krana posted:
   great work.. i would also liek to see more real market test and how it works .. as we all know demo is no proof :)

so please share your findings. ...


Thanks


Thanks Krana!

Yes, I also want to see more tests before trading real money with more than 1 microlot.

But I will disagree with 'demo is no proof' in this particular case. Considering the EA enters trades at normal times (as opposed to MDP which trades only on volatility spikes, for example), a demo and a live account would have almost the same exact fills.

Yes, depending on the broker you could get slipped a bit on the real account, which does not happen on the demo, and that's why I said 'almost the same exact fills'. I plan to move the EA to real money account(s) after it proves that the forward testing is profitable. Again, for the type of trading it does, the fact that forward testing is profitable on a demo account would be very much an indication that it would be profitable on the same broker on a live money account.

As a matter of fact, I might start a real account forward test sooner than I thought, just to test what I said above...

Cheers!

Thanks a lot, Will!

I'm 99% sure the algo does not use ticks. I have 2 reasons for saying that:

- processing for trades (open, change SL, close) is done at 15 minutes mark (exception: trades closed via SL, but that happens outside of the algo - the broker just closes the order because SL level is hit); that tells me the algo is only doing its calculation at the beginning of each new M15 bar, and does not care about intra-bar ticks
- backtesting was extremely fast; it was so fast that in the beginning I though something went wrong and the terminal just ran through the whole data just reporting errors. That event also supports the M15 processing idea, since there there were 500 TIMES LESS bars than ticks in the selected time interval

Me doing tickdata backtesting had more to do with the fact that I trust the quality of that data a lot more than the default MT4 history. Plus it's the best type of backtesting, in terms of accuracy, so even if I knew upfront that the EA only processes signals every 15 minutes, I would still have used tickdata.

As for real money account from the vendor, I did not see anything like that, which most like means he does not have one that he's willing to show. Would not be surprised to get a 'demo accounts are performing the same for this EA' type of answer, which I actually believe, for this kind of EA (see the answer I gave above to Krana's question).

Yes, the 2 accounts where I'm doing the forward testing took different trades, but with so little history I would not jump to any conclusions yet. The nice part is that the EA is indeed taking a lot of trades, so in 1-2 weeks we'll have a lot more data for evaluating the EA's performance.

Take care,

Remus

another tip. if you run BT in visual mode just minimize the visual chart. it will speed it up.

Good work, Remus! Thanks

This comment in general. I am just pointing some issues. I am not against or fan.
data from original web page:
BT;
EURUSD : 2007-2011
USDCAD: 2010-2011!!
GBPUSD: 2010-2011!!
AUDUSD: 2010-2011!!
USDCHF: 2010-2011!!

- 2 years of BT is not enough!!
- EURUSD what is the result for 2000-2006??

MDP has hard coded setup change in its mql file. which that made me stay away of using it.( this means MDP is a semiautomated EA if trader can manage to do it ) code for the new EA should be decompiled and checked for the same issue. hard coded setup change can also be done in DLL. to understand that the DLL function must be checked if any date/time data is been used.

fyi
walker