prize money. This privilege was got by those whose Expert Advisors contained fewer errors and had positive expected payoff.
However, it does not go like clockwork here, either. The Winner of the last-year Championship, Roman Zamozhniy (Rich), confesses that his victory is a matter of chance to a large extent. Indeed, if we look at how Participants alternated in the Top Ten, we will have to admit: there were elements of chance for everybody. For example, Hendrick was among the top participants along with Ldamiani for a long time. However, by the end of the contest Hendrick had been the 5th while Ldamiani had held down the 2nd line in the list. Such as there were, all results obtained within the framework of the Automated Trading Championship 2006 turned to be useful. Having won, Roman Zamozhniy remained unsatisfied with his Expert Advisor. He wrote a new EA for Championship'2007 having partly used the results provided by his previous Expert Advisor: "I have fully revised the list of indicators my strategy is based on." As often happens, a negative result is a result, too. Ahmed Soliman (CodersGuru) failed at the last Championship. However, he was not disappointed in automated trading. Ahmed admits that he "learned the lesson very well" and thinks that it was that fact that helped him to prepare for the new Championship better.
Championship 2006 changed much in the developers' views. Participants of the last-year Championship drew the conclusion that it was very important to review their trading strategies fundamentally, choose other analytical tools or systems of risk and money management. For example, Ahmed Soliman parted with technical analysis and developed an Expert Advisor working on news analysis. Now developers see a profitable Expert Advisor differently than one year ago. In 2006, they named a wide range of parameters to be the basis of profitability for an EA: from analytical tools to programming level. One year later, nobody mentions quality of programming. Everybody seems to be sure that any errors may cause fatal losses and to have pulled up their programming skills. Money management is more and more frequently mentioned as a factor making a positive impact on the profitability of an Expert Advisor. Andrei Moraru supposes the basis for trading success to be quite simple: It is "a competent money-management strategy and stable forecasting of successful entry and exit points… The attempts to create an automated Expert Advisor without these two qualities are doomed to failure." Al Parsai, the developer of commercial Expert Advisors and the author of the PipBoxer package, thinks that a perfect Expert Advisor must consist of successful strategy, risk-management and money-management systems. He emphasizes: "Without money management the odds of winning are very low." Ahmed Soliman is even more categorical: "Good money-management is undoubtedly the key to success, for any Expert Advisor." Well, after the 1st of October 2007 we will see whose money-management system is more effective. It must be noted that the level of all last-year Expert Advisors was still rather low. According to Thomas Bopp, this will look different in this new Championship: "Every programmer will do his or her best to develop an Expert Advisor that will get into a profitable area". As a result, competition will grow. The larger amount of Participants will also contribute to this. According to Ahmed Soliman, the most of the last-year Participants just tested their strategies and EAs: "The Idea of the Championship itself was a new one and I believe most of the participations weren't so serious; perhaps they were somehow doubtful about the idea. I'm sure this year will be 100% different in the quality of Expert Advisors and you will find the intention of winning very obvious." It is interesting, but not everybody sees the growth of competition in the same way. Roman Zamozhniy is sure that Participants should not be afraid of the competition: "We fight against the market, and not against each other." Well, we shall see. One thing is undoubted: The struggle will be more dynamic and exciting this year. Registration of Participants will be closed in a few days. As of the stage, 455 developers have already been registered. Less than two weeks are left till the beginning of the Automated Trading Championship 2007. Created: 2007.09.18 Author: MetaQuotes
yes, the publication of the testing results would be more than welcomed
varavee wrote: No. after 21 Sept Can you Rank Backtest of All Participants (maybe 1-20) Because I found something like 2007.09.19 16:36
varavee wrote:
after 21 Sept Can you Rank Backtest of All Participants (maybe 1-20) Because I found something like Some Expert has Good Backtest but fail in Forward Test and Some Expert has Fail Backtest but Goodl in Forward Test (maybe time for open server at home not enough) I hope to see Expert that have a Good for Backtest and Good for Forward Test. sorry for my poor English. Regards
Also of interest would be the ability to run backtest on the same period after the competition has finished to compare an EA in forward and backtest over what should be the 'same' data. Will the accounts be available for testing for a short while after the competition? The copy of a question from Russian version of article. goldtrader asks:
Question to Ahmed Soliman (to Al Parsai - correction by MetaQuotes): Are you going to participate in the competition with PipBoxer? It may be a good
chance to see the position of your product among the competitors? to prove its
quality and to promote it. I can harhly imagine the better way of public forward-testing
exept real-trading. It would be interesting to hear the answer to this question
from the other developers of comercial EA. 2007.09.19 10:38
after 21 Sept Can you Rank Backtest of All Participants (maybe 1-20) Because I found something like
Some Expert has Good Backtest but fail in Forward Test and Some Expert has Fail Backtest but Goodl in Forward Test (maybe time for open server at home not enough) I hope to see Expert that have a Good for Backtest and Good for Forward Test. sorry for my poor English. Regards |
|









The copy of a question from Russian version of article. goldtrader asks: