Saturday, November 8

Stat Boy Saturday: Looking at Playoff Schedules—Could It Be Bad?

Another week with our honorary stat boy, Zach Fein of Fein Sports. A contributor with as much stat muscle as anyone in the business. You may not understand what he's says and use it for your fantasy leagues, but if you're in the mood to be baffled and confused, then boy does he have you covered. Because life is one big spreadsheet. This week he gets the dirt on fantasy playoff schedules. So your league's trading deadline is coming up soon. If you're near the top of the standings, it would make sense to trade for a high upside player with an easy schedule in the fantasy football playoffs. Owners scrambling to make the playoffs would want to trade an injured player for one that would help them now. The Football Outsiders guys conveniently looked into this on Thursday, by spotting players with the easiest playoff schedules based on their now-to-the-end-of-season projections. Seems nice enough, right? That depends on if a team's early-season performance actual correlates with their fantasy playoff performance. It would make sense that if a team is bad from Weeks 1 through 12, they must be bad from Weeks 13 through 17, and vice versa for good teams. In the words of Lee Corso, not so fast, my friend. Using all teams since 2002, I checked the correlation of their regular season performance and their fantasy playoff performance. The results will shock you.

Correlation of Weeks 1-12 and 13-17
Stat r r2
Pts_all 0.215 0.046
PassYd_all 0.129 0.017
RushYd_all 0.381 0.145
Takeaways 0.111 0.012
Let me help you understand the numbers. The higher the r and r-squared (r2) are, the more related the stat is from the regular season to the playoffs. The r-squared effectively says how many of the teams' playoff performance were predicted based on their regular season performance. An r-squared of .7 or higher is considered good, anything under .5 is considered to be a bad set of data, and anything under .2 is considered to have no correlation whatsoever. In other words, only 4.6 percent of all 192 teams in the sample had a points allowed per game in the fantasy playoffs that could be predicted by their regular season points allowed per game. The numbers show that a team's points allowed, passing yards allowed, and turnovers are essentially a tossup in the fantasy playoffs. Rushing yards allowed does much better than those three, but it still is basically a coin flip as to whether a team's regular season play will predict their playoff performance. The next three tables show the correlations of teams ranked in the top 60 of all 192 teams in the sample (represents the top 10 in any given year), teams ranked in the bottom 60 (bottom 10), and all teams in the middle (teams ranked from 11-21) in each stat based on regular season play. I excluded takeaways because there were many ties and the sample size was too small (there's only 15 or so by a team in a year, so of course it would have a low correlation).
Top 10 Teams
Stat r r2
Pts_all 0.231 0.053
PassYd_all 0.088 0.008
RushYd_all 0.133 0.018
Teams ranked 11-21
Stat r r2
Pts_all 0.257 0.066
PassYd_all 0.032 0.001
RushYd_all 0.027 0.001
Bottom 10 Teams
Stat r r2
Pts_all 0.107 0.012
PassYd_all -0.048 0.002
RushYd_all 0.019 0.000
Again, no correlation whatsoever. In fact, you get better r's and r's-squared if you use every team than by splitting them up into tiers. By the way, there's a -0.048 r for passing yards allowed by the bottom 10 teams. That means as passing yards allowed in the regular season go up, it goes down in the playoffs. I find that hilarious, even if it's just -0.048. The main point of all of this is don't make trades or assess players based on their fantasy playoff schedule. The stats show that you can't predict at all a team's defensive play based on their early-season performance. Sorry, Trent Edwards, Marshawn Lynch, and Lee Evans owners. It doesn't even have to be a question. Just send us an e-mail.

No comments: