In case you haven't checked out the blog in a couple of days, here's a quick recap. On Friday, I explained the methodology I used to rank every regular season for every QB ever, and how I compiled his career rating. On Monday, I argued for Joey Harrington as the worst QB in NFL history, at least in a relative sense of receiving the most playing time while producing the least impressive stats. And then yesterday, I ranked the top 75 QBs in NFL history, leaving Dan Marino as the greatest regular season QB ever. Today, notwithstanding Jim Mora's wrath, I want to talk about playoffs. I'm not kidding you. Playoffs.
Whenever I've done a study analyzing NFL QBs, a frequent complaint is that I ignore post-season data. Part of the reason is simplicity -- I've got a long file of individual seasons, but a less complete and separate file of post-season seasons. It's easy to compare everyone across the board based on regular season numbers, while incorporating post-season data is tricky. Let me go through a few problems.
1) What weight do I give to post-season data? Do I weigh each playoff game equally, or give more weight to the later round games? What weights should be assigned? What standard should I compare the players against -- the league average for the post-season that year, or the league average for the regalar season that year? Should the QB's rating in that year's regular season matter?
2) Sample size issues. You might get one game against a terrific defense, or a game in a blizzard. Over the course of 10 regular seasons, I assume that those things even out. To the extent that they don't, I believe people subjective adjust the numbers, anyway (i.e., move Fouts and Anderson down because of their innovative offensive systems, move Elway up because of his weak supporting cast, and maybe move someone like Favre up because he plays in Green Bay).
3) I only have post-season data for QBs from 1967-2006, and I don't have sacks data for any of those games. I think that's a pretty significant problem, but it's not fatal. Pretty soon I'll get the 2007 data into my database, but I didn't want to let that delay me from posting.
4) Mostly, I thought that because the system was complicated enough, and because there would be some very real gripes with the method no matter what I chose, I took the hassle minimization approach. To be frank, I also feel that for the most part, post-season performances are overrated. Guys get way too much credit for post-season wins and losses, in my opinion, and it really overshadows the talent or ability of the QB.
But I thought it over some more. I don't have to focus on wins or losses at all. I could look simply at the QB's stats, and nothing more. For someone like Montana, the post-season makes up a very substantial part of his career -- he has 732 post-season attempts and 5,391 regular season attempts. So I'm effectively excluding over 10% of his career by ignoring the playoffs. Of course, you could do that and he still comes out as a top five QB.
That said, I'm going to go forward with some post-season analysis in the same way I did the best and worst QB analysis -- comparing the QB to the league average QB from the regular season that year. I'm going to assume, which is quite wrong, that no QB was ever sacked in the post-season from 1967 to 2006, and that no post-season games before 1967 were ever played. It's unfortunate, and often incomplete data is worse than no data, but I'm going to cave to demand here.
So... which QB had the best post-season of all time? No matter how you slice it, it was Joe Montana in 1989. But the guy who ranks #2 on the list might surprise you.
Read the rest of this entry »