Plus Minus for West Virginia Game

Once again, Wilson stands out in the plus minus but the real standout was McCormack. He was clearly the key player from a plus-minus perspective. It is always a good game when the top 7 players all have positive PM's. Thompson and Harris more than held their own from a +/- standpoint. McCormack had a delta of 1.22, meaning the team outscored West Virginia by 1.22 points more when he was in the game than when he was on the bench. Overall, a very nice game by all.

HE, great question. The answer to the 2nd question is I pick from the top 6 players in minutes. I put no minimum on minutes, but it is rare that the 6th man logs fewer than 15 minutes. This game was a rare case. But it looks at the top 6 in minutes and takes the best number.

I would like to give a little more in-depth explanation of the +/- analytic and how to make sense of it. Many coaches will wave their hands and dismiss the +/- as useful analytic, but in reality, they all use it and look at it. But they are able to decipher the value from it whereas a casual understanding of the analytic can lead to an overemphasis on it.

Think of it like this, there are 4 useful analytics of pieces of useful information that can be derived from the +/-. They are as I have in my table, 1) the overall +/-, 2) the offensive +/-, 3) the defensive +/- and 4) the delta. Each has a value but you have to keep them all in perspective.

1. Overall +/-: Obviously this is the scoring margin for a team during the time a player is in the game. As I've said before, the noise (stochastically speaking) in this metric can be significant for a single game. But over a larger dataset, this metric begins to filter out the noise and become more and more meaningful. Not that it is meaningless for and individual game, you just have to have the ability to filter out the noise, such as at what point was the player in and out of the game and for what reasons and who replaced him and so forth. My go to example in this stat is always Darnell Jackson and Landon Lucas. No one would peg either of these guys as offensive dynamos. Both would be considered very good defensively but not necessarily the lock down flashy defensive stopper. Both ranked a close 3rd on overall +/- in their senior seasons. Darnell ranked above Darrel Arthur and Brandon Rush. The conclusion is that over the entire season, Darnell and Landon did so many things that made the team better when they were on the floor. Darnell was a great rebounder, especially offensively his senior season. Landon, was a very smart offensive player in that he only shot when it was smart to do so and thus made a high percentage of his shots even though he would never be mistaken for a great offensive shooter. Landon also was the best screen setter and one of the best passers from the post that KU has had. Darnell had a skill that was always overlooked by casual fans but greatly appreciated by the coaches. Darnell was better at blocking out opponents on the glass but very often blocked out in a manner that allowed a teammate to grab the rebound. That 2008 team played together so long that they had a great sense for each other and Darnell improved his teammates rebounding numbers significantly without having anything show up in the box score save for +/-. So you see, this is an example where +/- is the only metric that gives us a glimpse of a players value. Darnell and London never had flashy numbers but they made the team much better. Another example, but one who did have a wee bit flashier numbers his senior season, was Travis Releford. Travis led the team in +/- by a pretty significant margin. Travis was not a ball thief defender as much as Mario or others, but his on the ball defense was a thing of beauty. Plus, he set the pace defensively for the team and made the team much better defensively. Where the +/- noise really comes in is when we normalize to minutes played. In other words, we divide the +/- by minutes played to get +/- per minute. Overall +/- is a pretty good measure a players value to the team. But when you normalize to minutes played it introduces some noise.

2. Offensive +/-: This is the points per minute scored by KU when a player is on the floor. Over Mitch Lightfoot's career, the KU points per minute is below the team average when Mitch has been on the floor. Mitch's game has mostly been defensive oriented. The team becomes better defensively when Mitch comes into the game but has suffered offensively. Conversely, in 2008, Sherron Collins was the 6th man. When he came into the game over the course of the entire season, the KU offense was a bit above their season average. But the defense suffered significantly. Mostly because he replaced Russell Robinson. Some players, like Darnell, have a positive impact on the team on both ends of the floor more or less equally, whether or not they put up flashy numbers. Darnell's offensive +/- was above average and his defensive +/- was above average. He made the team better on both ends of the floor. But some players are strong on only one end of the floor and weaker on the other. Early in the season 2 years ago, LaGerald Vick led the team in offensive +/- but was 5th in defensive despite playing with the starters virtually all of his minutes. By the middle of the season, he was not good in either the offensive of defensive +/- though. So offensive +/- is a reasonable measure at how a player impacts the offense when he is in the game.

3. Defensive +/-: Same as offensive only now on defense. In 2008, who led the team in defensive +/-? It was Mario, but 2nd was RussRob. Darnell was 3rd. Brandon was always touted as the best defender on that team and I am not questioning that. He was a great on the ball defender. But did he make the team an overall better defensive team, including keeping the opponent from offensive rebounds? I would argue from the defensive +/- that there were others on that team that were at least as good overall in making the team better defensively. But remember, that essentially, the +/- in all of its forms is a team stat. So everyone benefitted from Brandon's defense when he was in the game. this season, everyone is bunched up in a small spread on defensive +/- (starters between 1.63 and 1.68 so essentially no difference). But two players have somewhat better defensive numbers and one player has somewhat worse. Harris (1.54) and Lightfoot (1.57) are a bit better than the pack and Enaruna (1.80) is a bit worse than the pack. Is that meaningful? I think probably yes, to a degree. But you also have to look at which games those minutes came from. The starters are all in every game. What if a plyer like Harris played most of his minutes against weaker teams. That is the kind of noise you have to filter out. Given that Lightfoot's career has shown the team tends to play better defense when he is in the game, I am inclined to buy into his slightly better stat this season as well.

4. Delta -- This is a really difficult one to make sense of. It is essentially trying to mirror the stat many sports are developing variably named "Value over replacement" or something like that. It is the measure of the ratio of how a team performs (+/-) when a player is in the game to when a player is not in the game. There is a lot of noise in this stat not the least of which is number of minutes played and when. The worst delta I have recorded was for Frank Mason his senior season. He played 39.33 minutes. He rested for only 40 seconds for a brief period in the 2nd half. During that 40 second rest, Devonte' drained a 3-pointer. On the subsequent possession for the opponent, Devonte' stole the ball passed ahead to Svi for a fast break lay-in. A quick 5 points. The other team called timeout and afterward, Frank was back in the game. KU ended up winning the game by 4 points. So KU outscored the opponent by 4 but in the 39+ minutes Frank as in the game, KU outscored the opponent by -1 and while he was on the bench KU outscored the opponent by 5 or 7.5 per minute. That gave a delta of close to minus 8 points per minute. But was Frank's replacement that much better? Of course not. Just luck of the draw that Frank sat for those two key possessions. You really have to take the delta with a grain of salt in any given game. It has more validity over a larger data set more so than the other +/- metrics.

Great explanation of the meaning, value, and limitations of the different +/- stats.