Posted by Chase Stuart on February 7, 2007
I've often wondered how much worse quarterbacks play in the playoffs than in the regular season. With the 2006 season in the books, I now have data on the last forty years of NFL post-season play. It's safe to assume that in the aggregate, the statistics of players at any position decreases in the playoffs. This is true for two reasons, one obvious and one hidden. First, we know that the average playoff opponent has a better team, a better offense and a better defense than the average regular season opponent. Against better opponents, we'd expect poorer statistics.
The other idea is that teams that make the playoffs may not be as good as we think. Over at our other site, I do a lot of writing on rearview strength of schedule analysis, which adjusts regular season performance for strength of schedule. One thing I've noticed is that many times the quarterbacks that post the best regular season numbers are the ones that play the weakest schedules. I'd suspect there's at least a decent correlation between great quarterback statistics and qualifying for the post-season, so we might see QBs that aren't as good as they seem go up against much harder defenses in the playoffs. As a result, we can certainly expect to see QB Ratings and Adjusted Yards/Attempt dip in the post-season.
Let's start by looking at the wrong way to do it. Since 1967, all NFL QBs have completed 13,162 of 23,653 passes (55.65%) for 162,201 yards (6.86 Y/A) with 975 TD/ 1004 INT (5.36 AY/A, 73.3 QB Rating). For starters, we don't have too much to compare that to, although I think most of us know that those numbers are below the league averages these days.
What we really want to know is what happens to specific QBs in the playoffs. To do that, we need to compare every QB's regular season performance with their post-season performance. Let's get some administrative stuff out of the way early. Jarious Jackson, Jason Garrett, Bobby Hoying, Bill Musgrave, Dave Brown, Andre Ware and Rich Gannon threw post-season passes in years without throwing a regular season pass. Since there was no regular season to compare their post-season numbers to, I threw those statistics out. Additionally, three QBs -- Pete Beathard, Bernie Kosar and Steve Walsh -- played for two teams in the regular season in years they made the playoffs. I included their regular season numbers only from the team they made the playoffs with.
That leaves 909 games played by QBs in the post-season during the last 40 years. To determine how we might expect them to play, we need to get a weighted average of their number of post-season attempts by their regular season QB ratings. For example, take the small hypothetical list:
QB RegQBR Postseason Atts
QB1 100.0 100
QB2 90.0 20
QB3 70.0 70
In that hypothetical three quarterback world, we'd expect their aggregate QB rating in the playoffs (assuming that the playoffs were no different than the regular season) to be 87.89. That's because [ (100.0 * 100) + (90.0 * 20) + (70.0 * 70) ] / (100 + 70 + 20) = 87.89.
Now we can apply that logic to the actual results, only with several hundred QBs and several thousand attempts. Over the course of the 23,000+ post-season attempts, thrown by many great and some terrible quarterbacks, we'd have projected a QB rating of 83.6 and an AY/A of 6.30.
In reality, the aggregate post-season QB rating is 73.2 and the AY/A is 5.37 (the numbers are slightly different than before because of the QBs removed from the data). So we can say with a reasonable degree of confidence that "playing in the post-season" reduces a player's QB rating by about 12-13% and adjusted yards per attempt by 14-15%. I'm not sure if this reinforces what the average fan believes, but this seems pretty reasonable to me. I'm not sure how much the decline should be divided up among tougher weather conditions, teams with easy schedules making the playoffs, overall stiffer competition in the playoffs, and other distinctions.
Removing all QBs with fewer than 100 regular season attempts doesn't change the sample size much. Their actual aggregate performance compiled a 73.7 QB Rating and 5.41 AY/A, while we'd have projected an 84.5 QBR and 6.37 AY/A. Once again, a noticeable decrease in QB efficiency. But why not break things down even more.
Expected QBR and AY/A statistics by round
w 83.1 6.20
d 83.5 6.28
c 85.9 6.55
s 89.3 6.84
W, D, C and S stand for the Wildcard, Divisional, Conference Championship and Super Bowl rounds. As you can tell, the numbers increase slightly as we move to the later rounds, which is about what we'd expect. Generally the best regular season QBs move on to the later rounds, so we'd expect better QB Ratings and AY/A there. But of course, this assumes that defenses (and everything else) remains constant in the post-season, which we know it does not. So what actually happened in the four playoff rounds?
w 79.1 5.78
d 71.4 5.27
c 70.0 5.08
s 76.2 5.68
There have been a few standout performances in the Super Bowl, but in general we've seen actual performance go down as we move further into the playoffs. The conference championship round in particular has been rough for QBs, which may reflect that many of those are played in cold weather stadiums, whereas Super Bowl XLI was the first one with suboptimal conditions.
But by far the most significant result comes when looking at win/loss splits. Sure, we should expect winning QBs to have high QB ratings and losing QBs to have low QB ratings, but the split is enormous. Based on the specific QBs that won a playoff game, we'd expect an 86.6 QBR and 6.58 AY/A. They actually had a 92.9 QBR and averaged 7.17 AY/A.
Losing QBs would be expected to have, based on their regular season numbers, an 82.8 QBR and averaged 6.21 AY/A. In reality? A 58.4 QBR and an ugly 4.01 AY/A.
It's unclear which is the chicken (losing) and which is the egg (low QB Rating) but there's certainly a very high correlation.
After chewing on that for a minute, the natural follow up question popped into my head: what's the splits among winners and losers in the regular season? From 2001-2005, winning QBs in the regular season had a 94.1 QBR and averaged 7.04 AY/A. Losing QBs had a 68.2 QBR and averaged 4.68 AY/A. We're not drawing from the same population (regular season data is from five recent seasons, playoff data spans four decades), but the results are pretty similar. I'm not too sure what to make of it, but it wasn't what I was expecting. For what it's worth, in the post-season from 2001-2005 winning QBs scored a 100.2 QBR and averaged 7.56 AY/A, versus losing QBs having a 66.0 QBR and a 4.56 AY/A) average.