RazorStar 4,025 Posted March 5, 2017 Well I finally completed this long study on how certain QB’s fare when their defenses play to expectations or below expectations. Before I get into the meat of my findings I should first explain what exactly I was doing, and my methodology to achieve those findings. So the main goal was determining how good a QB was compared to the average player at his position. In order to go about that I needed to determine a relevant sample size and cut-off point for my data. 2004 may seem like an arbitrary year to start as my data collection point, but I chose it for two reasons. The first reason being the introduction of the Ty Law rule, which is technically just an extension of the Mel Blount rule, but called with greater severity thanks to his efforts. This greatly expanded the passing game in the NFL and created a new era of offense. The second reason is that 2004 was also the introduction of perhaps the greatest draft class for QB’s in league history. Ben Roehlisberger, Philip Rivers, Matt Schaub and Eli Manning were all drafted and all had impressive careers so far. Because this set a new era of passing, it only made sense for this to be the cutoff point. The second hurdle was determining how many QB’s should be on the study. This selection was a lot more arbitrary, I wanted players with 100 starts originally, but lowered it down to 90 once I realized that you can’t divide seasons into that evenly. Dropping the cutoff down to 90 starts let me put in some players who had retired a while back like Favre and McNabb, while also giving me relatively young guys like Cam Newton and Andy Dalton to put in the sample. With a healthy mix of 22 quarterbacks, I was ready to start. The last hiccup I had to deal with was determining which games should count as QB wins. In most cases it’s pretty easy because only one QB gets all the snaps. In some cases though, injury and resting starters can skew the results. It looks worse on the stat sheet when you’re credited for four extra losses because you started the first drive and then sat out the rest of the game to rest for the playoffs. So I had to get all the QB’s records from Pro-Football-Reference and go through the gamelogs to determine what games should and shouldn’t count as wins and losses for that QB. After a lot of picking and pecking, I finally had records for all 22 QB’s, including postseason performances. Once I had that, it was simply a matter of sorting each QB’s games into two categories. The first is when their defense gives up 23 or more points, the second is when they give up 22 or less. The reason I chose that as my threshold is because the average offense over the 13 years of the study score just a hair over 22 points a game on average. Therefore if your defense was allowing 22 or less points, a simply average performance should result in a win, while if the defense gave up more, an average QB performance would be a loss. Obviously that reasoning is flawed since there are games where both teams score above 23 points and games where both teams score below that, which is why I derived the second expected win percentage column for hat you'd reasonably expect from an average QB. With all that out of the way, here are the results, sorted by the QB’s winning percentage, the relative strengths of their defense, and finally the theoretical Wins Above Replacement they provide in a 16 game season. QB Win Percentage: Best Defenses: Wins above Replacement in a 16 Game Season. Google Doc: https://docs.google.com/spreadsheets/d/1wezMtoCffXvcLPmlYck1Flu4Si7-GRY1e9TQu2KQSwA/edit?usp=sharing I’ll share some of my conclusions and anecdotes on the data later, but for right now I just wanted to get this up and posted. The games I added and removed along with how I figured out what the average QB win percentage looks like are in the second tab. 4 Share this post Link to post Share on other sites
Vin+ 3,121 Posted March 5, 2017 Colt's defense is higher than I thought it would be. Share this post Link to post Share on other sites
RazorStar 4,025 Posted March 5, 2017 Peyton did get a few years with the Broncos defense as well, but I was pleasantly surprised with how well his defenses did. More importantly, Peyton was the best among QB's in games where his team gave up 22 points or less, going a staggering 106-6. You needed to put large scores on the board to stand a chance, there weren't going to be many defensive struggles. As for QB's on the more frustrating end of the spectrum, you've got Matt Stafford and Donovan McNabb, who do pretty well in shootout style games, yet have games where they just don't show up and can't back up good defensive performances. Ryan Fitzpatrick is just plain bad and the numbers confirm it, which I was really happy to see. Share this post Link to post Share on other sites
FartWaffles 1,857 Posted March 6, 2017 Colts defenses were probably ranked higher than their actual talent level, because a lot of the time Peyton would get the team up early so quickly, that opposing offenses would then become one-dimensional. (Tampa 2 defense takes advantage of that) 1 Share this post Link to post Share on other sites
RazorStar 4,025 Posted March 10, 2018 I've gone ahead and updated this for the new year. There was only one new QB to qualify for the sample data, and it was Russell Wilson, who has spotted himself right in behind Tony Romo and Ben Roethlisberger. Pretty good place to be early in your career. Cam Newton and Joe Flacco of all people ended up jumping quite a few spots up, but the Matt Schaub Eliteness Line (MSEL) is still holding strong. Russell Wilson has the best defense of any QB in this study so far, so it'll be interesting to see if he can keep up when his defense isn't consistently holding offenses to less than 23 points per game. So far every QB in the study save for Ryan Fitzpatrick is appreciably better than the average starter, but Alex Smith is actually inching closer to average after this season, which is peculiar to say the least. I still have to derive the new expected win percentages, but they shouldn't change all that much. https://docs.google.com/spreadsheets/d/1wezMtoCffXvcLPmlYck1Flu4Si7-GRY1e9TQu2KQSwA/edit?usp=sharing 1 Share this post Link to post Share on other sites