Hoop Math: Transition Off Defensive Boards, Updating Kobe Assists
John Beilein likes to say that the best defensive rebound is one by his point guard. Why? That's the best way to get out in transition. I decided to investigate Beilein's claim—at least as it applies to Michigan—by going through this season's play-by-plays and charting each defensive rebound.
In the (chart?) chart below, I've tracked each defensive rebound as well as any resulting fast break field goal attempts or drawn shooting fouls—a fast break, in this case, being defined as any shot coming within 10 seconds of the defensive rebound, so long as the ball remained in play the whole time. Also in the chart is how often each player gets a fast break assist or made basket off their own defensive rebound. "% Opp" is the percentage of individual defensive rebounds that result in fast break field goal attempts or drawn shooting fouls, and "% Conv" is the percentage of made fast break FGA and shooting fouls drawn.
SPOILER ALERT: Beilein's theory is correct.
PLAYER | Def. Reb. | FB FGA | FB FGM | Assist | Self Make | FT | % Opp | % Conv |
---|---|---|---|---|---|---|---|---|
Hardaway | 76 | 33 | 16 | 7 | 7 | 3 | 47.4 | 25.0 |
Morgan | 56 | 20 | 14 | 0 | 2 | 1 | 37.5 | 26.8 |
Robinson | 56 | 18 | 12 | 3 | 0 | 0 | 32.1 | 21.4 |
McGary | 54 | 25 | 14 | 3 | 0 | 1 | 48.1 | 27.8 |
Stauskas | 49 | 25 | 16 | 1 | 1 | 1 | 53.1 | 34.7 |
Burke | 42 | 23 | 11 | 3 | 7 | 2 | 59.5 | 28.6 |
Vogrich | 13 | 5 | 5 | 0 | 1 | 0 | 46.2 | 46.2 |
Horford | 13 | 2 | 1 | 0 | 0 | 0 | 15.4 | 7.7 |
Bielfeldt | 12 | 6 | 2 | 0 | 0 | 0 | 50.0 | 16.7 |
Albrecht | 10 | 3 | 0 | 0 | 0 | 0 | 30.0 | 0 |
LeVert | 7 | 3 | 0 | 0 | 0 | 0 | 42.9 | 0 |
Akunne | 6 | 2 | 1 | 1 | 0 | 0 | 33.3 | 16.7 |
McLimans | 5 | 1 | 0 | 0 | 0 | 0 | 20.0 | 0 |
Person | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
TOTAL | 400 | 166 | 92 | 18 | 18 | 8 | 43.5 | 25.0 |
Some takeaways:
GOOD PLAYERS ARE GOOD
Trey Burke is far and away the best on the team at turning defensive rebounds into transition opportunties, and the reasons are two-fold. For one, it's Trey Burke—you know, the guy you want running the fast break. Second, as you can see in the video at the top of the post, as a diminutive point guard many of Burke's rebounds come on shots that carom far away from the basket, providing a better chance to turn and run than a rebound in the charge circle.
Burke is also the best at converting his own rebound at the other end, with—surprise!—Tim Hardaway Jr. second in that regard; both have seven made baskets off their own rebounds while Hardaway has one more free throw opportunity... off 34 more defensive boards. Though Burke converts at a higher rate, Hardaway has the highest defensive rebound rate on the team by a non-center, and you can see just how valuable his newfound dedication to that area is to the team.
MCGARY'S OUTLET PASSING
Jordan Morgan (and, in small sample size territory, Jon Horford) has a rate well below the team average when it comes to turning defensive rebounds into transition opportunities, which is understandable: as a center, he's not turning and leading the break, and most of his boards come from right under the basket, where it's hardest to spark a transition opportunity.
That makes McGary's ability to turn 48.1% of his defensive rebounds into fast break chances—a better rate than Hardaway—all the more impressive. The difference, as far as I can tell, is in McGary's outlet passing; he's got surprisingly good court vision, which allows him to turn quickly off a rebound and find his point guard. This is one area where McGary has a decided edge on Morgan, especially since his defensive rebound rate is also higher.
GAP BETWEEN FRESHMEN: NOT THE ONE YOU'D EXPECT
What surprised me most when putting this together was the gap between Nik Stauskas (53.1% Opp) and Glenn Robinson III (32.1%). While Robinson matches up against bigger players, ending up closer to the hoop for rebound opportunities, he's also the more athletic of the two. It's Stauskas, however, who's the only player besides Burke to crack 50% in major minutes—this despite rarely being involved in the play at the other end of the floor.
Perhaps there's a lot of noise in these numbers given the sample size (I'd say yes—I'm mostly ignoring the "% Conv" figure because of this) but that doesn't entirely explain that large a gap. Like with the big men, I believe this has to do with the difference in court vision and passing ability; so far this season, Stauskas has proven himself the more adept passer. Meanwhile, Robinson still seems to be adjusting to the college game; in a year, I'd bet his transition rate will be better than Jordan Morgan's.
[Hit THE JUMP for an update on the Kobe Assist and Adjusted Points Per Shot numbers from last month.]
Revisiting Kobe Assists/Adj. PPS
Last month, I took a look at the concept of "Kobe Assists"—a missed shot that immediately results in a converted offensive rebound—and used that to come up with a stat called "adjusted points per shot". For further background, check out last month's post. I wanted to update the numbers now that conference play in under way. First, the shot chart (KA=Kobe Assist):
Player | 2PA | 2PM | OReb | KA | 3PA | 3PM | OReb | KA |
---|---|---|---|---|---|---|---|---|
Trey Burke | 153 | 87 | 25 | 11 | 85 | 34 | 17 | 2 |
Tim Hardaway | 114 | 63 | 17 | 4 | 84 | 33 | 20 | 3 |
Nik Stauskas | 49 | 24 | 10 | 6 | 95 | 47 | 13 | 3 |
Glenn Robinson III | 106 | 68 | 8 | 0 | 36 | 14 | 6 | 1 |
Jordan Morgan | 85 | 54 | 11 | 4 | 0 | 0 | 0 | 0 |
Mitch McGary | 74 | 45 | 12 | 0 | 0 | 0 | 0 | 0 |
Spike Albrecht | 9 | 4 | 3 | 2 | 16 | 6 | 4 | 0 |
Jon Horford | 15 | 7 | 6 | 1 | 0 | 0 | 0 | 0 |
Caris LeVert | 18 | 6 | 4 | 1 | 21 | 9 | 4 | 1 |
TEAM | 623 | 358 | 96 | 29 | 337 | 143 | 64 | 10 |
Standing out here, as usual, is Trey Burke; not only is Burke lethal inside the arc (56.9 2P%), but Michigan rebounds 37.9% of his two-point misses. In other words, on 73.2% of Burke's two-pointers, he either scores or the Wolverines retain possession. That figure is 60.0% for his three-pointers. Again, Trey Burke is good. You know this.
In fact, let's calculate those make/retain rates for Michigan's main rotation (players with >50 FGA):
Player | 2P Make/Retain % | 3P Make/Retain % |
---|---|---|
Trey Burke | 73.2 | 60.0 |
Tim Hardaway | 70.1 | 63.1 |
Nik Stauskas | 69.4 | 63.2 |
Glenn Robinson III | 71.7 | 55.6 |
Jordan Morgan | 76.5 | -- |
Mitch McGary | 77.0 | -- |
Again, sample size caveats apply, but it's interesting that Burke and Hardaway nearly make up the gap with Stauskas from beyond the arc despite both shooting around 10% worse from three. That doesn't mean a Burke/Hardaway three is as good as one from Stauskas—offensive rebounds are usually converted into twos, not threes—but it suggests that there's something different about the nature of Stauskas's threes. It's not that he's the one rebounding Burke/Hardaway misses—Stauskas has a lower OR% than Burke, in fact.
A potential explanation is how they get their threes. Burke, especially, and Hardaway get a lot of their three-point attempts off the pick-and-roll, which means a big man is rolling to the basket—and prime rebounding territory—as they shoot. Stauskas, meanwhile, is more of a spot-up shooter; the defense isn't scrambling as much when he's lining up his shot and there isn't a big man rolling towards the basket (EDIT: holy timing, UMHoops). That may help explain the disparity, though sample size is also an issue.
Before the adjusted points per shot chart, here's the explanation of the statistic from the original post:
The formula: (2*2FG + 3*3FG + 2*KobeAssists + [team's points per possession][KobePasses - Kobe Assists]) / FGA.
In layman's terms, add a player's points to the points scored directly off his misses and the (admittedly approximate) expected points scored when the team gets a rebound off his miss and doesn't immediately put a shot back up, all divided by his field goal attempts. This gives the expected number of points a team will score on a possession in which a certain player takes a shot.
Changes since December 11th are noted in parentheses:
Player | Adj. PPS (2P) | Adj. PPS (3P) | Adj. PPS (Total) |
---|---|---|---|
Trey Burke | 1.40 (+0.02) | 1.47 (+0.05) | 1.42 (+0.03) |
Tim Hardaway | 1.32 (+0.08) | 1.50 (+0.13) | 1.40 (+0.11) |
Nik Stauskas | 1.33 (+0.09) | 1.68 (-0.34) | 1.56 (-0.12) |
Glenn Robinson III | 1.38 (+0.08) | 1.40 (-0.01) | 1.38 (+0.05) |
Jordan Morgan | 1.47 (--) | - | 1.47 (--) |
Mitch McGary | 1.42 (-0.02) | - | 1.42 (-0.02) |
Spike Albrecht | 1.47 (+0.16) | 1.44 (+0.02) | 1.45 (+0.08) |
Jon Horford | 1.48 (+0.09) | - | 1.48 (+0.09) |
Caris LeVert* | 0.99 (--) | 1.56 (--) | 1.30 (--) |
*Caris LeVert wasn't included in last month's chart.
Notably, Stauskas has come back to earth with his three-point shooting; despite the drop, a three-point attempt by him is still easily the best shot this team can create, while his efficiency inside the arc has improved. Hardaway has rounded into form in recent weeks, while Burke has maintained an impressive pace. Caris LeVert, whom I didn't even bother charting the last time around, has great potential from outside, but needs to work on his shot selection from two-point range.
January 23rd, 2013 at 2:20 PM ^
who's seeing significant minutes gives an adjusted point per initial possession value greater than 1.30?! Am I reading this right? I guess that's how you end up with the #1 offense in the country. I have no way to put these numbers in context (what would they look like for Duke, or tOSU, or Northwestern, or Western Michigan?), but they seem incredible. All hail Beilein!
January 23rd, 2013 at 2:26 PM ^
but Spike looks to have a nice Adj. PPS (2P). If he can be one splitting minutes with Derrick next year and hover around that rate we should be in good shape. We should still have GRIII, Morgan, Mitch, and Stauskas so there will be plenty of rebounding/point generating power to pick up the misses. I'm excited about the other incoming freshman as well.
January 23rd, 2013 at 3:14 PM ^
Between your post and the one at UMHoops, that's some seriously outstanding basketball content today.
One note on the difference in rebound/transition effectiveness between Glenn Robinson, Jr. and Nik Stauskas: dribbling. Stauskas is an excellent ballhandler for a 2/3 (his turnover rate is extremely high, which reflects inexperience and poor decision-making at times, rather than an inability to handle the ball), while Robinson Jr. hasn't really demonstrated much ballhandling yet. It's the biggest reason why I think Robinson Jr. needs at least one more year of college - to demonstrate his ability to handle the ball, given that he'll be a small forward in the NBA. If I recall correctly, in Robinson Jr.'s recent 20-point outing he took *one* dribble total prior to all of his plays that led to points.
The disparity in those two players' ability to dribble is why I wasn't surprised at the results.
January 23rd, 2013 at 8:04 PM ^
Stauskis's TO rate is actually very low -- in the Burke range. Not sure what you're looking at.
January 23rd, 2013 at 9:36 PM ^
I (obviously) didn't check the stats first - thank you for pointing out my error, it was based solely on (poor!) observation.
That does further demonstrate my larger point though, which is that Staukas can really handle the ball, and it is those ballhandling skills that are driving the discrepancy that Ace found.
January 23rd, 2013 at 3:40 PM ^
Just a thought: Does an above average defensive-rebounding PG who transitions those rebounds into points, more often than not, have any affect on gameplan of shot selection for opposing teams? I guess what i'm asking specifically is, going back to Brian's mailbag from earlier, would this be effective in lowering an opponents proclivity to jack 3's
Great work Ace. Very Interesting. Also, Trey Burke is the man.
January 23rd, 2013 at 3:40 PM ^
Question from a layperson: Wouldn't a point guard naturally have the highest Kobe Assist Average given that their shots would normally occur when the big men are under the basket (as opposed to shots by big men which means one less big body on the glass)?
January 23rd, 2013 at 4:26 PM ^
Ideally you'd have a control sample (e.g. the rest of the conference) against which to compare our "Kobe assist" stats (a "difference-in-differences" approach), but in this case that's not possible because the data is too costly to obtain (cost = Ace losing sleep).
January 23rd, 2013 at 6:31 PM ^
I know some of us get annoyed with Trey's shot selection, but even with the position inflation, the Kobe Assist factor can still help us all sleep better knowing there's such a high probability that some frontcourt player will grab Trey's miss a lot of the time
January 23rd, 2013 at 9:03 PM ^
Ace, you owe us an obligatory Wes Unseld reference for the outlet pass discussion.
Comments