Monday, August 13, 2007

A Discussion I Keep Getting Into

It seems like approximately once per month, I end up on some message board responding to someone who says something to the effect of, "South Carolina will never be any good as long as they have Blake Mitchell at quarterback," or, "Blake Mitchell sucks." I always respond with a post defending Blake Mitchell and saying that he will be one of the better QBs in the SEC. Then I always get a response expressing incredulity at how I can possibly defend a quarterback as terrible as Blake Mitchell.

It has become conventional wisdom that Blake Mitchell is a mediocre QB. The conventional wisdom is wrong. Blake Mitchell is not a great QB, but he's a pretty good QB. Short of great, but better than average. The proof? Just look at his stats.

Of course, I have talked about evaluating quarterbacks statistically in the past. I look at rate stats exclusively, which I think is the proper thing to do. Of course, a team that throws the ball 400 times in a season may be less effective at it on a per-pass basis than a team that throws the ball 300 times due to the defenses keying on the pass, but I think the sins of looking at cumulative stats is even worse. If you look only at cumulative stats, you start overrating system quarterbacks or quarterbacks on teams that fell behind in a lot of games and had to pass a lot.

I developed a formula for evaluating QBs:
Rating = (yards passing + yards running -yards lost by sack + 20*Touchdowns -30*Turnovers)/(# of designed passing plays).
I would have preferred a more sophisticated formula that took account of first down throws, but the statistics just aren't available in a form I can use. This formula however gives results that are useful in that they account for yards per attempt, touchdowns, and interceptions without artificially inflating the value of short completions like the conventional passer rating does. It also accounts for the skills a running quarterback brings by looking at rushing yards, rushing touchdowns, etc.

In evaluating SEC quarterbacks from last season, I was surprised to discover that Blake Mitchell had the second best rate stats in the league, behind only Jamarcus Russell. How did this happen? Well, Mitchell had a very high yards-per-attempt statistic, second only to Russell. Higher than Andre Woodson's. Higher than Erik Ainge's. MUCH higher than John Parker Wilson's. Much higher than Brandon Cox's or Matthew Stafford's or Chris Leak's. Granted, he had much fewer attempts than any of those guys, because he wasn't the full-time starter. But, he also had a better ypa statistic than the OTHER quarterback on his team.

"But he is inconsistent," they say. Well, he was benched at one point for several games last season. That is true, but was he actually an inconsistent quarterback? Again, the statistics say no. At least, they say that if Blake Mitchell was inconsistent, so was every other quarterback in the SEC. Let's compare Mitchell to a few of the better QBs in the SEC last year: Leak, Russell, and Wilson.
  • Inaccurate games: Mitchell had one game last season with less than a 50% completion rate, a game in which he only had 11 attempts. Jamarcus had none. Leak had none, but had one game with exactly 50% and another very close to it. Wilson had two such games.
  • Inaccurate games, part 2: Mitchell had two games last year with more interceptions than touchdowns. Jamarcus had one. Leak had 3 games with more interceptions than touchdowns. Wilson had two.
  • Games with fewer than 200 yards passing: In his 7 games with greater than 20 attempts, Mitchell had 2 games with fewer than 200 yards passing. Jamarcus had 2 such games (with more starts, of course, and to be fair he was much closer to 200 in those games than Mitchell was in his). Leak had five (!) games with less than 200 yards passing with greater than 20 pass attempts. Wilson had two.
  • Games without a touchdown pass: This is where Mitchell really falls. In 4 of the 7 games in which he made 20 or more pass attempts, he failed to get a touchdown pass. In one of those, he managed to get a touchdown run, but to be consistent we should say he wasn't particularly good at putting the ball int he end zone consistently. Russell had one game where he failed to score. Leak also had one. Wilson had three.
OK, he's not Jamarcus Russell, but no one is. Mitchell is not quite in that league, but neither was Leak or Wilson. No one is, except maybe Woodson, who doesn't have the supporting cast around him. Mitchell's consistency compares favorably to that of a guy who won the national championship and a guy who many people believe is on his way to an All-SEC season.

The response is, "But he had Sidney Rice." Rice was a good receiver, picked in the 2nd round of the NFL draft. Russell had two 1st round receivers and another who likely will be. Leak had Dallas Baker, Caldwell, and Harvin. Wilson had DJ Hall. Ainge had Robert Meachum. Woodson had Keenan Burton. Chris Nickson had Earl Bennett. It's an argument that can be made of ANY half-way decent QB in the SEC last year, and therefore it really can't be used to single out Mitchell.

But why was he benched? Well, according to Spurrier, he was benched because his offensive line wasn't very good and he wanted a more effective running quarterback in there. When his offensive line solidified, Mitchell returned to the starter spot and completed almost 70% of his passes the rest of the season. He's going to be a senior and he's got the best team around him that he's ever had. He's set up for a potential run at the SEC East, which I'm hoping will happen. It's time someone other than the big three won the East.

I hope I'm done having this argument. There is really no basis in fact for the argument that Mitchell isn't any good. He's not great, but he's not bad either. And he's more than capable of leading his team to success, unless you just don't believe facts are facts.

No comments: