is this possible?!!!

after a long time of hard work , finally I wrote an expert wich was profitable , but today when I tested it again the results were changed . same expert , same parameters , but different statements . can anyone help me with this?
the first picture is the expert statement before today and the second picture is today's statement .

Are you sure the lots used are the same??? the patterns of the graphs are identical... but the difference is in the values...

yeah you're right . I double checked and lot sizes are the same . I used 0.1 in both tests . and the other odd thing is that when I test it with my demo account it even turns negative results!

Probably an issue of modeling quality. What was the modeling quality on the previous test(s) and what is it now? Assuming the dates you used are the same, check the time frame used and the modeling quality.

I don't understand rabid , what do you mean by "modeling quality"?

Flip thru the tabs after the backtest. You'll see where it gives you return stats and modeling quality.

What is modeling quality?

Consider this: You have an EA that goes long when the price breaks from the daily high. You backtest it using daily data... best it's going to do is go long on the day after the price breaks, not the day of, changing the entire strategy.

On the other hand if you use intra-day data you'd be able to take the trade sooner after the breakout. Depending on the grain of data you're using and how much of it you have, this might be anywhere from 1 minute to 4 hours after the breakout... a huge range w/ vastly different results at each.

That's where modeling quality comes into play. It's an attempt to compare how accurate your results are to real time. It's never going to be perfect, but the higher the mq the better. What this means is that you can effectively keep the same settings for a backtest, but if your data has changed any... it's a whole new ballgame.

Anyway the whole thing is a bit tricky. Use google or the search at the top to search for modeling quality, or "90% modeling quality" or something similar. There's a step by step guide out there that tells you how to find, import, translate data, then test your EA on it.

Your modeling quality should be 90% yours is N/A.
Open the history center and let it download the data for all timeframe for the pair you want to test. It might help to first erase all the data.

thanks I got it . I don't have the results of the previous statement but I think it had lower mq . in the new statement the model quality is 90% . I think I should forget this EA . thanks for your very useful information