clock menu more-arrow no yes mobile

Filed under:

NUMBERS DEEP DIVE: How SEC Tournament Runs Affect NCAA Tournament Performance

The numbers! What do they mean?!

NCAA Basketball: SEC Conference Tournament-Auburn vs Missouri Jim Brown-USA TODAY Sports

“I’d prefer my team rest up in the conference tournament. They’ve got a bid locked up in the Big Dance, there’s nothing to gain but everything to lose!”

“My team is on fire right now! They’re only an 8 seed overall, but they went on a run in the SEC tournament and now I think thy could win it all!”

What I’ve presented above is two very common arguments fans use when relating their team’s successes or failures in a conference tournament to the upcoming NCAA tournament. But what is the basis behind all of this? Is resting better if you can afford to do it? Does going on a hot streak mean you’ll continue that hot streak into the first few rounds of the NCAA tournament? Does it even really matter how a team performs in their conference tournament?

Last week, AUNerd proposed the question as something worth taking a deep dive into, and I figured it was about time to earn my “amateur statistician” epithet. I decided to gather all teams that played in both the SECT and NCAAT since the 2012-13 season, which is when the modern SECT was born. To keep the math simple, I counted the number of wins each team had in each tournament.

“But Ryan, that totally ignores which teams were actually just good and won games because of skill, not ‘rest’ or ‘hot streaks’. You dummy!”

Hey! I’m not done explaining yet!

“On with it, nerd!”

So first I’m a dummy, now I’m a nerd? Whatever...

Anyways, once I had the number of wins for each team, I was able to use the expected tournament win totals for each NCAAT seed (thanks to Bracket Odds) to set an expected performance baseline. In lieu of rigorous odds for the SECT, I calculated them experimentally just based off of win totals from the last 7 SECT. The results have fairly large standard deviations, but that’s the best I could do.

Now, with all of those pieces, I was able determine Percentile Performances for each team in the list. So for example, Auburn in the 2018 postseason was a 1 seed in the SECT and a 4 seed in the NCAAT. With 0 wins as a 1 seed in the SEC, that was good for a 1.5% percentile performance. Not good Tigers. Not good. With one win in the NCAAT as a 4 seed, that earned the team a 33.2% percentile performance. A little more in line with reality, but still not good. If Auburn had been a 7 seed in the SECT and a 12 seed in the NCAAT, the same performance (0 SECT wins and 1 NCAAT) would have yielded a much higher percentile performance. Keep in mind 50% would be the average performance.

Ryan Sterritt

As you can see, there’s little to no correlation between the SECT and NCAAT percentile performance. For those unfamilar, an R^2 value is generally explained as the percentage of the resulting data that can be explained by the experimentation at hand (as opposed to randomness). With an R^2 of 0.0029 or 0.29%, it’s pretty fair to say that SECT perfomance as calculated has little to no impact on a teams NCAAT performance.

Sure, you could argue with a South Carolina in 2017, who as a four seed in the SECT, went winless before advancing all the way to the Final Four in the NCAAT. Maybe their rest really did help them prepare for the tournament. But looking at the larger picture, you can see it’s just dumb luck rather than any sort of “resting” plan.

So what’s the moral of the story? Hot streaks mean nothing, and extra rest because you got upset by a 10 seed in your conference tournament means nothing. Case closed.