Conference Champions in March Madness
Do conference champions do better or worse than at-large bids in March Madness? The tricky thing about this question is that you could use your sports-intuition to justify either conclusion.
- Teams that won their conference tournament are hot and are playing well at the right time of the year. Therefore, we should expect them to over perform in March Madness.
- Teams that won their conference tournament are over-seed because they won a lot to end the year (recency bias) and so should under-perform relative to their seed. Even worse, they are likely to be beat up because of all the extra games they played.
Which of these is right? Which intuition about conference champions in March Madness is the correct intuition?
The answer, it turns out, is neither. Conference champions neither overperform nor do they underperform relative to their seed expectation. On average, going back over 40 years, we’ve found that conference champions have won roughly exactly as often as the teams with the same seed that weren’t conference champions.
This means that any March Madness strategy which relies on picking or avoiding conference champions is a strategy that is not backed by data. Feel free to use these techniques, but know that the numbers don’t support your claims.
The remainder of this article describes our methodology and how we came to the conclusions that conference champions perform no differently than at-large bids in March Madness.
To receive email updates when new articles are posted, use the subscription form below!
Problem Introduction
March madness consists of two types of teams: the automatic bids and the at-large bids. The automatic bids are teams that automatically qualify to make the tournament. They are guaranteed a spot. They earn this right by winning the conference they played in. Automatic bids make up about half the teams in the tournament.
The other teams in the tournament are the at-large bids. These teams earn their tournament spot based on their season-long resume. The at-large bids go to the other 36 teams that the selection committee thinks are the best teams that didn’t earn an automatic bid.
As described in the introduction, common wisdom is split on whether conference champions (the automatic bids) do better or worse in March Madness than the at-large bids. Some people think the automatic bids do better in March Madness. Some people think the at-large bids do worse.
The only way to settle this debate is by actually looking at the data. For all of the following analysis, we look at data back to the 1980 college basketball season.
The Naive Solution
The simplest way to look at this problem is to simply look at the win-loss record of conference champions in March. Since the 1980 season, conference champions are 915-1157. This means that conference champions only win about 44% of their tournament games. Does this mean that conference champions do worse in March Madness?
No.
This stat is extremely misleading. Most of the conference champions are 13, 14, 15, and 16 seeds. They are, by and large, the worst teams in the tournament. The point of our study is to understand whether conference champions over or underperform in March Madness. 13, 14, 15, and 16 seeds are expected to lose. Therefore, this really bad win-loss record is not indicative of underperformance but rather is a reflection of the fact that most conference champions are double digit seeds.
A better calculation is to look at how many games conference champions win relative to how many they’re supposed to win based on their seed. For example:
- Even if a 16 seed goes 0-1 and exits in the first round, they didn’t really underperform because that was the expected outcome.
- Even if a 1 seed goes 2-1 and makes the Sweet 16, they had a bad tournament relative to expectations
Therefore, to determine whether conference champions over or underperform in March Madness, we need to compare their records against their expectations for their seed.
March Madness Record by Seed
The first step in our proper analysis was to go through every tournament back to the 1980 season and compute the average wins by seed. The table below contains the expected number of wins for each seed in March Madness:
Seed | Expected Wins |
|---|---|
1 | 3.1 |
2 | 2.2 |
3 | 1.8 |
4 | 1.5 |
5 | 1.2 |
6 | 1.1 |
7 | 0.9 |
8 | 0.8 |
9 | 0.6 |
10 | 0.6 |
11 | 0.6 |
12 | 0.5 |
13 | 0.2 |
14 | 0.1 |
15 | 0.09 |
16 | 0.01 |
Let’s look at two examples to help interpret this data. A #1 seed has an expectation of about 3.1 wins in March Madness. That means they should on average win in the round of 64, round of 32, and sweet 16. An average performance for a one seed is an elite 8 appearance. If a 1 seed makes the final four, they over-perform. If they don’t make the elite 8, they underperform.
A 12 seed, on the other hand, has an expectation of 0.5 wins. This means that losing in the first round is “as expected” for a 12 seed. If they win one game, they over perform. If they win two, they significantly over perform.
We’re going to look at all the conference champions and see how many games they won in March Madness compared to the table above.
Performance of Conference Champions Relative to Seed Expectation
Since the 1980 season, 1174 conference champions have played in March Madness. These teams have played in 2072 games, with a record of 915 wins and 1157 losses. However, the true measure of how well conference champions do in March is comparing how many wins they recorded versus how many wins they were expected to have.
The figure below shows the distribution of wins above replacement for all conference champions since the 1980 March Madness. Positive values mean a team won more games than they were supposed to. Negative values mean a team under-performed.
The most common outcome is teams slightly underperforming relative to their expected number of wins. This is the large peak just to the left of 0. This is the most common outcome because of all the conference champions which get a 13, 14, 15, or 16 seed. These teams lose more than 75% of the time on average. However, their expected number of wins is in the 0-0.5 range. Therefore, whenever one of these teams lose (which is often), their wins relative to expectation is between -0.5 and 0.
The distribution above is interesting to look at, but the true metric which helps us determine whether conference champions over or underperform in March Madness is the average of this distribution. Since the 1980 season, conference champions won 0.02 games more than average after controlling for their seed.
This number is so close to 0 that this is not a statistically significant effect and we can comfortably conclude that conference champions do not win more or less often than at-large bids in March Madness.
