If you go preflop and have 75% chance of winning the hand and you win it, you win 100% of the chips in the pot. But ev defines that you only *deserve* 75% of the chips (because your chance to win them has been 75% only). You are Gain now is bigger than your EV gain.
If you lose in the same situatoin, you are gaining 0% of the chips in the pot, but would had *deserved* 75% of the chips. Now your Gain is lower than your EV Gain (EV line).
The EV line - if anything - will only tell you if you were lucky or not. It will not show you, if you played the hand well or not. Furthermore, you will need a gigantic sample size in order to make the EV line *telling the truth" that is because there is also short term variance. F.E. if you go all in preflop with KK and your opponent has AA it does not mean that your play has been bad - but right now - according to EV you wasted chips.
If you are good or bad at playing...you should find out by analyzing your sessions - not your stats.

I've been playing $7,$15 about 5000 games.(my overall ROI is 4-5%)

 
A result is not stabilized in case of 500 games, but the modestly stable if 1000 games.

If you just started playing there's a high liklihood you are a bad player (no offence intended at all).
In any case, how good you are cannot be determined by your EV line.
 
FYI: What the EV line actually shows
http://www.pokertracker.com/guides/PT3/tutorials/expected-value-ev-graphs
 
Focus on the best (most +EV) way to play each hand rather than your graph and resist the [natural] urge to be results-oriented.