I acquired all of the Division I team data since 2002, and from that we can observe some fascinating trends and data relationships in the data. This is a multipart series exploring some of that data.
As you may recall, Dean Smith defined the end of a possession as happening with a turnover, a trip to the free throw line, or the attempt of any field goal. The latter is in stark contrast to the "modern" method, but the result gives us some interesting information. At the end of a game, in Smith's eyes, one team likely had far more possessions than the other. That difference reveals the true rebounding advantage, and it tells the rebounding story more reliably than the raw rebounding stat in the box score.
The graph shows the Possession Differential (POSDIFF), by year, for the last 23 seasons, through the morning of February 24, 2024. The grey line shows the POSDIFF for all of Division I, while the green line shows the POSDIFF for teams that made the NCAA tournament that year. Over time the figure has held very flat, at just over 0.00 per team. This makes sense given that games were predominantly against each other.
As with many other stats, the performance difference between tournament teams and the national average has remained steady, about double the national average.
Ignoring the appearance of Batman in the graph, we see that UNC, shown in blue, has had significantly higher POSDIFF than the tournament teams in most years. This makes sense given the heavy emphasis that Roy Williams placed on Rebounding.
How important is this statistic? Despite Roy Williams' affinity for rebounding, its correlation factor with Winning Percentage is only 0.202, and its correlation with making the tournament is a miniscule 0.086.
Looking at this differently, things get more interesting, though. Raw Total Rebounds actually correlates with Winning Percentage in an average way, with a score of 0.473. Raw Opponents' total rebounds, however, has a 0.064 correlation score. The fact that Opponent's Rebounding correlation score is positive, and so much smaller than the team's rebounding score, is something that needs more explanation and research.
Despite the low correlation score, adding POSDIFF to a regression of P/P and p/p does increase the reliability. So, while its contribution to the regression concept is miniscule, POSDIFF isn't a bad statistic, it is probably more if a distraction than it is a key performance indicator.
Next up: Percentage of Missed Shots Rebounded
As you may recall, Dean Smith defined the end of a possession as happening with a turnover, a trip to the free throw line, or the attempt of any field goal. The latter is in stark contrast to the "modern" method, but the result gives us some interesting information. At the end of a game, in Smith's eyes, one team likely had far more possessions than the other. That difference reveals the true rebounding advantage, and it tells the rebounding story more reliably than the raw rebounding stat in the box score.
The graph shows the Possession Differential (POSDIFF), by year, for the last 23 seasons, through the morning of February 24, 2024. The grey line shows the POSDIFF for all of Division I, while the green line shows the POSDIFF for teams that made the NCAA tournament that year. Over time the figure has held very flat, at just over 0.00 per team. This makes sense given that games were predominantly against each other.
As with many other stats, the performance difference between tournament teams and the national average has remained steady, about double the national average.
Ignoring the appearance of Batman in the graph, we see that UNC, shown in blue, has had significantly higher POSDIFF than the tournament teams in most years. This makes sense given the heavy emphasis that Roy Williams placed on Rebounding.
How important is this statistic? Despite Roy Williams' affinity for rebounding, its correlation factor with Winning Percentage is only 0.202, and its correlation with making the tournament is a miniscule 0.086.
Looking at this differently, things get more interesting, though. Raw Total Rebounds actually correlates with Winning Percentage in an average way, with a score of 0.473. Raw Opponents' total rebounds, however, has a 0.064 correlation score. The fact that Opponent's Rebounding correlation score is positive, and so much smaller than the team's rebounding score, is something that needs more explanation and research.
Despite the low correlation score, adding POSDIFF to a regression of P/P and p/p does increase the reliability. So, while its contribution to the regression concept is miniscule, POSDIFF isn't a bad statistic, it is probably more if a distraction than it is a key performance indicator.
Next up: Percentage of Missed Shots Rebounded
Last edited:
