I know it's pretty a given that Fitz is streaky. We talk about it here. We hear about it on TV. Often if a news outlet puts out a single blurb on Fitz, it sums up Fitz's entirety by calling him one of the streakiest quarterbacks in the league.
But how much quantitative data do we have to demonstrate that? I haven't seen much out there, and since I'm a data junkie so I decided to take a quick cut at this (after the jump).Methodology: I took most/all of the QB's that played for most of the 2011/2012 seasons, and plotted out their weekly QB ratings. I also threw in the last season or so of Trent Edwards and Drew Bledsoe for good measure. Now, I know QB ratings don't tell the whole story; that I can't capture in-game streakiness; that some QB's played in more games than others, which can skew the stats, etc. But just bear with me.
Average QB rating: I first took a simple average of those weekly QB ratings. No surprises here. Fitz ranked 18th out of the 27 data points. Average to slightly below average. Keep in mind, however, the data set doesn't include some real crappy QB's that weren't able to start consistently for the past season and half. But still, Fitz could be better.
Standard Deviation: I then took a standard deviation of those QB ratings. The higher that number, the more variation there is from game-to-game. Here, Fitz ranks about 10th out of 27, which is high but isn't nearly as high as I would've expected. You can call him worse than average, but not much more than that. In fact, Fitz appears to be less volatile than Bledsoe during his last year with us.
StDev/Average: Finally I divided the standard deviation by the simple average. The idea is that, if you have a high QB rating, you inherently will be biased to have a higher absolute standard deviation, and vice versa. A better measurement would be the ratio of your standard deviation to your average. On this measure, Fitz jumps to #6 out of 27. Who's worse than him? Bledsoe in his last season with us, Ponder, Cutler, Cassel and Palmer.
Anyway, there's more than I could do, but I need to hop into a meeting. Here's the raw data for those curious: