ScreenHunter_16_Dec._19_10.37

The defensive value of Bernard James, John Henson, Josh Smith and Anthony Davis

ScreenHunter_12_Dec._19_10.26

Last season John Henson dominated the ACC Defensive Player of the Year voting. He received 66% of the media’s votes while FSU’s Chris Singleton received 22% and no one else even reached 3%. I had watched every FSU game and most UNC games and I felt the voting was backwards. And this I attributed to two things: ACC media bias, and Michael Rogner bias (I graduated from FSU).

Measuring individual defense in a team sport is nearly impossible. Tempo free data gives great insight into how teams perform, but for individuals we are primarily left with block and steal rates, and defensive rebounding. For the record, John Henson blocked 11.6% of his opponent’s shots when he was on the floor, recorded a steal in 1.2% of UNC’s defensive possessions, and grabbed 25.4% of the available defensive rebounds. Chris Singleton blocked half as many shots (5.9%), stole the ball three times as often (3.9%), and grabbed 17.1% of the defensive rebounds. Using those numbers alone, the edge would go to Henson. Of course, very (VERY) few members of the old media use tempo free stats. They cite blocks per game, steals per game, etc… But I’m not here to argue with them.

As I was wrestling with Chris Singleton’s defense and whether or not bias was clouding my ability to watch a game, the coach’s awards came out. In those, all 12 ACC coaches agreed on the Defensive Player of the Year, and all 12 picked Chris Singleton. With my hoops IQ vindicated, I moved on to a new year.

Rather than guess like last season, I wanted something more. So what I’m doing is tracking defense, so that when the voting happens I have some data to back up the opinion formed by watching the games.

The method is simple. Tedious, but simple. I’m tracking every defensive possession for several players, and recording the result. If the other team scores – that’s bad. If there’s a stop – that’s good. The end result is that I know how many defensive possessions these players are in games for, and how many points the other team scores. Then I compare that to how many possessions they spend on the bench, and how the other team performs while they’re out of the game. The results are illuminating.

A quick note about methods. When players are substituted around foul shots the possession is credited to where they were when the foul occurred. If they were on the floor and then get shuffled off before the FTs are made, they still count against that player (and the opposite is true as well). The other common situation is a team missing a shot, grabbing the offensive rebound, and calling a timeout. If the player I’m tracking comes out of the game during that timeout then the possession is removed as there is no sense crediting them (or punishing them) when the point scoring portion of the possession is going to occur with them on the bench.

Anyway, here are four examples:

THE SENIOR

Bernard James, Florida State Seminoles

6’10, 240 lbs

Bernard James isn’t your typical senior. Not only is he a high school dropout, but he never even played basketball while he was in high school. So when Leonard Hamilton spotted him playing in a military tournament he had to extend his imagination to see the potential. He convinced him to enroll at Tallahassee Community College, and then two years later provided the scholarship he’d promised to FSU. James didn’t play the recruiting game at JUCO, and so few people knew who he was. But it didn’t take long for them to figure it out.

In his first season at Florida State he finished with the second highest FG% in school history (65.7%) and the second most blocks (82). He’s already ninth on the career blocked shot list and he’s only played about 1 1/3 seasons.

But none of that matters. What I’m interested in is possessions and points. And through ten games here is what that looks like:

possessions points scored points per possession +/-
in game 456 341 0.75 23.3%
on bench 325 295 0.91

Florida State allows 0.75 points per possession when Bernard James is in the game, and 0.91 when he’s not, a difference of over 23%. What to make of this? First, his defense is not only legit, but it’s elite (as we’ll see when we look at other players). Second, these numbers can’t be compared on a straight 1-to-1 basis. The reason is that is doesn’t calculate a value over his replacement. FSU is a huge team, and when James leaves the floor the Noles often get bigger. Xavier Gibson is typically on the floor (6’11) and Jon Kreft comes off the bench (7’0). And herein lies the major challenge of interpreting individual defense in basketball.

The good news is that lots of people are working on it. The Kentucky blog Sea of Blue is tracking all of the Kentucky players using nearly identical methods that I’m using. And Luke Winn at Sports Illustrated initiated a group that is watching game film and breaking down every possession that way. The problem is that cracking the defensive code is going be a long and tedious process, so the end result will be a private firm collecting the data and selling it to those who can afford it. In other words, SI, Yahoo, CBS, etc… are going to have a corner on the market. So it goes.

THE JUNIOR

John Henson, North Carolina Tar Heels

6’11, 220 lbs

ScreenHunter_13_Dec._19_10.33

Everybody knew John Henson. His father played at Norfolk State and so knew what it took to be a college basketball player. John Henson was a 1st Team Parade All-American in high school. He played in the McDonald’s All-American game, the Jordan All-American Game and the Nike Hoop Summit. His skill set is almost completely unique.

But when Henson got to college he struggled. He clearly wasn’t comfortable in his body and his offensive game was raw. He didn’t even play 40% of his team’s minutes in his freshman season, and had a bad offensive rating (95.3) because he turned the ball over, couldn’t make FTs, and had the bad habit of jacking threes that were considered a moral victory when they hit iron. As a sophomore his offensive game showed plenty of new moves, but little advancement. But he wasn’t in the game for his offense. He was in the game for his defense. And he was named the ACC Defensive Player of the Year.

His junior season is a different story. His turnovers are down. He commits fewer fouls. And his offensive rating is above 111. But what about his defense? His block rate last year was 11.6%, which was 12th in the nation. It was 5th nationally if you limit it to major conference schools (Bernard James was 1st). This year its dipped slightly to 10.1%, good for 40th. His steal rate is almost non-existent (0.3%) and his defensive rebounding is still north of 25% making him one of the best rebounders in the country.

His possession data, through 11 games:

possessions points scored points per possession +/-
in game 593 533 0.90 12.2%
on bench 228 231 1.01

Clearly John Henson has a great effect on UNC’s defense, which is one of the reasons he’s on the floor more than any other Tar Heel besides Kendall Marshall. The numbers may not support a repeat of the DPOY award (which he’ll win) but, when combined with his offensive game, should make him a lock for 1st Team All-ACC.

THE SOPHOMORE

Josh Smith, UCLA Bruins

6’10, 305 lbs

ScreenHunter_15_Dec._19_10.35

Ahhh, Josh Smith. The enigma. Coming out of the Seattle area Smith was a 5* recruit and McDonald’s All-American. He also weighed 350 lbs when he graduated. His official weight was 305 as a freshman (which was probably close) and his official weight as a sophomore is also 305 (which is probably within 15 pounds). And that’s unfortunate, because Smith is one of the most skilled big men in the nation. He has great hands, knows how to score, and is fairly nimble for a man his size. In a perfect world Smith would have busted his ass all summer and come in weighing 295 with the ability to go full steam for 28 or 29 minutes a game. As it is he’s averaging 17 minutes and since the Kansas game (before Thanksgiving) hasn’t been on the floor a single time for more than 11 consecutive possessions. And his team is struggling. They’re 5-5 with one of those wins coming over an NAIA team, and they lost to Loyola Marymount, at home.

Since he’s so big he has trouble giving a proper hedge on ball screens and still getting back to the basket to guard against the roll. That criticism is legit. But a lot of the bashing he takes from announcers isn’t. He has good footwork and solid fundamentals. He just doesn’t have the body to execute.

His numbers:

possessions points scored points per possession +/-
in game 299 290 0.97 1.0%
on bench 351 343 0.98

There’s been a 1% difference for the better when Josh Smith is in the game. For an elite player who has a good grasp of the system, this is no good. He’s only a sophomore, but there are clearly enough red flags that you have to wonder if he’ll ever fulfill his vast potential.

THE FRESHMAN

Anthony Davis, Kentucky Wildcats

6’10, 220 pounds

Davis was an All-Everything recruit out of Chicago. Had he been allowed to go directly to the NBA he would have been a top-3 pick, if not top overall. As a player who’s grown immensely in the past couple years his offensive game is a work in progress. But his athleticism and potential are off the charts.

Through ten games he has the 2nd highest offensive rating on the Wildcats roster – though he uses less than 19% of the possessions. What this means is that his game is basically cleaning up around the basket. He’s the best offensive rebounder on the team (12th in the SEC), and he’s converting 70% of his 2s. His block% is 5th nationally, and he has a solid 3.0% steal rate. But what about Kentucky’s numbers while he’s on the floor?

possessions points scored points per possession +/-
in game 496 398 0.80 8.8%
on bench 223 193 0.87

Davis’s numbers are solid, though below John Henson’s and Bernard James. But beyond that, this illustrates a great challenge in interpreting solo defensive stats. Kentucky has elite players all over the place. Eloy Vargas, Terrence Jones and Darius Miller are all over 6’8. Michael Kidd-Gilchrist is a likely top-10 pick. Considering the value of  his replacements, how do you value that 8.8% increase versus someone who’s replacement player is a legitimate bench guy?

Quantcast