basketball defense

[Patrick Barron]

"Material possessions wear out and become a burden. Possessing the ball endures."

--Craig Ross (via Leon Blum)

At media day in 2018 Luke Yaklich said (something very close to) “what gets measured can get accomplished.” Looking at Michigan’s surprising (to me) 30-7 season, I wonder if a corollary might be even more salient; that “what isn’t measured, or isn’t measurable, isn’t appreciated.”

Michigan’s terrific season found the Wolverines to be pretty average in all elements of shooting the ball. Number 24 in offensive efficiency is hardly terrible, but the Wolverines were (slightly) below average from the line (70.1% versus 70.7%), slightly below average from 3 (34.2% versus 34.4%) and slightly above average from inside the line (51.8% versus 50.1%). These numbers are OK, but not this-team-is-winning 30 of its games OK. So, what’s up?

The answer, of course, is defense and the homily that “defense travels.” Shooting comes and goes, but the thinking is that defensive ability can remain within more narrow limits over a number of trials. Kenpom tracks defensive efficiency and Michigan (after clocking in at # 3 in 2018) was in the top five all year on this measure, finishing at # 2 (TTU, of course, was #1) and flirted with #1 over the last couple of months of the season.

Defense is a team game, of course, but I have had the luxury of sitting close to the floor and the trio of X, Matthews and Teske, was well, I dunno, enthralling for a basketball junkie. I watched Ricky Green and Gary Grant for their entire careers and I never thought I would see better defenders at the point in my lifetime of observing UM. I was wrong. X is a magician on D.

Charles Mathews has been a unique defender since he has the ability to erase a shooting guard or forward. He can cover a point. He won’t back down from a big man on an unhappy switch. From a defensive perspective, the only Michigan forward who matches Charles was Wayman Britt (Britt is the best UM defender I have ever seen) who could (and did) demolish Adrian Dantley (Britt was 6’ 2”) or switch and check a point with ease.

For the record, John Beilein says “Simpson and Matthews are the most driven defensive players I have ever coached.” By rumor, Beilein has coached for a long time.

Chris Webber was a phenomenal defender in his years at UM (let me get back to that) and Steve Fisher used to intentionally funnel ball carriers to the baseline due to Webber’s unusual ability to eliminate any ordinary advantage. Indeed, it was death to most possessions to try to beat those teams to the baseline. Of course, Webber was playing with Juwan Howard, a solid 6’ 10” quasi-post player who could help cover a player disengaged by Webber. The Fabs could play offense, but their defensive abilities weren’t far behind, if they were at all. Rose could defend. Jackson and King could defend. And, at the time, fouls were often called for breathing hard.

Taking Webber out of the comparative equation, I have been stunned by Jon Teske’s ability to play defense. It isn’t just me. When Mark Turgeon stepped to the podium after his impressive team’s loss in Ann Arbor, his first words were “Jon Teske” and how he limited Maryland in their inside game. “Teske,” Turgeon said, “was everywhere.” This is, I think, hard to appreciate if you are watching a game from distance or not watching in slow motion. Teske’s ability to cover the pick and roll---allowing the original defender to catch up---his ability to hedge off ball screens and still defend a big man in the paint, is at a level I have never seen at the NCAA; or maybe any level.

WHAT STATS WE HAVE

But this is pure holism, I concede. It is just an opinion, though I was gratified to see that Mark Turgeon shared it. So I began to wonder if there was some way to measure at least some elements of defensive ability and I came up with the construct of “Extra Possessions Created.” [I can’t swear someone else didn’t think about it before or isn’t tracking it comprehensively now.] I don’t suggest this is the entire way of measuring player value. Not even close. I don’t even claim it is a complete measure of defensive ability. It isn’t. But I think it may hold some kernel of “objectivity” concerning the “defensive” value of a player.

Start with a simple example. Team A shoots 47% as a matter of effective field goal percentage. This would be a dreadful number, what Northwestern did in 2018 at # 327 in the country. Assume they are playing Georgia, a team at 50.7 % and # 178, a very average number. HoeHow How many more shots does Northwestern have to take to balance the books? Assume an average number of shots per game (shot equivalents) is 65. I haven’t evaluated this, but I did run a few teams to see if this was close. If NU takes 65 shots and Georgia takes 65 shots NU loses by 5 points, 66 to 61. But what if NU can grind out 5 more shots? The game is now a push. Would we ordinarily consider 5 shots to be a big deal? Well, it seems like it might be.

This idea led me to consider the value of a “defensive” player in the number of “Excess Possessions Created.” This might be:

ORBs+Steals+ (%*Blocks)-TOs=EPC

Offensive rebounds seems obvious. You get an extra board? You just created an extra possession for your team. Same for a steal. Blocks are a bit more problematic, since not all blocks will lead to another possession. Sometimes the ball goes off the defender. Sometimes the ball lands in the hands of an offensive player. Originally I guessed about 1/3 of blocks resulted in an extra possession for the blocking team. But this has been studied in the NBA (“Kill Rate”) and it seems, at least with the pros, the percentage is between .5 and .68. For the moment my hypothesis is to work with .55, knowing this may be a placeholder and knowing that there is at least one other problem.

The problem is the (theoretical) “intelligence” of the guy blocking the shot. We have all seen players who are more interested in style points, the spectacular play. A ball being blocked about 5 rows into the stands is impressive. But less impressive blocks, those that fall into the hands of a teammate, are a lot more functional. Could it be that some players have a greater capacity to keep balls in play? Anthony Davis, in one study (50 blocks) had a 66% kill rate. Draymond Green? 50%. Andre Iguodala? 80%. My guess is you might have guessed this. On the other hand, the sample sizes are low. The macro data, averaging it all out, is around 55%.

Another problem is the question of TOs since TOs are of different quality. A charge, an illegal screen, or a ball thrown into the crowd isn’t as bad as being picked at mid-court. One results in an ordinary extra possession. The other is likely to lead to an easy run out. This leads to another question, whether it is logical to weight TOs and Steals equally. Since a steal is more likely to lead to a run out than a (say) charge, weighting this equally seems problematic. I am going to guess here, that a steal is about 10% better than a TO. With the caveat that this is a very preliminary stab, here’s my first idea for the construct for a “defensive” valuator of Excess Possessions Created. It is:

ORBs+ (1.1 * Steals) +(.55 * Blocks)-TOs=EPC

Using this, here are the numbers applied to some centers. Why centers? Good question. Partly because I mouthed off on radio and said, excepting Webber and Howard, Teske was the best center in UM history. Partly because I think the primary value of most centers is defensive. They have to be (a) the primary rim protector and (b) be able to get out on balls screens and make sure the offense doesn’t get easy looks off the pick and roll. Also, since most centers are not primary scorers, their value accrues in giving the offense the greatest number of opportunities. That’s my theory at least, with the admission that guys like Kareem or Olajuwan or even Ethan Happ have a value that doesn’t come close to being expressed in such an equation. Note that “Ad.EPC” normalizes to 1000 minutes.

Player OREBs Steals (1.1) Blocks (.55) TOs EPC
1 Jon Teske 73 30 41   126
2 Ethan Happ 77 42 24 104 39
3 Terry Mills[1] 81 41 15 75 62
4 Terry Mills[2] 71 22 27 77 43
5 Roy Tarpley[3] 105 31 36 87 85
6 Greg Oden[4] 113 20 58 64 127

Teske’s EPC amount to 3.4 extra possessions per game. Go back to the Northwestern and Georgia hypothetical to see how consequential this is.

Oden was tremendous in 2006-07. And his offensive numbers (1.63 points per shot compared to Teske’s 1.31 points per shot) are better than Teske’s despite the fact Teske was an above average three point shooter in Big Ten games (36.4%) though flopping out of conference (Teske was 3-22 OOC and Oden never attempted a three.) Holistically, I believe Teske creates more space on offense than Oden ever did, but, yeah, Oden. But Oden was known as a great defender and he was. Maybe as good as any center in the BT in a long time. Teske, however, is not all that far behind.

Webber? Well, let’s throw him into the chart.

Player OREBs Steals (1.1) Blocks (.55) TOs EPC
7 Chris Webber[5] 125[6] 54 50 105 124
8 Chris Webber[7] 117[8] 59 46 95 127

Not sure what to make of this at the moment. A work in progress. However, I will go through box scores to get a better number on C Webb’s offensive rebound numbers. Here’s a scan for all BT centers in the 2018-19 season:

EXTRA POSSESSIONS CREATED, 2018-'19 BIG TEN CENTERS

Rk Player School EPC
1 Jon Teske Michigan 126
2 Xavier Tillman MSU 115
3 Dererk Pardon Northwestern 96
4 JuWann Morgan Indiana 94
5 Myles Johnson Rutgers 81
6 Isaiah Roby Nebraska 80
7 Daniel Oturu Minnesota 81
8 Matt Haarms Purdue 68
9 Mike Watkins Penn State 68
10 Bruno Fernando Maryland 64
11 Kaleb Wesson Ohio State 52
12 Ethan Happ Wisconsin 39
13 Luka Garza Iowa 37
14 Giorgi Bezhanishvili Illinois 36
15 Nick Ward MSU 26

At the moment this is just the start of an idea of some creating some “objective” way of evaluating a center’s (at least) “defensive value,” with ORB added into “defense.” I will be curious about others impressions of the matter.

-----------------------------------------------------

Notes & Errata

  1. Senior Year
  2. Junior Season; ORBs estimated from his Junior Year
  3. Player of the Year in the BT. Estimated ORBs at .33.
  4. #1 Player in the NBA draft.
  5. Year 2.
  6. Estimated from 362 total rebounds
  7. Year 1
  8. Estimated
  9. No surprise to Brian Cook, on the Tillman train for some time.
now playing CB for UNC

YOU HAVE FORGOTTEN SOMETHING IMPORTANT FOR A SPECIAL PERSON: a sponsor note. Need a last-second gift? How about an actual piece of the Big House? Friend of the blog Martin Vloet got his hands on the original redwood Big House benches—the ones installed in 1927 and used until 2005—and had them made into limited edition pens, cufflinks, pendants, and bottle openers. He also claimed the old plastic seats and cut them up into magnets or pendants. The first 99 pens are reserved for Michigan football players, past or present, that want like to claim their jersey number. The rest of the pens will ship, #100 through #1927, on a first-ordered basis.

VictorsPen-Box(1024wp)

Use the code MGBFREESHIP and save on domestic shipping of any size order placed by 11:59 pm TODAY. As long as it goes out tomorrow, USPS Priority Mail should be able to make it to any US address by December 24.

Follow this man. Eric Shap on Michigan's defensive issues in their last two outings:

A combination of a December lull against teams that don't really have Michigan's attention and a reversion seemingly well past the mean; if holding Eric Paschall to 3/13 from two without doubling wasn't a realistic picture of Michigan's D, well neither is that last set of clips above.

If NET's taken as seriously as RPI that's fine. Weird article in the Washington Post trashing the NET rankings, which are wonky as any NCAA hodgepodge is going to be but hardly a disaster waiting to happen for tournament seeding. The article has three wrong premises. One is that NET is the be-all and end-all of selection and seeding:

You might not think such a discrepancy in the rankings would mean much, but consider how this could affect the NCAA tournament, where a team like Texas Tech would be given a No. 1 seed via its NET ranking, but plays more like a No. 3 seed, per its consensus ranking.

The committee still exists. We're still talking about quadrant one wins. There are still teamsheets. NET will be followed no more blindly than RPI was. Which was a little blindly, if we're being honest, but not to the point where a team gets a one seed solely because of a single number on the sheet.

Two is that a hodge-podge of computer rankings is an appropriate comparison point. Many, if not most, of the rankings in the giant compilation the author cites are predictive rankings that are inappropriate for selecting and seeding the field. At this point in the season many still have a significant preseason component—Kenpom won't be fully preseason-free until the end of January. If the season ended today a field selected and seeded by Kenpom alone would give Purdue, which is 6-5 and has just two B-level wins, a five seed. NET ranks Purdue 31st instead of 17th. NET's deviation from the average here is a positive. The article cites Houston's NET ranking (10th) vs their computer composite (23rd), but you could cherry-pick a weird outlier for almost every one of these ranking systems. ESPN's BPI has Michigan 11th.

Three is that NET won't be able to better distinguish between teams given an additional half-season of data. This is an absurd comparison to make:

Based on last year’s consensus rankings, a top-four consensus team had an average RPI ranking of 3.3. This year the average NET ranking of a top-four team is 5.5, almost identical to a team ranked between No. 5 and No. 8 in the consensus group. In other words, the NET rankings are incapable of distinguishing between a No. 1 or No. 2 seed in the NCAA tournament, a stark contrast compared to last year where, via RPI, there was a clear difference between the two.

For one, last year's RPI-influenced committee put Kenpom #9 Kansas and Kenpom #14(!!!) Xavier on the one line. As a group the two-seeds were stronger. For two, most teams have only played a third of their games so far. Of course there is going to be more disagreement amongst ranking systems when they have less data.

The only real question is "is NET better than RPI when tourney time nears?" Open question, but it would have to try real hard to be worse.

[After THE JUMP: more NCAA legal troubles, what is USC even doing, and a sudden 180]

something something water [Patrick Barron]

Sometimes I don't have a column. In these times: just bullets.

12/8/2018 – Michigan 89, South Carolina 78 – 10-0

A sloppy outing ends in an eleven point win as Michigan is clearly better, so that's nice. Items:

The un-Michigan game. Michigan 1) rebounded almost half their misses, 2) turned it over 16 times, 3) made 77% of their free throws, and allowed the bricklayers on the other team to shoot 53% from two. This was very un-Dude. Certain things did make sense in the recent history of Michigan basketball. South Carolina got up just 11 threes, didn't go to the line much, and Michigan burned the nets themselves.

The New York Football Knicks. Man, when South Carolina commits a foul it is not subtle. Iggy got flying shoulder thumps immediately preceding most of his FTAs. Michigan hit the bonus with about 12 minutes left in the second half and that felt late. In the first half I kept looking up at the scoreboard to see which South Carolina player had just earned an autobenching only to find out that Martin was rotating his guys so thoroughly that none of the hacking removed key players for long stretches. Both Alanzo Frink and Felipe Haase had three fouls in about ten minutes. Silva fouled out late; five other guys had two fouls.

Jon Teske got some back when he trucked Silva:

46182875802_da11d0fbf0_k

[Barron]

That was deemed a charge; at least Teske got his money's worth.

Illuminati Charles Matthews. Matthews was 2/4 from two, 2/4 from three, 2/4 from the line, and had 2 DREBs and 4 OREBs. Spooooooky.

The defense slowly comes back to earth. Michigan's early two-point D was always going to be unsustainable and things are creeping back up. The good news is that folks are still posting items about Michigan's defense in relation to some of the best in the last decade of college basketball; the bad news is that Michigan's regressing back towards the mean faster than those defenses:

All of those teams had giant block rates. The only reason all three didn't finish first is that 2015 Texas (20%!) beat out 2015 Kentucky (18.2). Last year's MSU (18.5) team had Jaren Jackson (sometimes) and also finished first. Michigan is currently 115th—not bad, but not a number that is going to see you finish the season with a historically good two-point D.

Even if the absurd two-point D was a bit of a mirage, Michigan's D is still very very good in a sustainable fashion. They're forcing more "other twos" than anyone other than San Diego. They're 9th in the country at preventing threes from getting up, and are still top 50 at preventing the opponent from getting to the rim. Teske may not be a super elite shotblocking center but he's also averaging just 3.3 fouls/40, which is a major part of Michigan's #1 ranking in free throw rate allowed.

Michigan is going to be a very good defense. They probably aren't going to be #1 at the end of the year. If they are it's going to be because they're pretty good at two point D and funnel everyone inside the line.

[After THE JUMP: on the other hand this is still a Beilein team.]