Is it not same between Strategy Tester and Real Trades ?

I've tried few times, and the results a lot differences trades between Strategy Tester and when I let my EA do the the trades lively.
In ST, model i used "every Tick", same time, same period, same EA with the real trades online, but I got a lot differences Trades and it makes me lose a lot of money. Sometime have same open time trades but difference close time. Sometimes the trades have same time but difference action (between BUY and SELL).
It makes me headache.

What happen ? and how to handle it ?

Hi, Zap
thanks for your refferences, I got lighten by its.
As I absorb from those refference, when we do the backtesting my EA got the data instantly to do trades ontime, but in real market we got delay our orders so thats make it difference (when my EA close a trade, maybe there's changing price in real market).
But sometimes in my EA when a trade is close but in real still running, is it because of the slippage that I set ? What is the realistic splippage in real market ? or because lag ? cos sometimes in my MT4's time is not smooth (sometimes from 1st second jump to 4th - miss the 2nd and 3rd) is it normal ? or because my network lagging ?

My conclusion : (please fell free to correct me if I'm wrong)
So we can't expect much by backtesting's resut, so I better looking for stable EA with bigger slipage for ensure the profitable EA.
To get almost same trades between backtesting and real market, in backtesting we must use actual time-1 or 2 (1 or 2 bar previous actual time) to test but in real market we use actual time. Will it work ? Which one better 1 or 2 bar prev to get nearly same with real market ?

In your statement in 1 of your refference, you can get almost same between backtesting and real. How ?
to get nearly realistic result, better backtesting in what time period ?

It is not the slippage which makes the big difference, but inaccurate data. M1 data will not be enough for accurate testing, because the ticks are simulated, and can be really different from real ticks.
You will need good quality historical tick data to perform accurate tests. This can cause extreme inaccuracy in some cases. (even 100000%). Of course all the above depend highly on your style of EA.

Even when using real ticks, the inefficiencies in realtime trading (slippage, changing spread) cannot be backtested, but as I saw it causes only around 10% inaccuracy. It cannot really be reduced, but it is acceptable for me.

The most important is the EA style itself which it using the real tick or using fix point/data, such as: data in previous bar etc will not have big effect as long not using ask or bid (ontime data), etc which the data unreliable and the result of backtesting is more reliable when using fix data. is it correct?

I think I have to built new EA which using fix data.

How about you? you said you can make EA that the result almost same between backtesting and real market.
If u would kindly share the strategy with me, you can email me at andy_goen(@)yahoo(. )com

Thanks 4 everything.