This is driving me crazy. I spent a lot of time demo'ing EAs on IBFX before going live. Now the results are significantly different. Same EA. Same settings. Any ideas?

Here's a few a several examples. For the period 12/4/06 to 1/5/07:

GBP M60:
Live +615 pips
Demo +192 pips

GBP M30:
Live +610 pips
Demo +21 pips

JPY M60:
Live -107 pips
Demo +163 pips

JPY M15:
Live +154 pips
Demo -43 pips

I've attached graphs

Attached Images

	P_007.gif (13.1 KB, 84 views)
	P_008.gif (19.0 KB, 47 views)
	P_009.gif (16.2 KB, 39 views)
	P_010.gif (17.1 KB, 45 views)

erm...both demo n live same time? or demo in dec and live in jan?

I found that starting the EA's at different time even 20 min apart, but especially at different prices can have a significant effect on the outcome. This all depends on the EA and what it's criteria is for entering and exiting a trade. But a lot of EA's open another trade as soon as one is closed, so by starting at a different level on the same EA will affect the outcome unless both start at exactly the same price (demo and live). Just one possibility, but from the looks of your charts, why are you complaining. :>)

Both are 12/4 to 1/5 like the graphs show, run concurrently on different computers (side-by-side, I have a demo PC and a live PC)

I could see the possibility of them being out of sync based on one being closed out slightly after another, causing one to reenter a follow-on position differently. But since they are the same EA with the same parameters, and being price driven, how they could differ to the point where some days one will trigger and the other won't. Since they are both price-data driven, how they could differ so much.

The other real issue is how reliable is demo data and optimizing around it. On the GBP M60, live was 615 and demo was 192 over the month, with the M30 live was 610 and demo 21. These are huge differences that getting slightly out-of-sync wouldn't explain.

did you check ur mt4 journal?

r both buying at same time?

same pc spec?

Depends on how much the stratergy depends on price data, the demo feed is missing some of the data in the live feed (as well as the other factors which could come into play, which have been mentioned above). I have come to the conclusion to really backtest well you need to do it with the exact broker live feed data, otherwise at lower timeframes backtesting, forward testing, it's all useless. There seem to be a lot of people bemoaning this situation on the net today, but Metaquotes just don't want to know.

when you open different accounts the history files are rebuilt from scratch. Its likely your demo history being older may have many segments of history from old builds, demo data direct , Alpari imports and from IBFX and metaquotes history. In addition your tester folders uses cached data from history folder.

To be sure your seeing fresh acurate data delete ALL history from your demo and live history folders, delete tester folders from demo and live and do this when MT is closed down. When you reboot MT it will remake a fresh tester folder from scratch. Now go to history settings and press the download button for 1 min thru to 1 month for demo and live and reboot AGAIN.

Now go to charts and open 1 min thru to month and drag back far as it will go for demo AND live this will get cached server data from IBFX and will resync history files.

Now when you re backtest everything should be from the same data.

Forgot to add that if you do this correctly backtest is not bad at all in fact very rare differences can be seen between live and forwards and backtest if everything is always done in 1 min tick testing. Only about 1 in 20 trades may have a stop hit in real forwards when backtest said it missed by 1 pip.

One last thing IBFX is not an honest broker if the forums are anything to go by. Demo feeds are untouched and usually most systems can show at least some profit. However, you start making any live profit over 1 full lot trading you will be met with requotes, delays, extra pips slipped against you and server pulled off line just when you need it. Not going into more then this as plenty already discussed elsewhere.

I don't understand why you need to do this after you have downloaded the history? can you expand on the quote?

Becuase your EA will use the brokers data not downloaded meaquotes data. By forcing all charts back it asks for broker data even though they only hold few weeks worth. The difference is subtle but if your MA's etc are looking at recent past history it might as well be your own brokers data.

What im trying to do here is make sure everything is in sync and as life like to real conditions as possible. Once you done that then the EA should perform the same.