Wednesday, October 30, 2013

Are Steve Alford And New Mexico Tournament Failures?

As we get closer to the first day of the regular season, college basketball social media is starting to kick into gear. And today, an interesting topic came up: the idea of a "regular season team" that fails in the postseason.

The context for this discussion is this article about New Mexico hoping that "a change in leadership yields a change in postseason success". I ended up debating the issue with Rob Dauster and Andy Glockner, and promised to write a blog post breaking down the math. So here we are.

Anyway, the crux of the issue is whether New Mexico has historically been a "postseason failure" and whether Steve Alford is an example of a "regular season coach" who has regular season success but can't find postseason success.

We know that Alford's Lobos had two pretty big Tournament failures in the last few years. And as Rob argued, the entire program has been to the Tournament 13 times and only won multiple games once, way back in 1974. In fact, because statistically analyzing the Tournament prior to the expansion to 64 teams is hard to do objectively, I'm only going to look at data from the 64/65/68 team era. And during that time, New Mexico has been to 11 NCAA Tournaments, never making a Sweet 16 and going 6-11 overall. That seems pretty damning.


Here's how I'm going to break this down. Total wins are a not-great way to define Tournament success. If you are a 14 seed every year, it's not a huge failure if you bow out in your opening game most years. The correct way to do this is PASE (Performance Against Seed Expectation). I'm going to use the data from here. So for example, the average 9 seed wins 0.59 games. If you are a 9 seed in ten consecutive years and go 6-10 then you will be exactly average, neither over- or under-performing what the average team has done with your seed.

So let's break down three different eras. New Mexico prior to Alford. New Mexico under Alford. And Alford at Division I schools other than New Mexico. What do we get?

New Mexico prior to Alford: 4-7 overall record. 6.75 projected wins according to PASE.
New Mexico under Alford: 2-4 overall record. 5.23 projected wins according to PASE.
Alford at other DI schools: 3-4 overall record. 3.77 projected wins according to PASE.

So overall, over a three decade stretch, New Mexico has won 6 games when they were supposed to win 12. Alford has won 5 games when he was supposed to win 9. So clearly, both have underperformed. But is that really significant?

I say no. And here's why. That sample size is just way too small. New Mexico prior to Alford won 2.75 too few games across two decades under three different coaches. Is that really supposed to be statistically meaningful?

And Alford before getting to New Mexico? He won 3 games when he was supposed to win 3.77. He lost famously on a crazy buzzerbeater in the opening round as a 3 seed, but he also took a 12 seed to the Sweet 16. The other two years he lost an opener as a 10 seed, and won a game as a 7 seed. Basically, he did fine.

What about this bad stretch for Alford at New Mexico? I'd argue that the PASE is deceptive here. As I've talked about many times, the Mountain West has become expert at "rigging" the RPI. It's very easy to inflate your RPI, and basically the whole conference does it. I spent all of last season getting excoriated in New Mexico fan forums and by New Mexico fans on twitter for pointing this out. But in the end they were only an 11 point favorite against Harvard in that 3/14 game. It was an upset, but nothing like a typical 3/14 upset. And in 2010? When a 3 seeded New Mexico lost to 11 seed Washington? That game was actually a pick'em in Vegas, and both Sagarin and Pomeroy had Washington the favorite. Why? Washington was a really strong 11 seed, and 3 seeded New Mexico was actually ranked 50th in KenPom. They were simply a fraudulent 3 seed.


So have Steve Alford and New Mexico, on average, underperformed in the NCAA Tournament? Yes. But not by much. And certainly not enough that statistical randomness and some bad seeding (such as due to Alford's Lobos coming up with a schedule that inflated their RPI) can't explain.

In general, the argument that "regular season teams" exist is without evidence. This post is in no way a rigorous statistical breakdown, but as far as I'm aware there has been no evidence that certain teams are more likely to struggle or succeed in the NCAA Tournament. The whole idea of ascribing an entire program over several decades as a "regular season team that fails in the postseason" seems to be proof of how irrational the concept is. Even if you believe that a certain style of basketball is more or less likely to succeed in March, what exactly is the style overlap between Dave Bliss, Steve Alford and Fran Fraschilla? Is there a magical potion in the Albuquerque water supply that causes basketball skills to deteriorate in March? Come on.

To me, this falls under the category of the "clutch" discussions I talk about all the time. Some athletes deal with clutch situations better than others, but the effect is a small one. And whenever anybody tries to statistically measure the impact of "clutch" play on a team's won/loss record, or future postseason success, they find none. The idea that some athletes or coaches can just "will" their teams to victory in clutch situations is nothing more than apophenia.

If the NCAA Tournament were played with a coin flip rather than on the basketball court, then by pure statistical randomness some teams would end up outperforming their seed more often while others would underperform. But this would mean nothing as far as projecting their future postseason success - it would still be a coin flip.

So if somebody has evidence that a certain style of basketball or coaching tends to outperform or underperform on average in the postseason, I'd love to see it. It's certainly believable and plausible. But without evidence it's nothing more than a guess. And if there's any effect at all, it's surely a small one.

No comments: