Regular vs. Post-Season QB WPA and EPA

As follow up to the recent post about QB 'clutch' performance in the playoffs, here is a comparison of QB performance in the regular season compared to the post-season. I've created a viz to see how each QB season stacks up. Each data point is a QB-year. The x-axis represents each QB's WPA per play in a regular season and the y-axis represents his WPA/P in the corresponding post-season. The trend line shows that there is, as expected, a systematic connection. Winning QBs in the regular season tend to continue to be winning QBs in the post-season. The residual, or vertical distance above the trend line, represents how much each QB over or under-performed in the post-season. I limited the cases to QBs who had at least 25 attempts in a post-season.

For example, Kerry Collins' 2002 post-season stands out in that he dramatically out-played his regular-season self. (It was only one game, but Collins was absolutely heroic in what was one of the most heart-breaking miscues in playoff history. With only a minute left Collins drove the Giants to the SF 23. A botched snap cost New York the game.)

On the other side of things was Tony Romo's 2006 loss to Seattle. Again, a bobbled snap cost the (anti-) hero the game. (Special teams plays are usually excluded in these kinds of analyses, but the play was technically a Romo run.)

Note that Joe Flacco's current playoff run ranks as the eighth biggest over-performing post-season (so far).
Hover over each point to reveal the QB-year. You might want to make use of the zoom controls.

Clicking on the WPA Rank tab shows a bar graph list of each QB's over/under-performance for each post-season. There is a similar analysis for EPA. One thing you'll notice is that the connection between regular and post-season EPA production is stronger. That's expected due to WPA's hypersensitivity to game situation.