The problem with playing the junglers that you named in your post is that a lot of other people play them, and do really well with them. Therefore, the grading system (let's take your Gragas game for example) compares your 21/2/9 over 50 minutes as jungle Gragas, and compares it to everyone else. Some people may have scored about the same but in less time. They're probably in a higher grading bracket. Maybe they had the same stats but in a 40 min game.
I'm not 100% sure, but I am a programmer, and I know that I'd code that system to grade based on (simplified for time) Score Per Minute. So I add up your (kills + assists + 0.5(CS))/round(minutes) = score. Take that score, and if it is better than the top 10%, give you an S-. Better than the top 7%? S. Better than the top 2%? S+.
Just an example, but you get what I mean. In my mock grading system above, let's pretend you have 200 cs. So your math would be 21+9-2 + 0.5(200) / 50. That would work out to be a score of 2.56/min. If I take those same stats at 40 min, it would be a score of 3.20/min. That's a pretty big difference, even though your score was still great.
Make sense?
