Just so I understand, are you judging the effectiveness of your QBA by whether or not the team with the higher rated QB wins?
If so, can you share with us the thought process behind that? While it looks to be a very good predictor, it is just surprising to me that QB performance alone would correlate so well with wins.
Or am I just totally off base in how I'm looking at your data?
The reason quarterback performance seems to correlate so well with wins is because so much of the team's offense is conducted through the quarterback. Even if he is handing the ball off on most downs he will typically have a higher TD % (the amount of total TDs he scores based on his own yardage) and a higher AYPP % (the average number of yards he generates on each dropback.)
Now, realistically I don't think it's possible to get a metric that could be accurate to much more than within 87 to 90% maximum without weighting for the actual score of the game, because many of the games that the metric fails to predict will be because the other QB was within a few QBA points of the other, making the outcome practically a dead heat. A perfect example of this is the the Patriots/Jaguars game last week that the Patriots nearly lost. The metric predicted the winner correctly, but only because Tom Brady had a 0.01 point advantage on Henne.
One thing I think my metric does do better than passer rating, and the reason it reflects more accurately on winning QB performances, is that it includes rushing yardage and also takes out meaningless plays like kneeldowns and spikes, and treats all yardage gained equally rather than judging a passing play to be more worthwhile than a rushing play.
But I wouldn't necessarily say that a QB performance makes up 85% of the outcome of a game, only that their good play will encourage trends from the rest of the team that will in turn reinforce his statistical advantages. So while I do judge the effectiveness of QBA in picking the winner it's not just picking the QB, but the total team offensive performance he led.