Seeds vs Placement: Analyzing 20 Years Of NCAA Results

Seeds vs Placement: Analyzing 20 Years Of NCAA Results

We analyze 20 years of NCAA Wrestling Championship Tournament results to see which teams did the best with regards to their seeds vs placement.

May 12, 2020 by Andrew Spey
null
For better or for worse, an NCAA wrestling team's performance is judged almost entirely by the Championship Tournament. As such, the success or failure of a season hinges on how the teams' NCAA qualifiers perform at the Big Dance in March. 

Unlock this article, live events, and more with a subscription!

Sign Up

Already a subscriber? Log In

For better or for worse, an NCAA wrestling team's performance is judged almost entirely by the Championship Tournament. As such, the success or failure of a season hinges on how the teams' NCAA qualifiers perform at the Big Dance in March. 

Because of the prominence of the NCAA tournament, a great amount of emphasis throughout the year is put on "peaking" for March. The rest of the season can all be reduced to one long qualification process, with the final crucible being the three days in Mid-March, where 10 champs and 80 All-Americans are determined. 

We need no reminder that the 2020 NCAA tournament was unfortunately canceled due to COVID-19. But we can always look back at the previous 21 tournaments, from 1999, when the current weight classes were put into place, to 2019. By focusing on the seeds and final placements, we can better understand which teams lived up to expectations, and which fell short. 

For our analysis, we will ignore bonus points and compare estimated tournament points scored based on seeds versus the estimated tournament points based on results. While using tournament points scored would be slightly more accurate (due to the difference in points awarded when advancing in either the championship or consolation side of the bracket), we do not have that data handy. The total discrepancies between actual and estimated points based on final placement would be so slight so as to not affect our confidence in the accuracy of our analysis. 

As an example, below is the 2019 Ohio State Buckeye lineup. All 10 starters qualified for the NCAAs, and in the table we compare how many points they would have scored based on their seed (i.e. if they finished in the same place as their seed) with the number of points (or very close to thereabouts) they actually scored (ignoring bonus points). 



With the 25th seed, Malik Heinselman was predicted to go 0-2 at the 2019 NCAAs. By winning a match, he scored half a team point more than expected. Myles Martin, conversely, was seeded first but finished third, and earned 6.5 team points less than expected. Adding up all 10 deltas between expected seed and estimated placement points and we get -1.5.

Though I'm sure the Buckeyes had a national championship as a goal and were thus disappointed by the final result, taken as a whole, everyone finished more or less where they were expected to, neither bust nor boffo. 

We can repeat the exercise for 2018, a tournament for which Ohio State also qualified all 10 starters.


With the net score coming in at -14.5, one can imagine the Buckeye faithful were a bit more disappointed in 2018 than 2019. And we didn't mean to single out a less than stellar tournament by Ohio State, they were just the only team that qualified all 10 for both 2018 and 2019, and we figured were best suited for demonstration purposes. 

In any event, we repeated that analytical process for every school from every NCAA tournament from 1999 to 2019. This net seasonal point totals will tell us of any team consistently over-performed over the years, and was thus better at peaking than their peers. 

Although our research in the past (article and follow-up) did not deliver any evidence that coaches have developed a particular skill for peaking their team, perhaps this time will be different. 

To test this peaking hypothesis, we picked two teams that qualified the most wrestlers over the last 21 years and plotted their net scores over time. Can you identify which schools we first analyzed from the graph below where we have removed the legend? 


Not much of a pattern. Here is the same graph with the legend included. 


Hands up to whoever had Penn State and Oklahoma State. The data appears very noisy, as if plotted at random, but Penn State does to a pretty solid job keeping their team in the net positives.

We can try the same exercise with the legend removed for two other schools. Any guesses?


That data once again appears as random as the previous graph. And now with the identifying legend. 


Curiously, the fortunes of the Big Ten rivals have been rising and falling in lockstep from 2015 to 2019, but headed in the exact opposite directions from 2008 to 2012. 

It's also worth pausing to note that the wrestling community uses the term "peaking" to mean two different things. There is the undeniably real concept of adjusting one's training schedule to optimize performance for a particular time of the year, or periodization. And then there is the concept that some teams are better are carrying out this strategy than others, and thus some coaches are said to be better at getting their respective teams to peak for March. The latter concept has proven far more elusive. 

Let's now look at the year-by-year net seed vs placement points for every team that averaged five qualifiers from 1999 to 2019. There were 26 such teams, and their stats are in the table below, listed in order of total number of qualifiers over the years. 


The cells shaded in a darker show over-performance, and the lighter shades show under-performance. We can tally up those scores across the era to get absolute numbers by team, which are below.

SCHOOL

DELTA

Penn State149
Lehigh114.5
Ohio State103.5
Edinboro97
Cornell74
Penn51.5
West Virginia48.5
Oklahoma42.5
Pittsburgh41.5
Illinois37.5
Purdue36.5
Northern Iowa34.5
Arizona State33
North Carolina31.5
Wisconsin27
Central Michigan21
Nebraska19.5
Hofstra16.5
Iowa State9
Oregon State9
Virginia Tech1
Minnesota-8.5
Michigan-29.5
Iowa-44
Missouri-65.5
Oklahoma State-77.5


Although one might be inclined to cite this table as evidence that Penn State has been the best at peaking and Oklahoma State the worst, but we don't think that would be accurate. The teams that qualify the most tend to have the worst net points, and the one team that did the best bucking that trend had a coaching change mid-way through the era. 

So we'd want to make adjustments for number of qualifiers (its easier to underperform than over-perform) and for coaching eras and probably lots of other stuff too. But it's a start, and fun to consider regardless of the findings (or lack thereof). 

What say you? Have we overlooked anything in the data? Any novel approaches to the question that we should consider? Send any thoughts on the subject our way, please!