Not to say the rest of your post isn't, but this bit is pure drivel. With a tiny sample size of games (n<30 for both with and without) you've been able to conclude that Bannan's performances must have been good due to a variation in winning percentage with and without him? Horse shart.
I agree it doesn't prove Bannan's performances were "good" but it does show his performances weren't, on average, bad. If his performances were bad that would mean the team's best form occurred whilst playing with one bad player. That doesn't make sense. We can't prove from that record he performed "good" but it does show he wasn't bad.
No it **** doesn't. It shows diddly squat. It's statistically insignificant.
It doesn't take into account the strength of opposition, the form of the other players or even Bannan's tangible contributions to the games.
I'm pretty sure it wasn't our best form either.