Phil
Stats Geek
- Joined
- Aug 25, 2011
- Messages
- 4,784
- Reaction Score
- 7,192
Start with the obvious caveats — not a single one of the games in the opening two rounds involving the number one seeds was expected to be a nailbiter and that's what happened. The expected margin of victory in these eight games was about 35 points, and the actual margin was closer to 45. It is very difficult teasing conclusions from lopsided games.
That said, what else am I going to do? Yes, it's possible that a 4 seed or maybe a 10 seed will be a challenge for a final four slot but history suggest that the ones most most likely to challenge for the final four spots of the top four seeds, so I will start by looking at how those games compared to expectations.
In the following table I have the first eight games. In the expected box, I list the expected score as estimated by Nancy which implies an expected margin of victory (EMOV). (For example UConn was expected to be UTSA 87 – 38 for margin of victory of 49)
In the actual box, I list the actual score and the calculated actual margin of victory (AMOV)
The difference between those two values is the Delta. If games exactly match expectations the Delta will be zero. If the team listed exceeds expectations the Delta will be positive, and if they fail to meet expectations the Delta will be negative.
I've also listed the bench minutes for each game which will become relevant later.
The first observation is that seven of the eight games have a positive Delta with only one negative value. While emphasizing the folly of teasing too much from a small sample size, this is not entirely unexpected. Massey is calculating the strength of the team over the season and estimating what will happen when those two teams meet up. As far as I know he does not assume that NCAA tournament games will turn out differently than regular season games. We know that in a game between a strong team and an overmatched opponent, the stronger team is likely to take a foot off the gas near the end of the game, and it is possible that this phenomenon will be less likely to happen in the winter go home atmosphere of the NCAA tournament, which would suggest that positive values are not surprising.
Another question is that while UConn has the lowest entry in the eight games, they also have the largest performance. That latter point is somewhat diminished by the observation that while 23 is larger than 21, it's unlikely to be statistically significant.
Overall, South Carolina comes away with the largest Delta for the first two games, followed by Texas. The lowest value is UCLA.
UConn fans will take part and the largest single value in a game that everybody knows was not as close as the final score suggests.
Finally I cooperated bench minutes as a crude estimate of the extent to which teams take a foot off the gas when the outcome is clear.
UConn has the highest bench minutes but Texas has virtually the same value. The main take away is that UCLA's Delta was the smallest of the four, and is not because they use a lot of the bench – the batch less than any of the other three.
Google sheet link
2026 NCAA first 2 rounds
That said, what else am I going to do? Yes, it's possible that a 4 seed or maybe a 10 seed will be a challenge for a final four slot but history suggest that the ones most most likely to challenge for the final four spots of the top four seeds, so I will start by looking at how those games compared to expectations.
In the following table I have the first eight games. In the expected box, I list the expected score as estimated by Nancy which implies an expected margin of victory (EMOV). (For example UConn was expected to be UTSA 87 – 38 for margin of victory of 49)
In the actual box, I list the actual score and the calculated actual margin of victory (AMOV)
The difference between those two values is the Delta. If games exactly match expectations the Delta will be zero. If the team listed exceeds expectations the Delta will be positive, and if they fail to meet expectations the Delta will be negative.
I've also listed the bench minutes for each game which will become relevant later.
The first observation is that seven of the eight games have a positive Delta with only one negative value. While emphasizing the folly of teasing too much from a small sample size, this is not entirely unexpected. Massey is calculating the strength of the team over the season and estimating what will happen when those two teams meet up. As far as I know he does not assume that NCAA tournament games will turn out differently than regular season games. We know that in a game between a strong team and an overmatched opponent, the stronger team is likely to take a foot off the gas near the end of the game, and it is possible that this phenomenon will be less likely to happen in the winter go home atmosphere of the NCAA tournament, which would suggest that positive values are not surprising.
Another question is that while UConn has the lowest entry in the eight games, they also have the largest performance. That latter point is somewhat diminished by the observation that while 23 is larger than 21, it's unlikely to be statistically significant.
Overall, South Carolina comes away with the largest Delta for the first two games, followed by Texas. The lowest value is UCLA.
UConn fans will take part and the largest single value in a game that everybody knows was not as close as the final score suggests.
Finally I cooperated bench minutes as a crude estimate of the extent to which teams take a foot off the gas when the outcome is clear.
UConn has the highest bench minutes but Texas has virtually the same value. The main take away is that UCLA's Delta was the smallest of the four, and is not because they use a lot of the bench – the batch less than any of the other three.
Google sheet link
2026 NCAA first 2 rounds
Last edited: