Friday, December 20, 2013

#MATH THO


I received an email this afternoon from a hero who would prefer to remain anonymous, alerting me to this gem. It's Insider (which is hilariously absurd, as you'll see), but the premise is going through each team in the playoff picture, and assessing their likelihood of making the playoffs, as well as which seed they should be expected to receive.

The article is written by Danny Tuccitto of Football Outsiders. He begins with the Broncos:

On the heels of a massive upset loss -- they were an 86 percent favorite -- Denver's probability of earning the No. 1 seed dropped from 76 percent to 63 percent. [...] But don't fret, Broncos fans. With remaining games coming against two of the worst three teams in the NFL (Oakland and Houston, according to weighted DVOA), our projection model says there's only a 15 percent chance Denver loses either game.

Well, this is worrisome. At 11-3, the Broncos are tied for the best record in the conference with the Chiefs. Denver owns the tiebreaker over Kansas City, having beaten them twice. So, stands to reason that if DEN wins out, they'll be the top seed.

With that in mind, let's revisit two things:

Denver's probability of earning the No. 1 seed dropped from 76 percent to 63 percent.

[O]ur projection model says there's only a 15 percent chance Denver loses either game.

Yeah, that's not gonna work. Even before considering the possibility that other teams may lose (more on that in a minute), Denver is 85% to get the one seed according to their W16 and W17 win probabilities. That same 63% number is here; maybe they're just splitting ties in half rather than, you know, considering that one team wins the tiebreaker 100% of the time?

To boot, the Patriots have a 74 percent chance of sweeping their last two games, which is only slightly lower than the Bengals' 81 percent.

New England is currently a two-point underdog in Baltimore on Sunday afternoon, so it seems like 74% to win out may be a bit high? FO's numbers famously don't "adjust for the future impact of recent injuries" so it seems possible that could be playing a role here? Not that there's any mention of that in the article, or that you could ever get to 74% by doing anything reasonable anyway.

According to our projection model, Indianapolis is a 19-to-1 underdog at Kansas City this weekend, while Cincinnati is a 19-to-1 favorite at home against Minnesota. Put those two odds together, and there's a 90 percent likelihood that the Colts will have nothing to play for in Week 17.

The best part of this is how casual he is about these insane numbers. The Colts are 6.5-point underdogs in Kansas City, and +233 at Pinnacle. But yeah, 19-1! Moving right along!

There is no point in any type of "analysis" like this if you're going to use 95% for the Chiefs' win % this week. If you actually think that, just bet as much as possible on Kansas City, and get very rich, very, very quickly.

According to our model, the Ravens are underdogs in both of their remaining matchups, with this week's game against New England presenting slightly less of a threat (42 percent win probability) than their Week 17 game at Cincinnati (31 percent).

The Ravens are 42% to win this week? Against the Patriots, who are 74% to win out? What? Where am I?

After Week 13, the Chiefs' likelihood of earning home-field advantage throughout the playoffs was a mere 2 percent; it's now 24 percent. Although that means Kansas City's still a 3-to-1 underdog, it nevertheless makes them Denver's No. 1 contender and twice as likely as New England to overtake the Broncos for the top spot. Both Kansas City and Denver are better than 90 percent favorites this weekend, so the battle for home field will likely come down to a pair of West Coast games in Week 17. The Chiefs will need to win as a slight favorite in San Diego (53 percent) and hope the Broncos lose as a huge favorite in Oakland (93 percent).

So, the Chiefs' HFA shot likely comes down to W17, where they need to win (53%) and have DEN lose (7%). Seems rather unlikely? Nope! There's a 24% chance that Kansas City gets the top seed.

I'm not even going to bother with the NFC. It's absolutely remarkable that the writer didn't pick up on any of this while writing and/or proofreading, and the fact that this then passed through an editor (assumption?) and was published on ESPN Insider (just $8.95 per month!) is mind blowing.

Keep building that brand though, guys. Build the shit out of it.

No comments:

Post a Comment