Ignore historical team tourney performances - unless Tom Izzo is involved
“I can’t pick Virginia to make the Final Four - they’ll just lose early again!”
“UConn is my Sweet 16 sleeper, they always show up in March!”
“Villanova always chokes in the Tournament, I can’t have them go far in my bracket!”
You may have heard phrases like the ones above uttered or written on a comments section shortly after the bracket was announced for the 2016 NCAA Tournament this past Sunday. Once the initial anger over the selection committee has passed and acceptance of the bracket in its final form has set in, college basketball fans grab hold of whatever information we have to fill our brackets to the best of our abilities. Naturally, one of our first instincts is to look for trends, but some of these trends are based more out of a media narrative than cold, hard data.
At bracketvoodoo.com, we work hard to separate the signal from the noise in order to train our advanced bracket optimization engine. In the process we’ve taken a hard look at recent tourney performance and whether it impacts results. Digging into tournament results since 2000 among schools who participated in at least 5 NCAA Tournaments, the following schools lead the 2016 field in tournament wins over expected per year of participation, based on their seeding. Several of these teams are popular picks again this March.
Biggest Tournament Overachievers in 2016 field, since 2000
.tg {border-collapse:collapse;border-spacing:0;} .tg td{font-family:Arial, sans-serif;font-size:14px;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;} .tg th{font-family:Arial, sans-serif;font-size:14px;font-weight:normal;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;} .tg .tg-yw4l{vertical-align:top}
Team # Appearances Average Seed Average Expected Wins Average Actual Wins Wins over Expected per year UConn 11 3.72 1.9 2.91 1.01 Butler 10 8 0.84 1.8 0.96 Michigan St. 16 5.19 1.55 2.5 0.95 Michigan 5 5.6 1.37 2 0.63 West Virginia 8 6 1.16 1.75 0.59
And here are the underachievers, led by another controversial pick in this year’s tourney.
Biggest Tournament Underachievers in 2016 field, since 2000
.tg {border-collapse:collapse;border-spacing:0;} .tg td{font-family:Arial, sans-serif;font-size:14px;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;} .tg th{font-family:Arial, sans-serif;font-size:14px;font-weight:normal;padding:10px 5px;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;} .tg .tg-yw4l{vertical-align:top}
Team # Appearances Average Seed Average Expected Wins Average Actual Wins Wins under Expected per year Virginia 5 4.4 1.8 0.8 1 Vanderbilt 6 5 1.29 0.83 0.44 Cincinnati 11 5.63 1.34 0.91 0.44 Pittsburgh 12 4.25 1.73 1.33 0.41 St. Joseph’s 5 7.6 1.19 0.8 0.4
Clearly, UConn, Butler and Michigan St. have had great success in the NCAA Tournament since 2000 while Virginia, Vanderbilt and Cincinnati are more often disappointed. But does that mean you should trust historically strong teams more when filling out your bracket this year?
We expanded the analysis and pored through all NCAA Tournament outcomes over the past 20 years to find whether or not prior tournament performance has any impact on current year success. We looked at teams that participated in at least 3 tournaments in the 5 years prior to a given tournament, and considered their average wins above expected in those 5 years as an indicator for tournament success. As it turns out, there’s virtually no correlation between past tournament success and current year tournament success:
The reasons for this are pretty clear: First, there’s a ton of turnover in the college game between both players and coaches, so using what a school did in the NCAA Tournament 4 years ago to project what they’ll do today doesn’t hold much weight. Also, the single elimination format of the NCAA Tournament brings plenty of variation, so much of the craziness that happens can be chalked up to statistical chance. Finally, we’re still working with pretty small sample sizes here, so while a performance like VCU reaching the Final Four in 2011 may stick in the minds of fans, it’s hard to extract any meaningful trends from one or two nice tournament runs.
Bottom line: When you fill out your bracket this year, don’t pay too much attention to the media narratives about how teams performed in the past. The data shows there’s not much predictive value there. If anything, this underperformance will lead many of your friends to steer away from these teams in your pool, making them even stronger contrarian plays.
If you’ve read to this point you may be wondering about a certain team that always seems to outperform expectations in March. Well, even though we can ignore past tourney performance for most teams, there’s evidence that this just isn’t the case for Tom Izzo and Michigan State. Since Izzo took over as the MSU coach in 1995 he’s won a title, been to the Final Four seven times, and Sparty hasn’t missed out on the Madness since 1998. Put simply, they’ve been arguably the most consistently great team in the NCAA Tournament since Izzo took over:
As noted above, Michigan St. has also been among the most overachieving teams in this time frame. They’ve earned more than 16 additional wins over expected in the NCAA Tournament since their streak began in 1998, which is the most in college basketball. Even with the amount of noise in the NCAA Tournament, our models would give Izzo less than a 1% chance of this happening randomly, which strongly suggests that Izzo makes his teams legitimately better in March.
Among teams with at least 5 NCAA Tournament appearances in this time frame, only UConn and Butler come close. However, Michigan St. separates themselves from these teams for a few reasons. First, they’ve been more reliant on consistently strong tournament play as opposed to being lifted by a few impressive runs (UConn’s 2014 championship as a 7 seed rates as the most unlikely NCAA Tournament event since 1986, and Butler’s run to the finals as an 8 seed in 2011 is more statistically improbable than any single run Michigan St. has had). The Spartans have outperformed tournament expectations over 72% of the time since 1998, compared to 61.5% for UConn and 54.5% for Butler.
Most importantly, Michigan St. still has their coach. We might be able to coin this the “Stevens Exception” if the ex-Butler coach hadn’t left for the Boston Celtics, and it’s worth noting that the Bulldogs haven’t enjoyed the same level of tournament success since Brandon Miller and Chris Holtmann took over. And while Kevin Ollie has delivered a title to Storrs in his brief time as head coach, much of UConn’s success came with Jim Calhoun and there’s not a large enough sample to say that Ollie’s tournament overachievement can carry past 2014. Despite NBA rumors, the 61 year old Izzo has remained in East Lansing, a fact that alone makes Sparty a dangerous opponent in March.
As a 2 seed this year, Michigan St. would typically be expected to win between 2 and 3 games, meaning a Sweet 16 appearance is expected and advancing to the Elite 8 would be a slight overachievement. Our models give the Spartans a 37% chance to reach the Final Four and a 13.6% chance at winning the title, so we shouldn’t be surprised to see Tom Izzo continue his remarkable stretch in the NCAA Tournament. Combined with the fact that Michigan St. rates highly on our playmaking concentration metric with their tendency to rely on Denzel Valentine, the Spartans have a strong chance to advance deep in the tournament this year, even if they aren’t the best play in the midwest region due to the amount of people picking them. Keep in this in mind while putting the finishing touches on your bracket but remember that for the vast majority of teams, past tournament performance is a factor that can be ignored. Don’t be wary of picking Virginia or Villanova to make a run due to their recent failures - we give those teams a 29% and 22% chance of reaching the Final Four, respectively.
