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.
The graph shows the national average in grey, the average for NCAA Tournament teams in green, and UNC's average in blue.
In the last 23 seasons of college basketball games each team has taken an average 20.06 free throw attempts per game. That number has precipitously declined, however, in the last 10 seasons for some reason. Do any of you have any theories on how this happened?
As we can see in the graph, NCAA Tournament teams take, on average, a free throw attempt more than the national average. UNC, however, has generally been much higher than either average, especially in years with strong big man play.
Is this stat relevant? We can all remember games where it felt that free throw shooting won or lost a game for our team, but the correlation between free throw attempts and winning percentage is only 0.342, just over half of what FG% is. This statistic is mildly important, and far too much time and attention is paid to it.
FT Advantage + Win: 400
FT Advantage + Loss: 42
No FT Advantage + Win: 301
No FT Advantage + Loss: 277
Therefore UNC won 91% of the time that UNC made more free throws than opponents attempted. In the cases where UNC made fewer free throws than opponents attempted, UNC won 57% of the time. UNC's overall winning percentage in that period has been 73%.
From UNC's opponents perspective, the numbers are different. Of the 127 times where UNC's opponents made more free throws than UNC attempted, the opponents only won 69% of the time. In the cases where opponents made fewer free throws than UNC attempted, the opponent only won 21% of the time.
Several years ago I tested this differential against Margin of Victory and found it to not be very useful; mainly because of the enormous discrepancy between UNC's experiences and its opponents experiences. I supposed that if we looked at several teams, we'd see a wide variance in the level of advantage this differential offers.
If this were a useful statistic, opponents should have seen a much higher rate of success when meeting the criteria being tested. This is likely a theory that holds truer for teams that play a certain style more than others.
Over the past 23 seasons, free throw shooting has improved by about 2.5 percentage points. Tournament teams have followed suit, and remained a point or so higher than the national average. UNC has excelled especially during times of great guard play.
Keep in mind, though, that this graph's origin is 60%, so these differences are exaggerated on the graph. A percentage point improvement in FT% only results in about 0.2 points per game.
Free Throw Percentage has only a 0.254 correlation factor with winning percentage. So it is not a particularly useful tool for evaluating teams. Be careful not to overemphasize this stat when evaluating performance.
Next up: Points Per Possession as a Predictor
Free Throw Attempts
The graph shows the national average in grey, the average for NCAA Tournament teams in green, and UNC's average in blue.
In the last 23 seasons of college basketball games each team has taken an average 20.06 free throw attempts per game. That number has precipitously declined, however, in the last 10 seasons for some reason. Do any of you have any theories on how this happened?
As we can see in the graph, NCAA Tournament teams take, on average, a free throw attempt more than the national average. UNC, however, has generally been much higher than either average, especially in years with strong big man play.
Is this stat relevant? We can all remember games where it felt that free throw shooting won or lost a game for our team, but the correlation between free throw attempts and winning percentage is only 0.342, just over half of what FG% is. This statistic is mildly important, and far too much time and attention is paid to it.
Billy Packer's Free Throw Theory
While commenting on games Billy Packer repeatedly claimed that teams that make more free throws than their opponents attempt win the game. If we look at case of UNC's last 1,015 games, we see that the theory is about 91% right. When there is a difference in made free throws and the opponent's attempts, there are four possible outcomes. In the 970 games where there was a difference, here is how that scenario played out:FT Advantage + Win: 400
FT Advantage + Loss: 42
No FT Advantage + Win: 301
No FT Advantage + Loss: 277
Therefore UNC won 91% of the time that UNC made more free throws than opponents attempted. In the cases where UNC made fewer free throws than opponents attempted, UNC won 57% of the time. UNC's overall winning percentage in that period has been 73%.
From UNC's opponents perspective, the numbers are different. Of the 127 times where UNC's opponents made more free throws than UNC attempted, the opponents only won 69% of the time. In the cases where opponents made fewer free throws than UNC attempted, the opponent only won 21% of the time.
Several years ago I tested this differential against Margin of Victory and found it to not be very useful; mainly because of the enormous discrepancy between UNC's experiences and its opponents experiences. I supposed that if we looked at several teams, we'd see a wide variance in the level of advantage this differential offers.
If this were a useful statistic, opponents should have seen a much higher rate of success when meeting the criteria being tested. This is likely a theory that holds truer for teams that play a certain style more than others.
Free Throw Percentage
Over the past 23 seasons, free throw shooting has improved by about 2.5 percentage points. Tournament teams have followed suit, and remained a point or so higher than the national average. UNC has excelled especially during times of great guard play.
Keep in mind, though, that this graph's origin is 60%, so these differences are exaggerated on the graph. A percentage point improvement in FT% only results in about 0.2 points per game.
Free Throw Percentage has only a 0.254 correlation factor with winning percentage. So it is not a particularly useful tool for evaluating teams. Be careful not to overemphasize this stat when evaluating performance.
Next up: Points Per Possession as a Predictor
Last edited:
