Sunday, March 26, 2017

Elite 8 Day 2 Picking The Lines

Anytime Bill Self loses in the NCAA Tournament, the Bill Self #HotTakes fire out from the people in the media that you expect:
I've debunked these sorts of narratives before, and the way to analyze NCAA Tournament performance is to use Performance Against Seed Expectation. How has Self done? As a Kansas coach, he's been expected to win 37.4 games and has won 33. If we throw in his other coaching stops he's been expected to win 45.3 games and has won 43. In other words, he's won ever so slightly fewer games than expected against some extraordinarily good seeds. In 14 seasons he's had a 1 seed seven times and has never been below a 4 seed. That is remarkable.

So don't fall for these sorts of dumb narratives about coaches. The reason Self's tournament losses are always upsets is because he's always favored. The coach who has the most losses in upsets, or who loses the most in late NCAA Tournament rounds, is by definition a fantastic coach. You've got to get to those games to lose them.

Anyway, let's get to today's games:

Yesterday ATS: 2-0-0
2017 Tournament ATS: 31-28-3 (53%)
2016 Tournament ATS: 36-30-1 (55%)
2010-15 ATS: 220-167-11 (57%)

Florida (-3.5) over South Carolina: Rather than breaking down match-ups, you really just have to ask yourself if you believe in South Carolina. Because South Carolina has played amazingly well for three games, but it kind of came out of nowhere:

Both Kansas and Xavier had been playing out of their minds for three games, and both regressed in their Elite 8 game. That doesn't mean South Carolina necessarily will regress too, but it's a reminder that "momentum" is not real, and a three game explosion is more than likely an anomaly. South Carolina's defense is excellent, of course, but the difference between a defense that gave up 0.93 PPP in SEC play and one that gave up 0.94 PPP in SEC play is not significant. Florida is the better team.

Kentucky (+2.5) over North Carolina: With two teams that play very high tempos and that score efficiently, this could be a very high scoring game unless John Calipari successfully slows the game down the way that he did against UCLA. I'm going with Kentucky here because I'm not exactly sure why North Carolina is supposed to be the better team. Statistically the two teams were almost exactly even this season. Also, as good as North Carolina's offense has been, a lot of their success has come from out-athleting teams - they led the nation in offensive rebounding percentage but were just 10th in the ACC in eFG% and 12th in FT%. If Kentucky can keep North Carolina contained on the glass (the Wildcats led the SEC in defensive rebounding percentage) then I like their chances to win.outright.


Anonymous said...

You say the proper way to evaluate Self is by Performance Above Seed Expection. Well according to T-Rank, he ranks 296th out of 302 coaches in PASE since 2002. That's not exactly stellar performance...


Jeff said...

That's a wildly deceptive factoid. PASE is a counting stat, so of course the extreme ends are going to be coaches who have coached a lot of games. Most of those 302 coaches have only participated in a handful of games, and many have coached just once. And of the coaches even lower ranked than Bill Self includes Coach K.

Anonymous said...

He's still fifth worst in PASE among coaches with at least 5 tournament appearances. The fact is his teams have underachieved compared to how you would expect them to perform on average. At some point it stops being just 'bad luck' and it's actually because you're being out-coached because you don't know how to adapt your style of play to maximize your chances of winning.

Jeff said...

Again... you are still using a counting metric. Cutting it off at 5 games isn't any less dumb than cutting it off at 1 game. Look at how the coaches at the extreme ends are all the super successful coaches who have been in the NCAA Tournament many times.

If we are to look at this from a Bayesian approach, the null hypothesis is that all coaches perform exactly well in the postseason as they do in the regular season. I am unaware of any scientific or statistical evidence of any kind that credibly challenges that hypothesis. Are you?

Anonymous said...

There are far more of the 'super successful' coaches at the positive end of the PASE spectrum than there are at the negative end. It's not equally distributed as you seem to be suggesting, which would support the idea that it's just random chance that's producing extremely high or low PASE values for successful coaches.

Successful coaches at the plus end of PASE (+3 or better): Roy Williams, Jim Calhoun, John Calipari, Billy Donovan, Jim Larranaga, Jim Boeheim, Rick Pitino, Brad Stevens, Tom Izzo, John Beilein, Bo Ryan, Sean Miller

Successful coaches at the negative end of PASE (-3 or worse): Coach K, Bill Self, Tony Bennett, Huggins, Dixon.

That's 12 coaches at the plus end and 5 on the negative end. Hardly equally distributed.

Your picks are also dog shit. Keep flipping those coins you developmentally disabled fuck stick. I pray to god no one is tailing those picks because they'll be filing for bankruptcy soon.

Jeff said...

You seem delightful.