Advanced Stats Primer

[Kedsy]

As many of you know, I've been giving an advanced stats breakdown after each Duke game this year. Billy Dat asked me to explain where I draw the "good/bad" line, and why. So in my usual habit of overkill, I've decided to post a thread explaining each of the advanced stats I use, what they are, how they're calculated, where I draw the good/bad line and why, and how Duke has done historically in each stat.

I plan to make a post for each stat, talk about that stat from both the offensive and defensive viewpoint (for most of them, anyway), and look at trends when possible. The Duke tables will come from a spreadsheet I've privately maintained, and the contextual (national) stat tables come from sports-reference.com (https://www.sports-reference.com/cbb/schools/). Any reference to 2019 Duke's performance were tabulated after the first 10 games of the 2018-19 season.

I will be focusing on advanced stats for a college basketball team. Advanced stats for individual players follow similar formulas, but are a bit more complicated to calculate (and are a bit less reliable due to the differing number of minutes a player is on the court, the unknown difference in what happens in a game from minute to minute, and the various complexities posed by who else is on the court with the player being evaluated by the stat).

Before we start, let's talk about what comprises an "advanced stat," and what makes it different from a counting stat. First, how many of a particular stat a team (or player) accumulates is dependent on pace -- meaning if a team has the ball 80 times in a game, it's easier to accumulate counting stats than if the team has the ball only 60 times in a game. Second, many stats depend on the number of opportunities, for example if a team only misses 10 shots in a game there are fewer opportunities to get rebounds than if they miss 40 shots in the game. Finally, though only applicable to individual player stats, how many stats you can accumulate is dependent to how many minutes you are on the court. The point of advanced is to normalize all those things. So these stats should be pace-independent, opportunity-independent, and minute-independent.

The stats I plan to explain can all be gleaned from a standard box score. There are also even more advanced stats, but they're more or less impossible to calculate from a box score so they're not widely available or used publicly. For our part, we can only look at what we have.

Finally, with one exception the stats I plan to present are all "raw stats," in other words they are not adjusted for competition. For example, if a really fast team plays a really slow team they'll probably have fewer possessions than they normally would; if a great offensive rebounding team plays a great defensive rebounding team, they'll probably have fewer o-boards than usual. But to adjust for all that would take a great deal more energy than I'm willing to expend, so all my numbers are unadjusted. If you want adjusted numbers, you can probably find a lot of them behind Pomeroy's paywall (though I don't subscribe, so I can't honestly say I know what he makes available).

Oh, one more thing. Lots of you guys know a lot more about statistics than I do. If you already know this stuff and I'm oversimplifying in some areas then (a) sorry, I'm doing my best; and (b) feel free to chime in and correct and/or clarify and/or amplify on any statement or stat.

Anyway, here goes...

[Kedsy]

Obviously, if you want pace-independent statistics, you need to be able to measure pace, which is generally measured by possessions per game. A possession ends one of three ways: (1) you shoot the ball; (2) you turn the ball over; or (3) you get fouled and take free throws. If you miss a shot (or free throw) and get the offensive rebound, it's not really a new possession, but rather a continuation of the same possession, even if it results in another shot, turnover, etc.

If every free throw was successful (and there were no and-ones) then the number of possessions represented by free throws would be (0.5)*FTA. Since neither of those assumptions is true, we have to come up with a multiplier other than (0.5). Most accurate would be to count these things individually from the play-by-play, but nobody on Earth is willing to do that so they've come up with an estimate based on statistical analysis of a sample of games. The commonly accepted college ball estimate is (0.475). Interestingly, the commonly accepted NBA estimate is (0.44); I'm not sure why, maybe there are more and-ones in the NBA?

In any event, the formula for possessions per game is: (FGA + (0.475*FTA) + TOs - ORs)/gms. The higher this number is, the faster you play.

Here's a table with Duke's possessions per game in every year since 1987 (which is when they first started tracking offensive rebounds, making this calculation possible). As you can see, for Duke, fast is generally good, as six of our fastest 7 teams made the Final Four. But slow isn't necessarily bad, as two of our five slowest teams won the national championship (though it's also worth noting that the other three of our slowest five teams got knocked out in the first round). If this year's team keeps up its currrent pace, we'll have the 5th-fastest team in "modern" Duke history.



Year Pace NCAA
1990 79.34 2
1989 78.91 4
1991 78.53 1
2002 77.19 16
2019 76.74
2001 76.65 1
1999 75.39 2
1988 75.02 4
1993 75.01 32
2000 74.82 16
2008 73.98 32
1992 73.67 1
1998 73.44 8
2003 72.96 16
1987 72.71 16
2006 71.55 16
2018 70.65 8
1994 70.28 2
2011 70.25 16
1995 70.18 n/a
2005 70.13 16
2004 69.95 4
1997 69.94 32
2009 69.61 16
2017 69.59 32
1996 68.98 64
2016 68.29 16
2013 68.25 8
2012 68.16 64
2010 66.55 1
2007 66.38 64
2015 65.78 1
2014 65.76 64


Here's a table showing context for the last five NCAA seasons. The reason there's a big jump in pace between 2015 and 2016 is that's when the shot clock was lowered from 35 seconds to 30 seconds. Apparently that change was worth about 4 possessions per game nationally. This year, it seems NCAA teams are running faster than in the past, but that may simply be a function of all games being November/December non-conference games.



Year 1 10 50 100 150 200 250 300 Last
2019 88.0 79.5 74.8 73.1 71.6 70.5 69.2 68.1 61.6
2018 83.6 75.6 72.4 71.0 70.0 69.2 68.0 66.9 60.6
2017 83.3 75.5 72.6 71.1 70.1 69.3 68.2 66.7 60.3
2016 81.2 74.8 72.2 70.7 69.7 68.7 67.7 66.4 61.5
2015 77.5 70.6 68.1 66.6 65.5 64.6 63.8 62.5 57.1

[Kedsy]

When we talk about offensive rating (oRating), we're really talking about points scored per possession (ppp). The formula is points scored/possessions. Obviously, more is good here. Duke generally ranks pretty high in this metric. I think good (for Duke) is 1.20 and above, and bad (for Duke) is anything below 1.10.



Year ppp NCAA
1999 1.218 2
2019 1.211
2015 1.197 1
2014 1.188 64
1992 1.186 1
2018 1.186 8
2001 1.180 1
2016 1.180 16
1998 1.166 8
2017 1.162 32
2010 1.157 1
2000 1.155 16
2011 1.151 16
2002 1.148 16
1993 1.143 32
2004 1.134 4
2013 1.131 8
1997 1.131 32
2006 1.130 16
2012 1.126 64
2008 1.121 32
1988 1.113 4
2005 1.111 16
1991 1.110 1
1990 1.109 2
2009 1.106 16
1994 1.104 2
2003 1.103 16
1989 1.096 4
1995 1.090 n/a
1996 1.084 64
2007 1.049 64
1987 1.046 16


The above table represents "raw" oRating, meaning it is not adjusted for competition. KenPom adjusts for competition; the table below shows Duke's adjusted performance since 2002 (the first year published by Pomeroy). Note that these adjusted numbers are post-tournament numbers, and such numbers are significantly affected by NCAA tourney performance.



Year oRtg NCAA
2014 1.247 64
2015 1.245 1
2018 1.224 8
2016 1.219 16
2017 1.210 32
2010 1.210 1
2002 1.210 16
2004 1.208 4
2019 1.203
2011 1.197 16
2006 1.194 16
2013 1.191 8
2012 1.177 64
2009 1.175 16
2008 1.163 32
2005 1.157 16
2003 1.150 16
2007 1.132 64


The tables below are contextual tables showing the last five years of NCAA team oRating, first unadjusted then adjusted.

"Raw"


Year 1 10 50 100 150 200 250 300 Last
2019 1.275 1.189 1.125 1.093 1.064 1.037 1.003 0.967 0.802
2018 1.222 1.156 1.109 1.082 1.053 1.039 1.016 0.988 0.863
2017 1.210 1.160 1.109 1.079 1.054 1.029 1.006 0.979 0.882
2016 1.192 1.154 1.107 1.076 1.057 1.038 1.011 0.976 0.827
2015 1.211 1.162 1.092 1.059 1.037 1.016 0.995 0.961 0.792


"Adjusted"


Year 1 10 50 100 150 200 250 300 Last
2019 1.218 1.160 1.098 1.069 1.041 1.018 0.994 0.961 0.862
2018 1.278 1.194 1.133 1.089 1.060 1.033 1.005 0.980 0.865
2017 1.260 1.198 1.122 1.085 1.055 1.030 1.004 0.971 0.857
2016 1.244 1.200 1.127 1.088 1.064 1.032 1.004 0.979 0.837
2015 1.290 1.194 1.119 1.078 1.055 1.030 1.003 0.969 0.789

[Billy Dat]

Thanks for the effort, Kedsy! I plan to refer to this, and the coming posts, often.

[Kedsy]

When we talk about defensive rating (dRating), we're really talking about points allowed per possession. The formula is points allowed/possessions. Obviously, less is good here (good (for Duke) is probably around 0.94 and down; bad (for Duke) is probably 0.96 and up). This year's team is crushing this stat.

Looking at the table, our top teams in defensive ppp had a lot of tournament success (6 of our top 10 teams made the Final Four, with an Elite Eight team as well). And our six teams with dRatings equal to or above 1.0 were generally tourney disasters. But oddly, our six teams worse than 0.95 but better than 1.0 did very well in the tourney, I suppose meaning that sometimes it's OK to give up points. All that said, a 1.0 dRating is still generally a top 100 defense.



Year ppp NCAA
2019 0.856
1998 0.873 8
1989 0.884 4
1999 0.891 2
2002 0.893 16
1987 0.912 16
1988 0.914 4
2010 0.917 1
2001 0.917 1
2011 0.921 16
2004 0.923 4
2007 0.924 64
2005 0.927 16
1991 0.929 1
2008 0.935 32
2000 0.936 16
1997 0.940 32
1993 0.942 32
2009 0.944 16
2003 0.947 16
2006 0.949 16
1990 0.953 2
2013 0.956 8
1994 0.958 2
2015 0.969 1
2018 0.979 8
1992 0.979 1
2012 1.000 64
1996 1.003 64
2017 1.009 32
2014 1.021 64
1995 1.033 n/a
2016 1.054 16


The above table represents "raw" dRating, meaning it is not adjusted for competition. As with oRating, KenPom adjusts dRating for competition; the table below shows Duke's adjusted performance since 2002 (the first year published by Pomeroy). Note again that these adjusted numbers are post-tournament numbers, and such numbers are significantly affected by NCAA tourney performance.



Year dRtg NCAA
2002 0.868 16
2010 0.877 1
2005 0.883 16
2004 0.885 4
2019 0.897
2008 0.899 32
2007 0.899 64
2011 0.913 16
2003 0.913 16
2006 0.918 16
2015 0.920 1
2013 0.921 8
2009 0.932 16
2018 0.936 8
2017 0.968 32
2012 0.980 64
2016 1.000 16
2014 1.004 64


The tables below are contextual tables showing the last five years of NCAA team dRating, first unadjusted then adjusted.

"Raw"


Year 1 10 50 100 150 200 250 300 Last
2019 0.769 0.850 0.915 0.954 0.980 1.007 1.036 1.075 1.198
2018 0.869 0.940 0.980 1.003 1.020 1.036 1.056 1.088 1.187
2017 0.867 0.930 0.971 0.999 1.015 1.036 1.059 1.081 1.174
2016 0.887 0.937 0.976 0.998 1.021 1.040 1.059 1.081 1.148
2015 0.844 0.914 0.953 0.985 1.005 1.023 1.043 1.067 1.153


"Adjusted"


Year 1 10 50 100 150 200 250 300 Last
2019 0.880 0.917 0.963 0.997 1.025 1.045 1.067 1.092 1.155
2018 0.856 0.938 0.975 1.012 1.036 1.067 1.086 1.112 1.250
2017 0.863 0.922 0.970 1.012 1.034 1.060 1.087 1.110 1.231
2016 0.887 0.929 0.978 1.008 1.037 1.064 1.087 1.116 1.193
2015 0.844 0.919 0.967 1.005 1.032 1.060 1.082 1.108 1.218

[Kedsy]

Effective field goal percentage (eFG%) is a way to take into account that 3-point shots are made at a lower rate than 2-point shots, but are worth a lot more. The formula can be expressed in several ways, including (FGM + (0.5*3pm))/FGA.

Generally I'd say "good" is above 55% and "bad" is below 50%. Five of the six Duke teams with eFG% above 56% have made the Final Four, including three champions. But our worst team ever in this stat (or at least since the 3-point shot was introduced in 1987) also won the national championship, so I guess there's lots of ways to succeed.



Year eFG% NCAA
1992 57.68% 1
1999 57.41% 2
1989 57.04% 4
2015 56.62% 1
2002 56.29% 16
2001 56.15% 1
2019 56.03%
2018 55.92% 8
2006 55.49% 16
1993 55.44% 32
2017 54.78% 32
1991 54.77% 1
2000 54.65% 16
2013 53.86% 8
2014 53.81% 64
2011 53.74% 16
2016 53.70% 16
2008 53.65% 32
1995 53.57% n/a
1998 53.45% 8
1997 53.35% 32
2004 53.19% 4
1994 53.05% 2
1988 52.87% 4
2012 52.79% 64
2007 52.51% 64
1990 52.17% 2
1987 52.14% 16
2005 52.11% 16
2003 51.44% 16
1996 50.67% 64
2009 50.53% 16
2010 50.51% 1


Note even though it's the worst Duke performance ever, the 2010 team's eFG% was 101st in the country. Below is the contextual table for the past five years of eFG% nationally.



Year 1 10 50 100 150 200 250 300 Last
2019 61.5 58.1 55.4 54.0 52.5 51.2 49.3 47.3 37.3
2018 60.5 57.4 54.4 52.9 51.8 50.7 49.6 48.3 42.6
2017 59.8 56.9 53.9 52.5 51.3 50.3 49.0 47.9 40.6
2016 58.7 56.1 53.0 51.8 50.7 49.7 48.4 46.9 41.7
2015 58.4 55.9 52.7 50.7 49.7 48.8 47.7 46.1 39.8

[Kedsy]

Opposing eFG% measures how well your opponent shoots against you. I'd say 46% or lower is good and anything close to 50% or above is bad. Eight of the top 10 Duke performances in this stat have at least made the Elite Eight, but our worst performance ever (1992) won the national championship.



Year eFG% NCAA
2005 42.22% 16
2019 42.33%
1999 42.98% 2
2010 43.69% 1
2011 44.48% 16
2004 44.78% 4
1994 44.90% 2
1998 44.96% 8
1989 45.21% 4
2013 45.49% 8
2001 45.79% 1
2007 46.05% 64
2000 46.11% 16
1997 46.15% 32
2002 46.17% 16
2006 46.21% 16
2018 46.38% 8
2015 46.54% 1
2012 47.12% 64
1993 47.16% 32
2017 47.45% 32
2008 47.58% 32
1987 47.64% 16
1990 47.71% 2
2009 47.77% 16
1991 47.83% 1
1988 48.04% 4
1996 48.23% 64
1995 48.56% n/a
2003 48.76% 16
2014 49.30% 64
2016 49.51% 16
1992 50.52% 1


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 38.7 41.7 45.4 47.4 48.8 50.2 51.6 53.4 59.6
2018 42.7 46.0 47.9 49.2 50.1 50.9 52.0 53.5 57.9
2017 41.1 45.1 47.5 48.6 49.7 50.7 51.9 52.9 57.7
2016 42.6 44.6 46.7 48.1 49.1 50.1 51.0 52.1 58.3
2015 39.6 43.7 45.6 47.1 48.3 49.4 50.5 51.9 55.9

[Kedsy]

Free throw rate is how often you go to the line. The formula is FTA/FGA. Our top four teams in this stat all went to the NCAA championship game, but there doesn't seem to be a strong correlation overall. Which is good because the 2019 team is our 4th-lowest in FT rate since 1987.

Still, it's probably better to go to the line than not. I'd rate 40% and up good and below 30% bad.



Year FTRate NCAA
1992 50.41% 1
1990 48.58% 2
1999 46.86% 2
1991 45.36% 1
2003 45.28% 16
2012 45.12% 64
2006 45.00% 16
1989 44.00% 4
1993 42.97% 32
2005 42.87% 16
1998 42.14% 8
2009 41.81% 16
1997 41.33% 32
2004 41.18% 4
2002 41.01% 16
2008 40.88% 32
1994 40.81% 2
2016 40.54% 16
1988 40.11% 4
2015 39.84% 1
2007 39.79% 64
2001 39.60% 1
2017 39.27% 32
2013 39.15% 8
2014 38.76% 64
2000 38.35% 16
2010 37.85% 1
1987 37.78% 16
2011 37.44% 16
2019 35.32%
2018 34.57% 8
1996 33.92% 64
1995 33.05% n/a


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 55.8 48.9 41.5 37.8 35.1 33.3 30.9 27.7 19.4
2018 46.3 42.5 38.7 36.2 34.3 32.5 31.1 29.0 22.5
2017 46.3 43.6 40.5 38.1 36.5 34.5 33.1 30.2 24.2
2016 49.5 44.9 41.9 39.5 37.5 36.1 34.0 31.7 25.9
2015 52.5 46.9 42.1 39.8 37.8 36.1 34.4 32.3 27.0

[Kedsy]

Opposing free throw rate is how often you put your opponent on the line. Again, there doesn't seem to be a strong correlation either way to NCAA success, possibly because there's a relationship between committing fouls and playing tough defense.

I'd say good is 25% and down and bad is 35% and up.



Year FTRate NCAA
2006 21.26% 16
2018 22.25% 8
2015 24.02% 1
2000 24.16% 16
2019 24.69%
2016 25.31% 16
1993 25.45% 32
1992 26.34% 1
1994 26.45% 2
2001 28.37% 1
1999 29.39% 2
2011 29.61% 16
2007 30.08% 64
2009 31.02% 16
1995 31.03% n/a
2017 31.24% 32
2008 31.90% 32
2004 31.94% 4
2005 32.05% 16
1990 32.13% 2
1991 32.41% 1
2002 32.64% 16
2013 32.75% 8
2012 32.92% 64
2010 34.03% 1
1998 34.83% 8
1997 36.08% 32
1996 36.28% 64
1988 37.14% 4
2003 37.64% 16
1989 38.16% 4
1987 38.99% 16
2014 40.75% 64




Year 1 10 50 100 150 200 250 300 Last
2019 19.2 21.8 26.6 29.2 31.5 33.7 35.9 40.1 56.5
2018 22.1 24.3 27.8 30.1 31.8 33.7 36.3 39.1 48.9
2017 22.1 25.4 29.2 31.9 33.6 35.7 38.5 41.0 56.4
2016 22.7 25.3 30.1 33.3 35.6 37.5 39.4 42.5 55.3
2015 22.4 26.1 31.0 33.2 35.1 37.6 40.2 43.0 55.5

[cato]

If there were an advanced stat for posting useful stuff, Kedsy would be killing it.

[Kedsy]

Offensive rebounding percentage (OR%) is the percentage you grab of offensive rebounding opportunites. The formula is OR/(OR + opp DR). For Duke teams, this stat seems to closely correlate with NCAA success. Possibly because Duke is usually so good on offense that giving them multiple shots in a possession is hard to counter. This year's team is doing extremely well in this category, and currently ranks 5th in the nation.

I'd say 38% and up is good and below 35% is bad.



Year OR% NCAA
1999 44.34% 2
2019 40.92%
1990 40.89% 2
1988 40.53% 4
2010 40.33% 1
1998 39.67% 8
1992 39.54% 1
2004 39.20% 4
2018 38.56% 8
1996 38.27% 64
1991 37.99% 1
1994 37.96% 2
1987 37.55% 16
2000 37.39% 16
2009 37.38% 16
2005 37.22% 16
2003 37.06% 16
2001 37.03% 1
1995 36.94% n/a
1989 36.88% 4
1993 36.74% 32
2007 36.41% 64
1997 36.21% 32
2015 35.77% 1
2011 35.06% 16
2012 34.70% 64
2014 34.45% 64
2002 34.43% 16
2008 33.92% 32
2016 32.97% 16
2017 31.55% 32
2006 30.86% 16
2013 28.78% 8


Here's the contextual table for the NCAA in the last five seasons. It's worth noting that historically, teams have been getting fewer and fewer offensive rebounds. I'm not sure why.



Year 1 10 50 100 150 200 250 300 Last
2019 44.3 38.9 34.6 32.0 30.0 28.6 26.8 25.0 16.6
2018 38.6 35.6 33.1 31.1 29.6 28.2 26.8 24.5 18.7
2017 41.3 37.3 33.6 31.5 30.0 28.9 27.3 25.4 15.7
2016 41.9 38.4 34.2 32.1 30.6 29.1 27.9 25.6 17.6
2015 42.4 38.7 35.4 33.6 31.8 30.6 29.3 27.3 19.3

[Kedsy]

Defensive rebounding percentage (DR%) is the percentage you grab of defensive rebounding opportunites. The formula is DR/(DR + opp OR). Duke has been historically bad at this, presumably because of the way Coach K's defense works. I guess if you're constantly jumping into the passing lanes and your big men are constantly helping if the risks fail, it's hard to get into good rebounding position. And when I say historically bad, I mean really bad. Our best DR% performances are still outside the top 100 nationally, and usually outside the top 200. So far this year, we're around #300.

For Duke, I'd say 70% and up is good and below 67% is bad.




Year DR% NCAA
2017 70.44% 32
2018 70.39% 8
2015 69.84% 1
2007 69.68% 64
2019 69.21%
2014 69.06% 64
1989 68.78% 4
2010 67.92% 1
2011 66.83% 16
2009 66.82% 16
2008 66.27% 32
2016 65.61% 16
2002 65.49% 16
2012 65.38% 64
2013 65.33% 8
2003 65.33% 16
1999 65.10% 2
1987 64.79% 16
1988 64.78% 4
1995 64.60% n/a
1998 64.60% 8
2001 63.83% 1
1991 63.74% 1
2005 63.23% 16
2004 62.87% 4
1992 62.83% 1
2000 62.32% 16
2006 62.29% 16
1993 61.51% 32
1994 61.36% 2
1996 60.72% 64
1990 60.31% 2
1997 59.13% 32


Here's the contextual table for the NCAA in the last five seasons. Historically, for some reason, defensive rebounds have been on the rise. Hopefully that explains how three of our national champions (plus three other Final Four teams) managed to be below 64% in this stat.



Year 1 10 50 100 150 200 250 300 Last
2019 81.3 79.2 76.1 74.0 72.7 71.6 70.1 68.5 60.5
2018 81.8 77.5 74.8 73.3 72.2 71.2 70.3 68.7 62.7
2017 78.2 76.2 74.2 72.9 71.9 70.7 69.4 68.0 63.3
2016 78.2 75.7 73.7 72.1 71.0 70.0 69.0 67.6 61.7
2015 77.6 74.8 72.1 70.7 69.6 68.7 67.8 66.2 60.1

[Kedsy]

Turnover percentage is how many turnovers you commit per possession. Oddly, people calculate this percentage in two different ways. Some people simply divide TOs by total possessions. This is how I've been doing it in the game threads, and the calculation on which the table below is based. Others, including sports-reference.com, don't deduct offensive rebounds from the denominator (and thus their formula is TO/(FGA + (.475*FTA) + TO)). I guess the rationale is the offensive rebounder can turn it over and thus it should count as a separate possession, but this makes no sense to me. Especially since the same people who don't deduct ORs from TO% go ahead and do deduct it from steals percentage.

Anyway, if you calculate it TO/poss, then I'd say good is below 17% and bad is above 20%. Despite that, 7 of our 10 teams with the worst TO% have made the Final Four, suggesting that pushing the envelope on offense is probably a good thing even if it leads to turnovers.



Year TO% NCAA
2014 14.20% 64
2016 14.30% 16
2013 15.59% 8
2010 16.12% 1
2019 16.16%
2015 16.31% 1
2017 16.35% 32
1998 16.79% 8
2011 17.20% 16
2018 17.48% 8
2009 17.57% 16
2001 17.71% 1
2012 17.78% 64
1997 18.19% 32
2000 18.22% 16
2008 18.30% 32
2002 18.44% 16
2003 18.96% 16
2006 19.11% 16
1996 19.16% 64
1999 19.28% 2
2005 19.33% 16
1993 19.43% 32
2004 19.57% 4
1992 20.11% 1
1988 20.15% 4
1994 20.17% 2
1995 20.22% n/a
1991 20.93% 1
1990 21.61% 2
1987 22.75% 16
1989 22.81% 4
2007 22.97% 64


As I explained earlier, I don't have contextual tables for this, because my source for those calculates this stat differently than I do.

[Kedsy]

Opposing turnover percentage is how many turnovers you force, with the same oddities and caveats as offensive TO%. The classic Duke defenses of olde were really good at forcing turnovers, presumably how they justified being such bad defensive rebounders. Lately, not so much. Nine of our ten lowest performances in this stat have come in the past nine seasons. This year's team seems to be breaking this mold to an extent, but our opp TO% this season is still the 12th lowest since 1987.

These days I'd say above 20% is good and below 20% I wish we'd have done better.



Year TO% NCAA
1987 27.10% 16
1998 26.74% 8
1988 25.69% 4
1997 25.67% 32
1989 25.56% 4
2002 25.49% 16
1991 25.11% 1
1990 25.01% 2
2001 24.84% 1
2008 24.76% 32
2004 24.37% 4
2003 24.15% 16
2009 23.80% 16
1992 23.78% 1
1993 23.56% 32
2000 23.35% 16
1999 23.06% 2
1996 22.78% 64
2006 22.44% 16
2007 22.12% 64
2005 21.91% 16
2019 21.76%
1994 21.38% 2
2010 21.37% 1
2011 20.97% 16
2013 20.43% 8
1995 19.09% n/a
2015 18.63% 1
2012 18.59% 64
2014 18.40% 64
2017 17.36% 32
2016 17.25% 16
2018 17.21% 8


As I explained earlier, I don't have contextual tables for this, because my source for these calculates this stat differently than I do.

[Kedsy]

Assist percentage, or %assists, is the percentage of your field goals that were assisted. The formula is assists/FGM. Duke generally ranks low nationally is this category, presumably because K "let's them play" and we usually have a lot of guys who are good at one-on-one scoring.

Seven of our top nine performances in this stat have at least made the Elite Eight, including three champions, so I guess it matters (presumably because it means less "hero ball"). That said, the 1999 team was our 9th lowest in this category at just over 50%.

For Duke, I'd say above 55% is good and below 50% is bad.



Year Asst% NCAA
1992 58.94% 1
1989 58.81% 4
1987 57.93% 16
2001 57.60% 1
2018 57.60% 8
2002 57.18% 16
1988 57.12% 4
1990 56.62% 2
1991 56.15% 1
2000 55.89% 16
2006 55.77% 16
1994 55.68% 2
2014 54.79% 64
2019 53.78%
1995 53.72% n/a
2015 53.50% 1
1993 53.49% 32
2010 52.91% 1
1996 52.89% 64
1997 52.74% 32
2007 52.68% 64
2004 52.65% 4
2013 52.08% 8
2008 51.56% 32
1999 50.88% 2
2011 50.87% 16
2009 50.67% 16
2003 49.50% 16
2005 49.36% 16
2016 48.63% 16
2012 48.40% 64
1998 47.55% 8
2017 47.50% 32


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 73.7 62.4 58.6 55.4 53.1 51.4 48.8 46.0 37.6
2018 67.6 61.3 58.1 55.3 53.4 51.6 49.7 47.4 34.7
2017 66.8 62.6 58.2 55.7 53.5 51.3 49.3 46.9 39.7
2016 71.3 64.1 58.2 55.7 53.4 51.3 49.4 46.7 36.7
2015 69.6 63.6 59.5 56.7 54.3 52.0 49.9 47.6 38.8

[Kedsy]

Opposing assist percentage is how many of our opponents' baskets are assisted. In this case, however, "better" might not necessarily be better. Only one of our "best" seven performances in this stat have gone past the Sweet 16 (including two first-round exits) and five of our "worst" six have gotten to at least the Elite Eight. Possibly the explanation is that if teams are going one-on-one against us that means our defense has broken down. If they can only score when they run a good play, they're not scoring as much. As such it's probably not worth it assigning "good" and "bad" to this stat, which is fortunate because our current team has the highest percentage of any Duke team since at least 1987.



Year Asst% NCAA
2013 40.02% 8
2005 41.89% 16
2009 42.87% 16
2012 43.98% 64
2014 44.81% 64
2006 46.01% 16
2017 46.61% 32
2015 46.99% 1
2008 47.61% 32
1991 47.93% 1
1999 48.15% 2
2003 48.27% 16
1995 48.57% n/a
2011 48.88% 16
2010 48.98% 1
2007 49.02% 64
2002 49.66% 16
1994 49.89% 2
2004 49.89% 4
1993 50.33% 32
2000 50.43% 16
2001 50.58% 1
1997 50.95% 32
1989 51.16% 4
1996 51.61% 64
2016 52.40% 16
1990 53.00% 2
1988 53.62% 4
1992 53.92% 1
1987 54.71% 16
1998 55.05% 8
2018 56.11% 8
2019 56.15%


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 32.8 38.2 44.7 48.2 50.6 52.3 55.1 57.9 72.2
2018 37.9 42.3 47.3 49.4 51.2 52.7 55.2 57.5 73.8
2017 40.2 43.4 47.1 49.7 51.4 52.9 54.7 57.6 67.1
2016 40.4 43.1 47.0 49.4 51.1 53.0 54.9 57.7 72.7
2015 40.0 44.0 48.2 50.4 51.9 53.8 55.6 57.9 69.4

[Kedsy]

Assist to turnover ratio is not an advanced stat, but people seem to like to follow it so I include it in the game breakdowns. Four of our national champions are in the top 10 here, but there doesn't really seem to be a strong correlation to anything.

Having said that, obviously if you're below 1.0 in this stat something isn't working on offense.



Year a/to NCAA
2014 1.552 64
2019 1.492
2018 1.409 8
2015 1.397 1
2016 1.350 16
2001 1.320 1
2013 1.305 8
2010 1.294 1
2002 1.250 16
2000 1.237 16
1992 1.216 1
2011 1.183 16
1998 1.158 8
2017 1.150 32
1988 1.149 4
1993 1.126 32
1999 1.116 2
1997 1.116 32
2006 1.105 16
1994 1.098 2
2009 1.084 16
2008 1.076 32
1991 1.068 1
1995 1.065 n/a
1996 1.063 64
1989 1.056 4
2004 1.051 4
2012 1.019 64
1990 1.005 2
1987 0.982 16
2003 0.965 16
2005 0.938 16
2007 0.851 64

[Nugget]

I know this is jumping the gun a bit, but when you get to OR and DR rates, will you include how team and deadball rebounds are accounted for?

[Kedsy]

Opposing assist to turnover ratio is also not an advanced stat, and also not worth assigning "good" and "bad" because at best this is simply indirectly measuring how many turnovers you force.



Year a/to NCAA
2009 0.631 16
1989 0.640 4
2002 0.641 16
2005 0.644 16
1998 0.655 8
2008 0.670 32
1991 0.673 1
1997 0.673 32
1987 0.677 16
2013 0.679 8
2004 0.690 4
1999 0.690 2
2003 0.691 16
2001 0.698 1
1988 0.712 4
2006 0.755 16
2010 0.761 1
2007 0.765 64
1990 0.774 2
2000 0.779 16
2011 0.804 16
1993 0.805 32
1996 0.817 64
2019 0.820
1994 0.857 2
2012 0.876 64
1992 0.877 1
2014 0.894 64
2015 0.940 1
1995 0.964 n/a
2017 1.000 32
2018 1.185 8
2016 1.225 16

[Kedsy]

Block percentage is how many blocks you get per 2-point shot. The formula is blocks/2ptFGA. Teams that block three-point shots thus get a small boost, but I guess we just have to live with that. In the game threads I give percentages for blocks per total FGA and blocks per 2ptFGA, and I try to deduct if I noticed we blocked a 3pt attempt. Putting that aside, this year's team so far is our best shot-blocking team ever (and is tied for 6th nationally).

I'd say good is above 12% and bad is below 8%.



Year block% NCAA
2019 17.65%
2004 14.87% 4
2005 13.87% 16
1999 13.25% 2
2003 12.58% 16
1994 12.53% 2
2018 12.39% 8
1998 11.95% 8
2006 11.58% 16
2007 11.45% 64
2000 11.32% 16
2017 10.89% 32
2001 10.49% 1
2010 9.85% 1
2016 9.75% 16
2011 9.71% 16
1993 9.70% 32
2009 9.46% 16
1995 9.18% n/a
1997 9.14% 32
2015 9.04% 1
2013 9.01% 8
2012 8.98% 64
2008 8.76% 32
1996 8.63% 64
1989 8.08% 4
1991 8.07% 1
1992 8.00% 1
1987 7.47% 16
2014 7.44% 64
2002 7.25% 16
1990 7.03% 2
1988 7.03% 4


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 22.3 17.1 13.1 11.4 10.0 8.9 7.6 6.4 3.2
2018 18.6 15.1 12.1 10.5 9.6 8.9 7.8 6.6 3.9
2017 17.2 14.5 11.9 10.7 9.6 8.5 7.4 6.6 4.4
2016 15.5 14.2 11.8 10.5 9.6 8.7 7.6 6.6 4.3
2015 20.0 14.9 12.7 11.1 10.1 9.0 8.1 6.8 2.9


Note that I don't bother tracking opposing blocks, because when we're on offense it seems unimportant, merely a dramatic missed shot attempt. But when we're doing it on defense, it's fun.

[Kedsy]

Steals percentage is how many steals you get per possession. The formula is steals/poss. Oddly, even the people who use a different formula for the denominator in TO% use the sensible one here. I don't list this stat in my game recaps because I generally think steals are just another turnover, but the way this year's team turns steals into dunks, maybe I should track it. This year's team so far is our best stealing team ever (and is tied for 2nd nationally).

I'd say good is above 11% and bad is below 9%.



Year steals% NCAA
2019 14.46%
2001 13.70% 1
2004 13.05% 4
2006 12.96% 16
2002 12.91% 16
2000 12.85% 16
1998 12.82% 8
2005 12.35% 16
1997 12.04% 32
2009 12.03% 16
1999 11.87% 2
2003 11.79% 16
1991 11.74% 1
2008 11.68% 32
1990 11.39% 2
1988 11.27% 4
1989 11.23% 4
1993 11.16% 32
1987 11.09% 16
1992 11.01% 1
2007 10.92% 64
2015 10.77% 1
2014 10.56% 64
2018 10.45% 8
2011 10.35% 16
1994 10.17% 2
2010 9.99% 1
2013 9.44% 8
2016 9.13% 16
2012 9.00% 64
2017 8.78% 32
1996 8.63% 64
1995 8.01% n/a


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 15.2 13.1 11.3 10.3 9.4 8.7 8.1 7.3 4.5
2018 14.0 11.6 10.4 9.7 9.0 8.5 7.9 7.2 5.3
2017 15.3 12.1 10.4 9.6 9.1 8.4 7.8 7.2 5.2
2016 13.8 12.0 10.4 9.4 8.9 8.5 7.9 7.2 4.2
2015 15.5 12.9 11.1 10.3 9.7 9.2 8.6 8.0 5.2


Note that I don't bother tracking opposing steals, because when we're on offense it seems unimportant, just another turnover. But when we're doing it on defense, like the block it's fun.

[Kedsy]

Three-point percentage is obviously not an advanced stat, but it's an important one so we track it. Duke is usually really good at this. Our top 15 three-point shooting performances include all five national champions and a couple runners-up, but our bottom ten performances include four Final Fours and a couple Elite Eights, so not hitting from outside doesn't seem to be a deal-breaker. This is good because this year's team's 3-point percentage is by far Duke's worst ever.

I'd say good is 40% and up and bad is below 34%.



Year 3pt% NCAA
1992 43.40% 1
1987 40.22% 16
2013 39.94% 8
1993 39.74% 32
1999 39.65% 2
2014 39.45% 64
1997 38.94% 32
1990 38.90% 2
2006 38.76% 16
2015 38.66% 1
2016 38.54% 16
2001 38.51% 1
2010 38.49% 1
1991 38.34% 1
2000 38.27% 16
1995 38.14% n/a
2007 38.08% 64
2005 38.01% 16
2017 37.83% 32
2008 37.75% 32
1996 37.56% 64
2011 37.40% 16
1988 37.25% 4
2018 37.18% 8
2012 37.08% 64
1998 36.94% 8
1989 36.52% 4
1994 36.52% 2
2004 36.43% 4
2003 36.31% 16
2002 36.27% 16
2009 34.86% 16
2019 33.20%


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 48.2 41.6 38.4 36.6 35.3 33.9 31.7 29.9 20.3
2018 43.4 40.3 37.8 36.4 35.5 34.5 33.6 32.6 27.9
2017 42.9 40.3 38.0 36.6 35.5 34.6 33.5 32.0 27.9
2016 43.4 40.2 37.3 36.3 35.2 34.1 33.2 31.7 27.2
2015 44.7 40.0 37.8 35.9 34.9 33.7 32.6 31.3 25.5

[Kedsy]

Likewise, opposing 3-point percentage is not an advanced stat but is important enough to track. So far this year's team is Duke's best-ever in this category. I'd say good is 30% and down and bad is over 37%, but there doesn't really seem to be a meaningful correlation here. Three of our top six played in the national championship game, but also three of our national champions were in the bottom seven in this category.



Year 3pt% NCAA
2019 28.19%
2010 28.26% 1
1993 28.85% 32
2013 29.04% 8
2017 29.26% 32
1994 29.65% 2
1999 30.08% 2
2002 30.27% 16
2006 30.36% 16
1998 30.47% 8
2005 30.48% 16
2014 30.73% 64
2015 31.44% 1
2007 31.52% 64
2012 31.74% 64
1989 31.97% 4
2018 32.03% 8
2011 32.41% 16
2004 32.41% 4
2008 32.94% 32
1987 33.47% 16
1996 33.54% 64
1990 33.68% 2
2009 33.76% 16
2016 33.98% 16
1997 34.02% 32
2001 34.39% 1
2003 34.60% 16
1991 34.67% 1
1988 35.08% 4
2000 35.75% 16
1992 37.69% 1
1995 38.96% n/a


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 20.4 26.7 29.3 31.2 32.6 33.6 35.2 37.2 45.0
2018 27.8 31.0 32.7 33.5 34.4 35.2 36.0 37.2 41.2
2017 28.6 30.5 32.3 33.5 34.3 35.3 36.1 37.3 41.8
2016 29.1 30.3 32.1 33.1 34.0 34.9 35.7 36.9 41.3
2015 26.5 29.9 31.6 32.8 33.7 34.6 35.5 36.9 39.8

[Kedsy]

2-point percentage is not an advanced stat, but also not regularly reported with the basic stats so it's probably worth looking at. For example, our current team may be Duke's worst three-point shooting team ever, but it's also our best two-point shooting team ever, which I suppose balances everything out.

There is a much stronger correlation to NCAAT success here than in three-point shooting. Seven of our top nine teams in this category at least made the Elite Eight, including four national champions. But while eight of our bottom nine teams lost in the Sweet 16 or earlier (including three first round exits), our worst ever team in this category won the 2010 national championship.

I'd say 55% and up is good and below 50% is bad.



Year 2pt% NCAA
2019 59.59%
1989 57.55% 4
2002 57.43% 16
1999 56.51% 2
2018 56.01% 8
1992 55.94% 1
2015 55.93% 1
2001 54.99% 1
1993 54.14% 32
1991 54.12% 1
2006 54.06% 16
2017 53.56% 32
2000 53.22% 16
1998 52.50% 8
1994 52.50% 2
2004 52.47% 4
2011 52.45% 16
1988 52.08% 4
1995 51.88% n/a
2008 51.74% 32
2012 51.02% 64
2016 50.97% 16
1990 50.93% 2
2013 50.85% 8
2007 50.57% 64
1997 50.41% 32
2014 50.29% 64
1987 50.22% 16
2003 49.89% 16
2009 49.58% 16
2005 48.87% 16
1996 47.88% 64
2010 46.96% 1

[Kedsy]

Opposing two-point percentage holds similar interest to our own 2pt%, though the correlation to NCAAT success isn't as strong. I'd say good is 45% and down and bad is over 50%.

Interestingly, the 1992 Duke national champion team ranked 2nd-to-worst in Duke history (or at least since the 3-point shot was introduced in 1987) in both opposing 2pt% and opposing 3pt%.



Year 2pt% NCAA
2005 41.30% 16
1999 42.24% 2
2019 42.35%
2011 43.14% 16
2004 43.49% 4
2000 43.66% 16
2001 43.93% 1
2010 44.13% 1
1989 44.60% 4
1997 44.66% 32
1998 44.71% 8
1994 45.02% 2
1995 45.41% n/a
2018 45.41% 8
2007 45.66% 64
2013 46.19% 8
2015 46.30% 1
2006 46.39% 16
2002 46.44% 16
2009 46.76% 16
1991 46.83% 1
2012 46.96% 64
2008 46.97% 32
1988 47.15% 4
1990 47.20% 2
1987 47.23% 16
1996 47.46% 64
2003 47.70% 16
1993 48.12% 32
2017 48.89% 32
2016 48.89% 16
1992 48.98% 1
2014 50.32% 64

[Kedsy]

The percentage of shots which are three-point shots (3pa/FGA) is an interesting one, because it's not a situation where more is better or less is worse. Though the 2001 team had the highest percentage of any Duke team and won the natty, none of the next eight teams on this list got past the Sweet 16. Of the bottom eight teams, six made at least the Elite Eight (though the other two lost in the first round). Still, you probably don't want too few and you don't want too many. For Duke, it would seem the Goldilocks zone is around 33%.

This year's team is therefore shooting too many threes, especially since we hit so few of them.



Year %three NCAA
2001 41.78% 1
2005 39.84% 16
2016 39.84% 16
2014 39.65% 64
2008 39.16% 32
2012 38.59% 64
2017 38.34% 32
2002 37.57% 16
1997 36.84% 32
2019 36.34%
2018 36.32% 8
2011 35.27% 16
2006 35.16% 16
2009 35.04% 16
2000 34.16% 16
2003 33.92% 16
2004 33.41% 4
2015 33.41% 1
2013 33.25% 8
1996 32.96% 64
2010 32.93% 1
1998 32.44% 8
1999 30.51% 2
2007 29.65% 64


Coach K seemingly changed his strategy regarding the three-point shot after his injured season of 1995. So I separated the years before 1996 into the chart below. Considering the way the game is played today, the chart has no significance, I just included it for completeness sake.



Year %three NCAA
1995 31.65% n/a
1994 24.12% 2
1993 23.74% 32
1988 20.70% 4
1991 19.12% 1
1992 19.04% 1
1987 18.94% 16
1989 18.33% 4
1990 16.75% 2


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 59.2 49.7 44.5 41.5 39.5 37.6 35.7 33.5 20.7
2018 55.5 47.5 43.6 40.1 38.1 36.2 34.8 32.0 24.5
2017 58.2 46.6 41.9 39.5 37.3 35.6 33.1 30.5 19.7
2016 52.7 46.5 41.5 38.2 36.2 34.3 32.1 29.7 23.7
2015 52.0 45.6 39.7 36.8 35.0 33.1 31.2 28.7 19.3

[Kedsy]

Opposing percent threes is another one that's hard to read, because if teams aren't taking threes it could be because twos are so easy. Thus only one of our top 10 teams in this category got past the Sweet 16, while three of the ten lost in the 1st round and two others lost in the 2nd round. But that doesn't mean allowing a ton of opposing threes is the right strategy, either. Again it seems there's a Goldilocks zone and for Duke this might be around 25%. Our current team is allowing opponents to take the 2nd-most percentage of three-point shots since 1996, which could be dangerous if an opponent gets super-hot from out there at the wrong time.



Year %three NCAA
2005 20.94% 16
2006 21.26% 16
1997 23.41% 32
2014 24.14% 64
2012 24.15% 64
2007 24.34% 64
2001 24.36% 1
2011 24.55% 16
2000 24.61% 16
2008 25.04% 32
1998 25.05% 8
2003 25.09% 16
2004 25.17% 4
2010 25.36% 1
1999 25.56% 2
2009 25.98% 16
2002 26.25% 16
2013 26.89% 8
1996 27.00% 64
2015 27.75% 1
2017 28.81% 32
2016 29.63% 16
2019 34.82%
2018 36.84% 8


I again split out the years prior to 1996, because things were different then.



Year %three NCAA
1987 13.92% 16
1990 15.38% 2
1988 16.33% 4
1989 18.19% 4
1991 19.35% 1
1993 19.86% 32
1992 20.38% 1
1994 23.10% 2
1995 24.18% n/a


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 23.8 30.1 34.0 36.0 37.8 39.4 41.4 43.5 53.5
2018 24.7 30.6 33.7 35.5 36.9 38.0 39.6 41.7 48.7
2017 26.8 28.9 32.8 34.3 35.7 36.9 38.7 40.2 46.8
2016 25.1 28.9 31.7 33.4 34.9 36.1 37.1 39.5 47.2
2015 21.8 27.2 29.9 31.9 33.6 35.3 36.5 38.4 45.3

[Kedsy]

True shooting percentage is a way to combine free throw shooting and field goal shooting. The formula is points/((2*FGA)+(.475*FTA)). For an individual offensive player I think this is an interesting stat. For a team, not so much, because other than how often you put your opponent on the line (covered by FT rate), you don't have any control over this. But while I don't include it in my game report, I figured it can't hurt to include it here. Above 57% is good; below 54% is bad.



Year TS% NCAA
1992 61.75% 1
1999 60.47% 2
2006 59.84% 16
1989 59.46% 4
2015 59.31% 1
2002 58.95% 16
1993 58.90% 32
2001 58.85% 1
2017 58.79% 32
1991 58.61% 1
2018 58.58% 8
2000 58.26% 16
2019 57.81%
2013 57.61% 8
2011 57.59% 16
1990 57.46% 2
2014 57.38% 64
2016 57.31% 16
2004 57.16% 4
2008 56.86% 32
1997 56.70% 32
1994 56.59% 2
1988 56.51% 4
2012 56.51% 64
1998 56.36% 8
2005 55.99% 16
1995 55.87% n/a
2007 55.72% 64
2003 55.61% 16
1987 55.57% 16
2010 54.94% 1
2009 54.54% 16
1996 53.91% 64


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 63.3 60.1 58.3 56.9 55.6 54.2 52.6 50.4 41.5
2018 63.9 60.1 57.5 56.1 54.9 53.9 53.0 51.7 46.1
2017 61.8 59.3 57.1 55.6 54.7 53.7 52.6 51.4 44.4
2016 61.2 59.2 56.5 55.1 54.1 53.2 52.3 50.8 45.8
2015 61.2 58.9 56.3 54.1 53.2 52.3 51.3 49.9 43.7

[Kedsy]

Free throw percentage is another non-advanced stat. I included it here only to point out that our current team is Duke's worst free-throw shooting team since at least 1987, by a fair margin.



Year FT% NCAA
1990 76.35% 2
2017 76.25% 32
2006 76.13% 16
2010 75.86% 1
2011 75.30% 16
1992 74.78% 1
2000 74.19% 16
2013 73.87% 8
2004 73.60% 4
2014 72.92% 64
1991 72.64% 1
2016 72.29% 16
1993 72.08% 32
2009 71.99% 16
1988 71.85% 4
2005 71.25% 16
2003 71.24% 16
1994 71.14% 2
2018 71.05% 8
1987 70.98% 16
1999 70.48% 2
1996 70.30% 64
2012 70.14% 64
1997 70.04% 32
2015 69.87% 1
2008 69.72% 32
2001 69.56% 1
2002 68.98% 16
2007 68.91% 64
1989 67.47% 4
1998 67.40% 8
1995 67.05% n/a
2019 65.02%


Here's the contextual table for the NCAA in the last five seasons:



Year 1 10 50 100 150 200 250 300 Last
2019 81.4% 78.0% 74.4% 72.1% 70.4% 68.8% 67.1% 64.3% 53.9%
2018 81.0% 77.9% 75.2% 73.4% 71.8% 70.8% 69.6% 67.5% 60.8%
2017 80.0% 77.7% 74.4% 72.5% 70.9% 69.4% 68.2% 66.1% 60.5%
2016 79.3% 76.6% 73.3% 71.9% 70.3% 69.2% 68.1% 66.5% 58.9%
2015 79.4% 76.5% 73.0% 71.4% 69.8% 68.6% 67.1% 65.5% 57.5%

[Kedsy]

Margin doesn't get included in the game reports, because for one game the margin is simply the difference in the final score. But over the course of the season it's an interesting stat, so I'm going to end on it. It's also interesting to see how "raw" margin changes as it becomes more of an advanced stat.

This first table is the raw scoring margin for all Duke teams since 1987. As you can see, this year's team has the largest margin by a... (ahem) large margin. It makes sense that the teams that win by (on average) the most points are the better teams and the ones on the opposite end of the scale are the worst.



Year margin NCAA
2019 27.20
1999 24.67 2
1998 21.50 8
2001 20.21 1
2002 19.74 16
1989 16.75 4
2000 16.68 16
2011 16.19 16
2010 15.95 1
1992 15.33 1
1993 15.22 32
2015 15.10 1
1988 14.94 4
2004 14.81 4
2018 14.73 8
1991 14.28 1
2008 13.79 32
1997 13.45 32
2006 13.00 16
2005 12.97 16
1990 12.53 2
2013 11.89 8
2003 11.48 16
2009 11.32 16
2014 11.03 64
2017 10.65 32
1994 10.29 2
1987 9.88 16
2012 8.65 64
2016 8.64 16
2007 8.39 64
1996 5.58 64
1995 4.06 n/a


This second table controls for pace. It shows average margin per 100 possessions. So teams who played at a slower pace, like the 2015 team, move up in the rankings. Again, though, the current team is on top, though a large part of that is probably that this team hasn't played any ACC games yet.



Year margin NCAA
2019 35.44
1999 32.72 2
1998 29.27 8
2001 26.28 1
2002 25.49 16
2010 23.97 1
2011 23.04 16
2015 22.81 1
2000 21.89 16
1989 21.23 4
2004 21.03 4
2018 20.71 8
1992 20.67 1
1993 20.13 32
1988 19.85 4
1997 19.09 32
2008 18.58 32
2005 18.42 16
2006 18.11 16
1991 18.07 1
2013 17.42 8
2014 16.71 64
2009 16.21 16
2003 15.62 16
1990 15.58 2
2017 15.30 32
1994 14.65 2
1987 13.38 16
2012 12.59 64
2016 12.56 16
2007 12.50 64
1996 8.03 64
1995 5.70 n/a


This final table controls for both pace and quality of opponent. These numbers are taken from Pomeroy (and thus only go back to 2002). Here, you see our current team only ranks 5th in the past 18 seasons (though that's still pretty good, as the four teams ahead of this year are the outstanding teams from 2002, 2010, 2015, and 2004).



Year margin NCAA
2002 34.20 16
2010 33.30 1
2015 32.50 1
2004 32.30 4
2019 30.60
2018 28.80 8
2011 28.40 16
2006 27.60 16
2005 27.40 16
2013 27.00 8
2008 26.40 32
2014 24.30 64
2009 24.30 16
2017 24.20 32
2003 23.70 16
2007 23.30 64
2016 21.90 16
2012 19.70 64

[JetpackJesus]

If there were an advanced stat for posting useful stuff, Kedsy would be killing it.
Some ideas:
PER: Poster Efficiency Rating
SARP: Sporks Above Replacement Poster
TS%: True Spork Percentage
eS%: Effective Spork Percentage
Now someone just needs to figure out the math...

[CDu]

I know this is jumping the gun a bit, but when you get to OR and DR rates, will you include how team and deadball rebounds are accounted for?

Team rebounds are simply those missed shots that aren’t collected by a player and go out of bounds off the other team. These count towards team rebounding % just like those attributed to individual players do.

I am not sure what you mean by “deadball” rebound. Do you mean what you see in play by play on ESPN for (e.g.) missing the first of two free three throws? Those don’t count towards anything at all. It is just how ESPN chooses to handle presenting the play by play.

[-jk]

Team rebounds are simply those missed shots that aren’t collected by a player and go out of bounds off the other team. These count towards team rebounding % just like those attributed to individual players do.

I am not sure what you mean by “deadball” rebound. Do you mean what you see in play by play on ESPN for (e.g.) missing the first of two free three throws? Those don’t count towards anything at all. It is just how ESPN chooses to handle presenting the play by play.

Deadball rebounds are a stat recognized for decades, however weird...

-jk

[Nugget]

Team rebounds are simply those missed shots that aren’t collected by a player and go out of bounds off the other team. These count towards team rebounding % just like those attributed to individual players do.

I am not sure what you mean by “deadball” rebound. Do you mean what you see in play by play on ESPN for (e.g.) missing the first of two free three throws? Those don’t count towards anything at all. It is just how ESPN chooses to handle presenting the play by play.

I'd been erroneously thinking about this by comparing OR totals from box scores against Duke's missed FG/FT totals and trying to figure out how best to figure out of many of the missed FTs weren't OR opportunities since most box scores don't spell out the number of deadball rebounds. But, Kedsy's explanation of the calculation OR / (OR + Opp. DR) makes clear I was focusing on the wrong thing.

However, in terms of the "team rebounds" it does seem essential to know what kind of box score you are looking at -- for example, the CBS Sports.com ones only list up individual player totals and then the bottom line "team" numbers don't appear to include team/deadball rebounds (i.e. their "team" numbers just total up the individual numbers), whereas the ESPN.com box scores -- while they do not say so explicitly -- have "team" numbers that do include the "team rebounds" along with the individual rebound totals (as indicated by the fact that if you add up the individual rebound numbers it doesn't match what they print as the total for the team). The deadball rebounds don't show up in either box score (and, absurdly, the NCAA.com box scores don't break out OR vs. DR splits at all).

It could make a big difference -- for instance, the CBS Sports.com box score last night for Penn-Villanova has just the individual players' rebounds listed and shows as "team totals" for Penn 9 OR and 18 DR (27 total rebounds) and for Villanova 6 OR and 15 DR (21 total rebounds). Using those would result in Kedsy's calculation reflecting Penn with a 37.5% OR rate (9/24).

But, ESPN's box score has as the team totals 13 OR and 22 DR for Penn and 6 OR and 16 DR for Villanova -- i.e., Penn got a relatively whopping 8 "team rebounds" (4 Offensive) to just 1 for Nova. When those team rebounds are added in it pushes Penn's OR rate to 44% (13/29).

What's the most reliable source for box scores?

[Nugget]

2-point percentage is not an advanced stat, but also not regularly reported with the basic stats so it's probably worth looking at. For example, our current team may be Duke's worst three-point shooting team ever, but it's also our best two-point shooting team ever, which I suppose balances everything out.

There is a much stronger correlation to NCAAT success here than in three-point shooting. Seven of our top nine teams in this category at least made the Elite Eight, including four national champions. But while eight of our bottom nine teams lost in the Sweet 16 or earlier (including three first round exits), our worst ever team in this category won the 2010 national championship.

I'd say 55% and up is good and below 50% is bad.

It makes sense that this would correlate strongly with overall success since shooting percentage would presumably be driven less by skill variances than the quality of shots taken -- i.e., the better teams get more easy shots in transition or, more generally, work the ball for better shots and shoot relatively fewer contested mid-range jumpers.

And, the odd-ball 2010 team presumably made up for the relatively poor 2 point shooting percentage with its abnormally strong OR prowess and very low TO rate, so we got more shots up and could afford to make a lower than usual percentage of them.

[CDu]

I'd been erroneously thinking about this by comparing OR totals from box scores against Duke's missed FG/FT totals and trying to figure out how best to figure out of many of the missed FTs weren't OR opportunities since most box scores don't spell out the number of deadball rebounds. But, Kedsy's explanation of the calculation OR / (OR + Opp. DR) makes clear I was focusing on the wrong thing.

However, in terms of the "team rebounds" it does seem essential to know what kind of box score you are looking at -- for example, the CBS Sports.com ones only list up individual player totals and then the bottom line "team" numbers don't appear to include team/deadball rebounds (i.e. their "team" numbers just total up the individual numbers), whereas the ESPN.com box scores -- while they do not say so explicitly -- have "team" numbers that do include the "team rebounds" along with the individual rebound totals (as indicated by the fact that if you add up the individual rebound numbers it doesn't match what they print as the total for the team). The deadball rebounds don't show up in either box score (and, absurdly, the NCAA.com box scores don't break out OR vs. DR splits at all).

It could make a big difference -- for instance, the CBS Sports.com box score last night for Penn-Villanova has just the individual players' rebounds listed and shows as "team totals" for Penn 9 OR and 18 DR (27 total rebounds) and for Villanova 6 OR and 15 DR (21 total rebounds). Using those would result in Kedsy's calculation reflecting Penn with a 37.5% OR rate (9/24).

But, ESPN's box score has as the team totals 13 OR and 22 DR for Penn and 6 OR and 16 DR for Villanova -- i.e., Penn got a relatively whopping 8 "team rebounds" (4 Offensive) to just 1 for Nova. When those team rebounds are added in it pushes Penn's OR rate to 44% (13/29).

What's the most reliable source for box scores?

If there is no presentation of team rebounds in the box score, it is wrong. Total rebounds are the sum of individual and team rebounds. CBS does include team rebounds, just as a separate combined category (1 for Nova and 8 for Penn). ESPN breaks them out. So if you are wanting to look at OR% or DR%, ESPN is better.

[cato]

Here's a table with Duke's possessions per game in every year since 1987 (which is when they first started tracking offensive rebounds, making this calculation possible). As you can see, for Duke, fast is generally good, as six of our fastest 7 teams made the Final Four. But slow isn't necessarily bad, as two of our five slowest teams won the national championship (though it's also worth noting that the other three of our slowest five teams got knocked out in the first round). If this year's team keeps up its currrent pace, we'll have the 5th-fastest team in "modern" Duke history.



Year Pace NCAA
1990 79.34 2
1989 78.91 4
1991 78.53 1
2002 77.19 16
2019 76.74
2001 76.65 1
1999 75.39 2
1988 75.02 4
1993 75.01 32
2000 74.82 16
2008 73.98 32
1992 73.67 1
1998 73.44 8
2003 72.96 16
1987 72.71 16
2006 71.55 16
2018 70.65 8
1994 70.28 2
2011 70.25 16
1995 70.18 n/a
2005 70.13 16
2004 69.95 4
1997 69.94 32
2009 69.61 16
2017 69.59 32
1996 68.98 64
2016 68.29 16
2013 68.25 8
2012 68.16 64
2010 66.55 1
2007 66.38 64
2015 65.78 1
2014 65.76 64






Interesting that Hurley and Duhon were applying pressure on the ball handler (and the team was great in transition) during all of the fastest years.

[Kedsy]

Interesting that Hurley and Duhon were applying pressure on the ball handler (and the team was great in transition) during all of the fastest years.

Yeah, who the lead guard is appears to be very important for a fast pace. Though it's interesting that two of the fastest years (1989 and 1988) featured Quin Snyder as our PG.

[cato]

Yeah, who the lead guard is appears to be very important for a fast pace. Though it's interesting that two of the fastest years (1989 and 1988) featured Quin Snyder as our PG.

Good point. Billy King was there in 88. Hurley did not arrive until 90. Was the D particularly disruptive in 89?

Interesting team that kind of gets lost in the conversation.

Ferry/Laettner
Henderson/Abdelnaby/Koubek
Brickey
Henderson
Snyder

[Kedsy]

Good point. Billy King was there in 88. Hurley did not arrive until 90. Was the D particularly disruptive in 89?

Interesting team that kind of gets lost in the conversation.

Ferry/Laettner
Henderson/Abdelnaby/Koubek
Brickey
Henderson
Snyder

1989 team had Duke's 3rd best unadjusted dRating and 5th highest opposing TO%, so pretty disruptive, I guess.

[cato]

1989 team had Duke's 3rd best unadjusted dRating and 5th highest opposing TO%, so pretty disruptive, I guess.

So: the teams that did the best job picking-up the ball handler before he is used to it/getting out in the passing lanes have played at the fastest pace.

[robed deity]

Interesting the '92 team unadjusted def efficiency was relatively high, a full point above the '15 team, who didn't figure it out until the end. Maybe the schedule was tough that year, everyone was scoring at a higher clip at that time, or that team had an on/off switch. Maybe all of the above.

[Jaymf7]

Thanks for this great work. Any chance you can post the historical stats for team oRating - dRating? Seems like that would have a very high correlation with success.

[CDu]

Interesting the '92 team unadjusted def efficiency was relatively high, a full point above the '15 team, who didn't figure it out until the end. Maybe the schedule was tough that year, everyone was scoring at a higher clip at that time, or that team had an on/off switch. Maybe all of the above.

Both pace of play and defensive quality have changed a lot over the decades. Not coincidentally, as teams have made more and more of an emphasis on preventing fast breaks (which have a higher probability of scoring than half-court offense).

[robed deity]

Both pace of play and defensive quality have changed a lot over the decades. Not coincidentally, as teams have made more and more of an emphasis on preventing fast breaks (which have a higher probability of scoring than half-court offense).

Yeah, probably has a lot to do with it. Still, the '92 team was not near the top in pace, and the '91 and even '90 teams were significantly better in raw Defensive Efficiency. Just a little surprising, as '92 (and '99, of course) had reached mythical status in my brain. Maybe they just didn't blow teams out on the regular.

[House P]

Yeah, probably has a lot to do with it. Still, the '92 team was not near the top in pace, and the '91 and even '90 teams were significantly better in raw Defensive Efficiency. Just a little surprising, as '92 (and '99, of course) had reached mythical status in my brain. Maybe they just didn't blow teams out on the regular.

The 1992 team is a conundrum from an advanced analytics standpoint. In my mind, the team was truly dominant, but the advanced metric indicate that they were "merely" very good. For example, the 92 team finished the year #3 in SRS (https://www.sports-reference.com/cbb/seasons/1992-ratings.html), significantly behind Indiana and with essentially the same SRS rating as Kansas, Ohio St, and Arizona. KenPom ratings don't go back that far, but I distinctly remember the 92 team not being ranked #1 in the final Sagarin ratings.

However, one thing to consider about the 1992 team is that Bobby Hurley and Grant Hill were injured for key parts of the season. In fact, Duke's only losses came without Hurley (Wake) or with Hurley playing on a broken foot (UNC). Duke's less than stellar 1 point win vs a mediocre Clemson team came without Grant.

The 92 team wasn't particularly deep, so I expect that these injuries play a role in the team's less-than-historic advanced metrics. I wonder what the advanced metrics would look like in games where Hurley and Hill were 100% healthy.

[MChambers]

The 1992 team is a conundrum from an advanced analytics standpoint. In my mind, the team was truly dominant, but the advanced metric indicate that they were "merely" very good. For example, the 92 team finished the year #3 in SRS (https://www.sports-reference.com/cbb/seasons/1992-ratings.html), significantly behind Indiana and with essentially the same SRS rating as Kansas, Ohio St, and Arizona. KenPom ratings don't go back that far, but I distinctly remember the 92 team not being ranked #1 in the final Sagarin ratings.

However, one thing to consider about the 1992 team is that Bobby Hurley and Grant Hill were injured for key parts of the season. In fact, Duke's only losses came without Hurley (Wake) or with Hurley playing on a broken foot (UNC). Duke's less than stellar 1 point win vs a mediocre Clemson team came without Grant.

The 92 team wasn't particularly deep, so I expect that these injuries play a role in the team's less-than-historic advanced metrics. I wonder what the advanced metrics would look like in games where Hurley and Hill were 100% healthy.

And Cherokee Parks also was injured.

I still think 92 was a great team, but it sure didn't roll thru the NCAA tournament.

[CDu]

The 1992 team had some serious NBA talent: Laettner, Hill, Hurley, and Parks were all lottery picks and regular starters, with Laettner being a reasonable NBA star and Hill being arguably the best NBA player Duke has ever produced (with injury being the only thing standing in the way of that being certain). Guys like Brian Davis and Antonio Lang also had brief NBA careers. That's a pretty stout 6, with both top-end talent and quality depth. And that ignores Thomas Hill, a strong role player who got drafted but didn't play in the league.

As a group, the 1992 team compares favorably with the 1999 team in terms of NBA careers. Brand is arguably #2 to Hill's #1, with injury again doing him in way too early. Maggette is on the short list of most productive NBAers from Duke too, probably in the same neighborhood as Laettner in the pros. Trajan Langdon and Shane Battier were terrific role players (though Langdon decided to leave the NBA for more money in Russia). The 1999 squad would be comparable at the top, maybe a tad better in the middle, but not quite as deep.

Would be a fun matchup. Certainly hard to compare across years, in part because the ACC sort of stunk in 1999 but was quite good in 1992, so the margins were a bit inflated for 1999 and deflated for 1992 (especially when you account for Hurley's injury).

[Troublemaker]

Yeah, probably has a lot to do with it. Still, the '92 team was not near the top in pace, and the '91 and even '90 teams were significantly better in raw Defensive Efficiency. Just a little surprising, as '92 (and '99, of course) had reached mythical status in my brain. Maybe they just didn't blow teams out on the regular.


The 1992 team is a conundrum from an advanced analytics standpoint. In my mind, the team was truly dominant, but the advanced metric indicate that they were "merely" very good. For example, the 92 team finished the year #3 in SRS (https://www.sports-reference.com/cbb/seasons/1992-ratings.html), significantly behind Indiana and with essentially the same SRS rating as Kansas, Ohio St, and Arizona. KenPom ratings don't go back that far, but I distinctly remember the 92 team not being ranked #1 in the final Sagarin ratings.

However, one thing to consider about the 1992 team is that Bobby Hurley and Grant Hill were injured for key parts of the season. In fact, Duke's only losses came without Hurley (Wake) or with Hurley playing on a broken foot (UNC). Duke's less than stellar 1 point win vs a mediocre Clemson team came without Grant.

The 92 team wasn't particularly deep, so I expect that these injuries play a role in the team's less-than-historic advanced metrics. I wonder what the advanced metrics would look like in games where Hurley and Hill were 100% healthy.


And Cherokee Parks also was injured.

I still think 92 was a great team, but it sure didn't roll thru the NCAA tournament.

In addition to the injuries, institutional program fatigue was probably a factor. '92 had made four Final Fours in a row (and would obviously end up making it 5 in a row), and the quest for back-to-back titles had make Duke THE program in the country for the first time. They were rock stars, adored by suburban teenage girls nationwide, and the program had never dealt with that kind of intense national interest and scrutiny before. (It's old hat for us now, in 2018). This was also pre-back surgery Coach K, so he hadn't learned the value of not being everything to everyone yet, a lesson that would benefit both himself and the players he coached in the future.

As a result, Duke in '92 was sort of like an NBA team that was a veteran of playoff wars, a champion, and couldn't quite summon the energy needed to dominate in the regular season game after game. Think about how often the Warriors this season or the Lebron Cavs would underperform their talent in the regular season. Duke '92 was sort of like that. Sort of like that, not exactly that. I mean, we were much better than the Warriors and Cavs -- shoot, '92 Duke finished the season 34-2 afterall -- but opponents that we could've blown out, we settled for keeping at arm's length. And some of those opponents broke through the arm's length and made it close games every now and then. But '92 Duke was clutch. 14-2 in games decided by 10 points or less (not counting the massacre of St John's that only ended up landing on 10 by the end), 5-1 in games decided by 3 points or less. Which helped maintained our sparkling record.

Ultimately, Duke '92 was a team that summoned their best play only when they really needed it or wanted it. The ACC championship blowout of UNC. That 2nd half run against Indiana in the Final Four. The second half of the national title game against Michigan. All of those were displays of beautiful basketball on both ends of the court.

No matter what the stats say, there's no way I take another Duke team over '92.

[Kedsy]

Thanks for this great work. Any chance you can post the historical stats for team oRating - dRating? Seems like that would have a very high correlation with success.

It's the middle table (unadjusted) in post #30, back on page 2 of this thread, or the bottom table (adjusted) in the same post.

[Kedsy]

However, one thing to consider about the 1992 team is that Bobby Hurley and Grant Hill were injured for key parts of the season. In fact, Duke's only losses came without Hurley (Wake) or with Hurley playing on a broken foot (UNC). Duke's less than stellar 1 point win vs a mediocre Clemson team came without Grant.

The main thing that made the 1992 team less than stellar statistically was the fact that our opponents shot so well against them. The team's season-long eFG% of 50.5% was Duke's worst ever (or at least worst since 1987). While it's true the team's D was even worse when either Bobby H or Grant (or both) were injured -- from the first UNC game (where Hurley got hurt) through the 2nd Clemson game (last game Grant missed), Duke's opposing eFG% was 53.0%, compared to the opposing eFG% of 49.8% in Duke's other games. But 49.8% would still be Duke's worst ever opposing eFG%, so it really wasn't worse enough while Bobby and/or Grant were out to explain away the problem.

I guess Troublemaker's "they were only trying hard enough to win" explanation might be the right one, but it's also possible that they just weren't that great a defensive team.

[BandAlum83]

Obviously, if you want pace-independent statistics, you need to be able to measure pace, which is generally measured by possessions per game. A possession ends one of three ways: (1) you shoot the ball; (2) you turn the ball over; or (3) you get fouled and take free throws. If you miss a shot (or free throw) and get the offensive rebound, it's not really a new possession, but rather a continuation of the same possession, even if it results in another shot, turnover, etc.

If every free throw was successful (and there were no and-ones) then the number of possessions represented by free throws would be (0.5)*FTA. Since neither of those assumptions is true, we have to come up with a multiplier other than (0.5). Most accurate would be to count these things individually from the play-by-play, but nobody on Earth is willing to do that so they've come up with an estimate based on statistical analysis of a sample of games. The commonly accepted college ball estimate is (0.475). Interestingly, the commonly accepted NBA estimate is (0.44); I'm not sure why, maybe there are more and-ones in the NBA?

In any event, the formula for possessions per game is: (FGA + (0.475*FTA) + TOs - ORs)/gms. The higher this number is, the faster you play.

Here's a table with Duke's possessions per game in every year since 1987 (which is when they first started tracking offensive rebounds, making this calculation possible). As you can see, for Duke, fast is generally good, as six of our fastest 7 teams made the Final Four. But slow isn't necessarily bad, as two of our five slowest teams won the national championship (though it's also worth noting that the other three of our slowest five teams got knocked out in the first round). If this year's team keeps up its currrent pace, we'll have the 5th-fastest team in "modern" Duke history.



Year Pace NCAA
1990 79.34 2
1989 78.91 4
1991 78.53 1
2002 77.19 16
2019 76.74
2001 76.65 1
1999 75.39 2
1988 75.02 4
1993 75.01 32
2000 74.82 16
2008 73.98 32
1992 73.67 1
1998 73.44 8
2003 72.96 16
1987 72.71 16
2006 71.55 16
2018 70.65 8
1994 70.28 2
2011 70.25 16
1995 70.18 n/a
2005 70.13 16
2004 69.95 4
1997 69.94 32
2009 69.61 16
2017 69.59 32
1996 68.98 64
2016 68.29 16
2013 68.25 8
2012 68.16 64
2010 66.55 1
2007 66.38 64
2015 65.78 1
2014 65.76 64


Here's a table showing context for the last five NCAA seasons. The reason there's a big jump in pace between 2015 and 2016 is that's when the shot clock was lowered from 35 seconds to 30 seconds. Apparently that change was worth about 4 possessions per game nationally. This year, it seems NCAA teams are running faster than in the past, but that may simply be a function of all games being November/December non-conference games.



Year 1 10 50 100 150 200 250 300 Last
2019 88.0 79.5 74.8 73.1 71.6 70.5 69.2 68.1 61.6
2018 83.6 75.6 72.4 71.0 70.0 69.2 68.0 66.9 60.6
2017 83.3 75.5 72.6 71.1 70.1 69.3 68.2 66.7 60.3
2016 81.2 74.8 72.2 70.7 69.7 68.7 67.7 66.4 61.5
2015 77.5 70.6 68.1 66.6 65.5 64.6 63.8 62.5 57.1


I have a question, professor:

It would seem, by definition, each team should have the same number of possessions at the end of the game, or be within one possession of each other. Is this true, or is there something I might be missing?

BTW, million thanks for doing this thread!

[Kedsy]

I have a question, professor:

It would seem, by definition, each team should have the same number of possessions at the end of the game, or be within one possession of each other. Is this true, or is there something I might be missing?

BTW, million thanks for doing this thread!

Technically, they should be within two (if, for instance, the same team has both the first and last possessions in both halves). Since the calculation is an estimate, however (due to free throws: we use .475*FTA because that's the closest estimator in general but in a particular game it could be off), the two teams' possession stats could differ by a little more (so far this season, Duke has had three games where the calculations differ by 3+, with a largest difference of 3.37). For a one game possession calculation, you're supposed to calculate possessions for each team and then take the average, which I do when I give the game stats. I imagine you could do the same thing for season stats, but I'm not sure if anybody bothers to do that (figuring over the course of 30 to 40 games it will all average out). FWIW, I did not average team and opponent possessions for the season-long data in this thread.

[Troublemaker]

I guess Troublemaker's "they were only trying hard enough to win" explanation might be the right one, but it's also possible that they just weren't that great a defensive team.

In either case, the data would look the same. That's what makes it difficult to decide which is the right answer.

Ultimately, I think you just have to ask yourself if a team with Bobby at PG and GHill, Brian, and THill covering the 4 thru 2 on the wings was a great defensive team. I think they absolutely were, and they displayed it when it was really needed.

Very nice thread and explanations, btw. Thanks.