I have downloaded some of the source code for the winners of the strategy contest - why are the results I obtained vastly different?

That's right. The statistics shows the amount of trades, but the contest rules require to have at least 10 positions. A position can be opened with one trade and lately closed with 5 or 10 partial trades. But it's still considered that there was only 1 position.

I will also intervene to this discussion. Number 8 in the ranking was using @Configurable parameter and it is forbidden. So already 4 bad guys in top 10.

It depends on what period of time you are testing it. It is possible that the contestant started the strategy on a different date of the month.

That's true. We haven't noticed @Configurable parameter. This is a bad strategy for July. We will contact the contest participant in order to get some additional comments on this issue. Thanks for noticing the annotation but what are other 3 bad guys? We don't see anyone else.

those bad guys which people were talking about in previous posts. Number 6, number 9 and number 2 in the final table have less than 20 trades. It means less than 10 open positions. please eliminate them. Many recent posts were about this issue written by 'trader', 'cebesaru' and 'contest support'.

good luck

jaguar, There are many things that the contest organizers are ignoring and I do not quite understand their way of decision. Since I am not supposed to question their judgment, so I did not counter question them.

According to them, if you have order closing code in onStop, you are disqualified. But the winner of July's contest has that code. May be I do not understand their rules properly and so I try to code my strategies in a safe manner so that they do not fall under any rules which I do not understand properly.

My point is if you have a rule which says you cannot have something in your code, then there should not be 'if' or 'but' statement in it. That makes it a strong rule and a fair competition and judgment.

We do not provide with any online statistics but we monitor when the winning strategies were started and stopped. We have this information in our internal logs.

Regarding the rule that "the strategy must be running at least 10 trading days of the month."

Is there a minimum number of hours that the strategy must run, or is it sufficient that it ran at some point during the day?

Yes, the strategy must run 10 x 24 hours. That is, 240 hours totally.

Hi,

What is the reason that only 1 simultaneous positions may be opned ?
'◦The maximum amount of simultaneously opened positions is 1. '
I would agree that for one instrument, but why I cant open 1 position on EURUSD and one on AUDCAD ?

Also, if the strategy uses position.close() is this considered manual close ?

Thank you.

1) The idea is to prevent gambling and extreme positioning.

2) Position.close() is not considered a manual close if it s not in onStop.

personally, i don't think having a 1 order limit prevents gambling...
just as an example:
suppose your program would make one big trade at the beginning.
according to the rules, you are allowed to have 50% profit, so you take all your equity and bet it on something using take-profit 50, stop-loss 100.
with that one trade, going purely for luck, you can get to an equity of 150.000.
then you have to fulfill the other rules:
let the program run for 10 days and in this period let it make another 9+ trades with tight stop-loss and minimum budget.

that is a pretty stupid client which one could implement within a few minutes and it would give you a chance of more than 50% (because of the stop-loss) of getting into the top10.

i think, a rule like the following would prevent gambling better than the 1 order limit+10 trades:
The turnover of your trades within any period of 24h should not be bigger than 1/10th of your total turnover.

this would prevent making one big trade followed by an arbitrary number of tiny trades as well as making many equal, small trades at one particular point in time.

Dear Ocire,

I don't think your assumption is completely correct, since if you do one big trade with the biggest possible amount you be able to open, it would most probably end in a margin-call. This is because the market is not moving straight in one direction, it might be moving first a few pips against you, which is then causing a margin-call already.
The rules are more to stop extreme cases, but if you want to introduce a 1/10th rule, everyone has to implement this check, which I personally don't like. Everyone should be able to use his own style and methods as much as possible, the rules are just to have mutual basis on a fair playground.

Kind regards,
Sacrelege

PS: Otherwise, I would like to see you script in action next month

Single-position mode spurs everybody to look for a better entry/exit points. Additionally, it just does not allow you to increase the exposure and to average the trade price. Since the new rules were adopted, the performance of winners has apparently decreased but the amount of winners has grown.

@Ocire, we consider that your comments in the previous post were very interesting and valuable for us!
We acknowledge that the current rules are not perfect and allow anyone to success in the contest with the above mentioned strategy.

We discussed the strategy within our contest jury and we've decided to update the rules in the following way:
1) the contestant must keep the trade amount constant, e.g. all trades must be done with the same ammount. Though the trades can still be closed partially.
2) The maximum amount of profit on a single position should not exceed 25% of the actual balance (e.g. with the 1st position the balance can only grow to 125 000, with the 2nd one to 156250 et.).

These amendments will be added to the rules within 1-2 days and the new contest regulation will be adopted for September.

With these changes, the rules will be very close to its final version. But we will keep changing the rules until the gambling will be fully eliminated.

You may leave own suggestions and comments to the contest, like Ocire did, and we will scrutinize your feedback with a great interest.

@sacrelege:
you are right, i didn't fully account for the margin cut. however, i just implemented the strategy as proposed, to see which kind of influence that has. i used a take profit of 50 pips (around 35%, enough to be in the top10) and stop loss of 100 pips and ran it through the historical tester for the months of 2009/01 to 2010/07. In those 19 months, a buy order would make a profit in 11 occasions, a sell order in 13. As you can see, there are months where both types of orders will make a profit and the margin cut seems to only apply if you have a loss of more than 50% of your equity.
there is another thing though: the margin call trade doesn't get closed automatically, so in 2 occasions the profit was actually made by the margin call itself. (i find that a bit weird, however my knowledge of those things isn't enough to determine whether that is how the order should behave)

@Contest Support:
So that means, we have a one concurrent trade AND a trade amount based limit?

@Ocire, yes. We have one concurrent trade, one amount based limit on trades and we also have a limit on profit. Even if the trader will manage to close the first 10 million position with 25 pip profit, the equity will grow up to 125 000 USD. But there are some spread and commission costs on the other 9 trades that the participant needs to pay... If the spread is 1.5 pips and the standard commission is 18 usd/mio than the transactions costs will be about 15 120 USD, i.e. the end balance will be about 109 880 USD which, according to the last month results, does not guarantee you more than 500 usd prize

PS: Please note that the amount of trade is selected by a contestant. The strategy has a choice to open any trade amount but it has a requirement to keep the trade size fixed so that all positions will be more or less similar.