CDu
02-19-2016, 01:37 PM
We've talked ad nauseum about the inconsistency between Kennard's 3pt shooting reputation and (so far) his results. The general consensus seems to be that he's inconsistent. What I find interesting, though, is that his inconsistency is starting to seem somewhat consistent.
For the season, Kennard is shooting 34% (40/118). That obviously excludes 2 preseason games in which he shot 8-18. Now, for the purposes of this discussion, I'm going to include those data. My rationale is this: I don't think most of Kennard's misses are due to defensive pressure. He has been missing (often just barely) open 3s. Thus, I see no reason to exclude data points against vastly inferior competition. Further support for this is as follows: (1) some of his worst shooting performances have come against some of our weakest opponents (1-5 against Bryant; 1-7 against Ga Southern; 1-5 against BC; 1-4 against LBSU); and (2) his 3pt shooting in ACC play has been pretty good overall (22/56, or 39.2%). So for the purposes of this discussion, I posit that the quality of opponent has had very little (if not nothing) to do with Kennard's shooting performance.
With that hypothesis explained and hopefully well-supported, I'll go to the numbers. Including all 40-minute games we've played (regular and exhibition), Kennard is 48/136 (35.2%). That is 40/118 in regular season games and 8/18 in preseason exhibitions (he shot 2-7 against FL Southern and 6-11 against Livingstone). If we divide those 136 attempts into 3 relatively equal segments, we see the following:
1. 16/46 (35.2%) through the Utah State game
2. 14/45 (31.1%) through the Clemson game
3. 18/45 (40%) the rest of the way
Now, I see two important things to note here. First, we're still dealing with very small sample sizes: 2 more makes in that middle segment and he's shooting 35.6% for that stretch; 2 more makes in the first stretch and he's shooting 39.1%; 2 more misses in that last stretch and he's back at 35%. He's 4 makes overall from being a 37% shooter on the season, or 4 misses from being a 31% shooter. Second, despite his wildly inconsistent shooting from game to game, there is an oddly consistent feel to his shooting results when segmented this way. Some of the 45 or 46 attempts result in awful games (like the 0-7 against Syracuse). Others will be magnificent (like the 6-11 against State, 3-4 vs UNC; 4-5 vs Notre dame; 4-6 vs Utah State; 6-11 against Livingstone). But on average, he's been a roughly 35% shooter this season.
If we compare this 35% to the "priors" that Kennard was an elite shooter (supported by his high school numbers, his FT%, and Coach K's willingness to let him continue to be a high-volume shooter throughout the year), the 35% corresponds to our priors moreso than the ~28-30% he was shooting at other points this season. So it could be that he really has been just a bit unfortunate thus far, and that his percentage is still low. For example, he's still just 4-5 makes shy of being a 37-38% 3pt shooter. Given that he's had quite a few shots rattle in and out this year, I think that's a real possible explanation of the data we've seen.
There are, of course, other possibilities. I can certainly accept the possibility that the 1+ foot increase in distance of the 3pt line could potentially bring his numbers down from the 40% percent he shot in high school to around 35% as a freshman. I could also buy that the reduction in touches compared to his high school days and the increase in defensive intensity required to play at Duke are driving his numbers down. But it could also be that we're still just looking at too small a sample, and that his results are being driven down by just a few too many unlucky bounces.
I guess whether you fall in the former camp or the latter camp may depend on whether you have a Bayesian or frequentist view of the world. That's a separate discussion altogether. I just found the data to be really interesting and a little surprising given how up and down he's been this year from 3.
For the season, Kennard is shooting 34% (40/118). That obviously excludes 2 preseason games in which he shot 8-18. Now, for the purposes of this discussion, I'm going to include those data. My rationale is this: I don't think most of Kennard's misses are due to defensive pressure. He has been missing (often just barely) open 3s. Thus, I see no reason to exclude data points against vastly inferior competition. Further support for this is as follows: (1) some of his worst shooting performances have come against some of our weakest opponents (1-5 against Bryant; 1-7 against Ga Southern; 1-5 against BC; 1-4 against LBSU); and (2) his 3pt shooting in ACC play has been pretty good overall (22/56, or 39.2%). So for the purposes of this discussion, I posit that the quality of opponent has had very little (if not nothing) to do with Kennard's shooting performance.
With that hypothesis explained and hopefully well-supported, I'll go to the numbers. Including all 40-minute games we've played (regular and exhibition), Kennard is 48/136 (35.2%). That is 40/118 in regular season games and 8/18 in preseason exhibitions (he shot 2-7 against FL Southern and 6-11 against Livingstone). If we divide those 136 attempts into 3 relatively equal segments, we see the following:
1. 16/46 (35.2%) through the Utah State game
2. 14/45 (31.1%) through the Clemson game
3. 18/45 (40%) the rest of the way
Now, I see two important things to note here. First, we're still dealing with very small sample sizes: 2 more makes in that middle segment and he's shooting 35.6% for that stretch; 2 more makes in the first stretch and he's shooting 39.1%; 2 more misses in that last stretch and he's back at 35%. He's 4 makes overall from being a 37% shooter on the season, or 4 misses from being a 31% shooter. Second, despite his wildly inconsistent shooting from game to game, there is an oddly consistent feel to his shooting results when segmented this way. Some of the 45 or 46 attempts result in awful games (like the 0-7 against Syracuse). Others will be magnificent (like the 6-11 against State, 3-4 vs UNC; 4-5 vs Notre dame; 4-6 vs Utah State; 6-11 against Livingstone). But on average, he's been a roughly 35% shooter this season.
If we compare this 35% to the "priors" that Kennard was an elite shooter (supported by his high school numbers, his FT%, and Coach K's willingness to let him continue to be a high-volume shooter throughout the year), the 35% corresponds to our priors moreso than the ~28-30% he was shooting at other points this season. So it could be that he really has been just a bit unfortunate thus far, and that his percentage is still low. For example, he's still just 4-5 makes shy of being a 37-38% 3pt shooter. Given that he's had quite a few shots rattle in and out this year, I think that's a real possible explanation of the data we've seen.
There are, of course, other possibilities. I can certainly accept the possibility that the 1+ foot increase in distance of the 3pt line could potentially bring his numbers down from the 40% percent he shot in high school to around 35% as a freshman. I could also buy that the reduction in touches compared to his high school days and the increase in defensive intensity required to play at Duke are driving his numbers down. But it could also be that we're still just looking at too small a sample, and that his results are being driven down by just a few too many unlucky bounces.
I guess whether you fall in the former camp or the latter camp may depend on whether you have a Bayesian or frequentist view of the world. That's a separate discussion altogether. I just found the data to be really interesting and a little surprising given how up and down he's been this year from 3.