I know this isn't an enourmous sample size, but maybe big enough to have some significance:

My OPR for stars.fr, most field sizes approx. 50%: 80 – 240 players, 35%: 300 – 1000 players and 15% 1000 – 6000 players.

My GM rating for 2015: 99,79% | played 4000 | ROI 23%

gold rating:     40,20% | played 600  | ROI -20%

silver rating:    99,88% | played 2500 | ROI 50%

bronze rating:  99,99% | played 900  | ROI 120%

Bronze games I play are mainly 3€ and 1€+R (turbo and 3x turbo).

Silver are mainly 5€+R and 10€ buyin, some 20€ and a few 2€+R and 3€+R

Gold are almost exclusivly 30€ tourneys, besides the daily 25€ turbo and the daily 10€+R. I feel like I have been running very bad in the turbos for the last two months (losing the important flips). I am looking forward to get a bigger sample. But the sample should be large enough to indicate some major leaks, even if I run below expected value.

Now, here are some questions: What would be a good idea to focus on when trying to crush the higher limits better (much better please). I know, this sounds very general, but what could be a good start to get an idea what I should focus on more. I use HEM2 and have all the data, just need an approach to access it.

I feel like my early game has much more leaks, whereas my push-fold game in situations with a small stack are near equilibrium but could be more exploitive.

Another question: How reliable is the OPR ranking and what do my numbers indicate to an experienced player?

Thanks