Here are the historic percentages of each seed to win at each round:
The above are percentages of the TOTAL not the winning percentage for those seeds that got that far, ie:
- 15.97% of #1 seeds won the NC (23 out of 144).
- 3.47% of #2 seeds won the NC (5 out of 144).
- etc
Below are the winning percentages by seed in each round:
#1 seeds win at least three out of five games in EACH round. They are above 80% in each of the first three rounds and right around 60% in the E8, F4, and NC.
There are two big distinctions that I want to identify, explore, and explain:
First, #1 seeds are a breed apart:
They are MUCH better than everybody else. I think this makes sense intuitively. IMHO there are usually between about two and six teams in the country that are just a LOT better than everybody else. These teams make up generally the #1 seeds. Sometimes there aren't enough of them and you get one or two weaker (relatively) #1 seeds or one or two VERY GOOD #2 seeds but, in general, the #1 seeds are just REALLY good. Even with the #2 seeds there are chinks in the armor. They are good teams, I'm not saying that they aren't, but they have identifiable weaknesses.
#1 seeds have the best winning percentage of any seed in each of the first three rounds. Thus, an individual #1 seed has a nearly 70% chance to make the E8. This is nearly double the chance that an individual team of any other seed will make the E8.
The E8 is the first round in which the #1 seeds do not have the best winning percentage but even there they are fourth and there are two things to note about the seeds that do better in the E8:
First, the seeds that do better than #1 in the E8 are:
- #5 seeds win 77.8% of their E8 games (7-2)
- #4 seeds win 61.9% of their E8 games (13-8)
- #8 seeds win 62.5% of their E8 games (5-3)
Does anything stand out to you about those particular seeds?
Well, none of them have to play a #1 in the E8 because #8 seeds already did in the second round and the #5 and #4 seeds already either beat a #1 in the S16 or beat an 8/9 that beat a #1.
Secondly there is a lack of data issue. #1 seeds have won 60% of their E8 games (60-40). There have been 100 instances of a #1 playing an E8 game. The three seeds that are "better" in the E8 have only played a combined total of 38.
Moving on to the other big gap:
The 5/12 upset is well known to anyone with even a passing interest in CBB but here is the thing: 5/12 upsets aren't more likely than 6/11 upsets, they are slightly less likely than 6/11 upsets but 5/12 upsets are MUCH more likely than 4/13 upsets. Why?
Well, there are two theoretical possibilities, either:
- #4 seeds are substantially better than #5 seeds, or
- #12 seeds are substantially better than #13 seeds.
Which is it? It is the latter, #12 seeds are substantially better than #13 seeds, but why? Well, I'm going to tell you but first lets disprove the idea that #4 seeds are substantially better than #5 seeds:
- In the second round the gap between #4 and #5 tightens while the gap between #12 and #13 widens.
- In the S16 both the #4's and the #5's have very poor winning percentages (because they are almost always playing #1) but the gap isn't that big, the #13's have NEVER won a S16 game while the #12's have won two.
- In the E8 the #5's actually do BETTER than the #4's and they also do better in the semi-final before crashing they've never won a NC.
So then, why are #12's so much better than #13's? I'm glad you asked, here is the key reason:
The above shows all 32 leagues that will get an auto-bid. The top-15 (down through Am East) each have at least one team in the top-68 which I'm using as a rough proxy for "tournament quality" because, of course, 68 teams make the tournament. After that the next few leagues listed have at least one team that is at least borderline "tournament quality". The top teams in the SoCon, Summit League, MAAC, and MAC are in the 70's in the NET rankings. Then there is a fairly large drop to the MWC whose best team (NMST) is #86. Then there is a humongous drop to the Ivy League whose best team (Princeton) is #109).
The bottom 12 leagues (starting with the Ivy) do not have any teams even remotely good enough to be involved in the National Championship tournament. Their Champions are purely "tallest midgets" and nearly all of them will get unceremoniously dismissed in the first round. A lucky few will pull off first round upsets and once in a while one will win a second round game to make the S16 but in the 36 Tournaments since expansion to 64 teams (1985-2021 not including 2020) no #13 or lower has EVER won a second weekend NCAA Tournament game.
Think about that for a minute. There have been 576 #13-#16 seeded teams and those 576 teams have won a grand combined total of ZERO second weekend NCAA games. These teams simply don't belong.
There are actually two sources of REALLY bad NCAA Tournament teams:
- As demonstrated above, there are about 12 leagues where it simply doesn't matter who wins the league, none of the teams are any good, and
- There are, of course, league tournament upsets. Nebraska (#166) or Georgia (#216) could get hot and win the B1G or SEC. This, however, isn't likely because either of them would have to beat a slew of actual tournament quality teams to pull that off. In the middling leagues, however, you have situations like the Am East where the best team is a respectable Vermont team (#63) but the second best team is UMBC (#235). Thus, if Vermont gets upset in the league tournament then the Am East will send a REALLY bad team to the NCAA. Similarly in the Summit League the best team is a decent SoDakSt (#71) but the second best is Oral Roberts (#146).
Note that this does not vary all that much from year to year. Here is the data for this year and the last two tournaments:
The bottom ten leagues didn't have a team in the top-68 in 2022, 2021, or 2019. Then, depending on the year, there were another six or seven leagues that didn't have a top-68 team in that particular year.
Based on that and the upsets, there are around 12-18 teams in the tournament every year that are just flat awful. These teams make up roughly the #13-#16 seeds.
