Plus/Minus as Another Tool in Analyzing the Bruins Performance
by Patroclus on Jan 27, 2010 11:00 AM PST
in Basketball
Email
PrintAt many points over the course of this basketball season, whether it be during an exhibition of Drago's matador defense, of JA not taking an open jumper, or Reeves missing yet another free throw, I have wondered what impact each of our players have had on the team's performance. One way to attempt to separate out the contributions of each individual player is through the Plus/Minus statistic, which compares the team's performance when each player is on the court to when that player is on the bench. The statistic has its origins in Hockey, but is beginning to gain traction in Basketball as a way to gleam insights not otherwise apparent from a boxscore. Rivals.com recently published an explanation of Plus/Minus in terms of its application to basketball.
In an attempt to provide greater insight into the performance of the basketball team, I have decided to begin calculating these Plus/Minus (+/-) figures for the Bruins, having now gone through the past few weeks of game records and compiled +/- stats for our Pac-10 conference games played to date. While the Rivals article referenced above describes several different ways to count and to analyze Plus/Minus, for now I am taking the basic +/- analysis, simply looking at how many points are scored and how many are allowed while each UCLA player is on the court, to come up with a +/- statistic for each player in each game.
The numbers, and a couple of surprises after the jump.
| ASU | UA | @Cal | @Sta | USC | UW | WSU | Total | Minutes Played | Minutes Share | Share/ GP | Adj +/- | |
| Lee | 0 | -8 | +1 | -9 | -24 | -1 | +14 | -27 | 253 | .888 | - | -30.42 |
| Roll | -1 | -10 | +1 | -14 | -15 | +6 | +12 | -21 | 255 | .895 | - | -23.47 |
| Honeycutt | +7 | -7 | +9 | -2 | -18 | +6 | +1 | -4 | 185 | .649 | - | -6.16 |
| Dragovic | +14 | -18 | +6 | -13 | -12 | -7 | +17 | -13 | 233 | .818 | - | -15.90 |
| Nelson | -8 | -8 | -1 | -14 | -16 | +4 | +10 | -33 | 190 | .667 | - | -49.50 |
| Anderson | 0 | -17 | -8 | -8 | -4 | - | - | -37 | 125 | .439 | .610 | -60.26 |
| Keefe | +1 | -2 | -4 | 0 | -9 | 0 | +2 | -12 | 102 | .358 | - | -33.53 |
| Abdul-Hamid | - | - | 0 | - | 0 | -8 | +4 | -4 | 41 | .144 | .248 | |
| Lane | -3 | +6 | -1 | 0 | -5 | +5 | - | +2 | 28 | .098 | .114 | |
| Morgan | 0 | -6 | +2 | +5 | - | - | - | +1 | 12 | .042 | .073 | |
| Moser | - | - | - | - | -2 | - | - | -2 | 1 | .004 | .025 |
The first thing that stands out about the team's +/- to date in conference is that Nestor's favorite player is not at the bottom of the chart. When accounting for the share of minutes that each player accounts for, Drago's -13 rates him as the rotation's second most valuable player since the start of conference play. Despite a hesitance to take many shots himself, or maybe due to his willingness to sacrifice his points scored in order to pass the ball and set himself up to grab the occasional offensive rebound, Tyler Honeycutt has been the team's most positive influence during Pac-10 play, with the team allowing (only) 4 more points that they have scored while he is on the court.
While Jerime Anderson's team-trailing -37 rating likely does not come as a surprise to regular readers of this blog (despite missing the last 2 games due to injury), I was surprised to find that Reeves Nelson comes up a close second - and before JA's injury was the team's clear leader (?) - in this category with a -33 rating. While RN's struggles at the free throw line have some effect on the magnitude of this negative figure (if he hit even 2/3 of his FT's, he would be in Lee/Roll's +/- range), it remains an aspect of his game, and until he shows an improved ability to hit from the line, something that has to be considered when analyzing his effectiveness. While his offensive skills have rarely been displayed, James Keefe's more advanced fundamental understanding and grasp of Coach Howland's system gives him value in the rotation. When broken down by shares of playing time, Keefe's -12 rating scales to be a few points better than RN to this point, and does seem to justify the 15-ish minutes per game that he is seeing, whether starting or coming off the bench.
Interestingly, the only two players that have a positive +/- rating over the course of Pac-10 play are Brendan Lane and J'mison Morgan. For both players, the quick hook often given to them - and the resulting limited time on the court - has been explaned by some by an inability of these players to contribute much in meaningful game situations. While their positive figures are derived from a very small sample size, the fact that they have not seen the team fall apart when present on the floor could cause one to think that these two should get some extended opportunities as the season continues (in Morgan's case, when he is healthy enough to rejoin the lineup).
Edited (2:20pm): I have edited the above chart to include the minutes played by each player. I have also included the share of the team's total minutes that each player has participated in. For players that have not appeared in every game, I have also included the share of minutes that those players particpated in for the games in which they appeared; probably not relevant for this post, but something that may be of general interest. I also added an adjusted +/-, reflecting the share of total minutes played by each player (appearing in at least 25% of total minutes) and scaled up as if each player participated in 100% of minutes played. Take that one with a large grain of salt; just throwing that one out there for thought.
49 comments
|
0 recs |
Do you like this story?
Comments
One word of caution, however.
A lot of times, +/- can be less than accurate because of WHOM a person plays with.
For example, if ML is counted on to be the bail-out guy for a rotation lineup that is incapable of scoring on their own or in making open shots, his +/- may be inaccurate, as opponents may be double and triple-teaming him, contributing to a high minus number.
Overall, however, a great statistical analysis of the team, one which should fuel discussion for weeks. Bravo, P.
M
"In this program your passion bucket must be full to play SC." -- CRN, to Dan Patrick, 1/2008
Yeah, it is certainly not a be-all stat, especially in basic form
Just something to look at, among other things in looking at performance. +/- can also be used in terms of combination of players, particularly player pairs (ie. Drago/Nelson, or Honeycutt/Keefe …) to give a slightly better idea as to how a player’s +/- is affected by the teammates on the floor. I might give that a try in the future, but just creating (and double-checking) the basic +/- for these games took enough time for now.
formerly bruinhoo
being double and triple teamed isn't a problem for +/-
as +/- would be much less affected than other stats (like ‘points’ or ‘FG%’, which we typically use without comment…).
BUT, the +/- would be skewed if (say) ML always played point guard whenever the bench players were in, while JA played some minutes at point with the starters while ML sat. Which is basically the point you’re making.
by britishbruin on Jan 27, 2010 1:14 PM PST up reply actions
I think I got a C in Stats 50
way back when, so I have no idea if this has any relation or not, but for the final scores the team is at minus 32 over the games referenced. I’m sure some south campus people will let me know whether comparing a player’s +/- to the team’s has any logic to it or is just plain stupid.
I don’t think BL’s or Bobo’s minutes allow much analysis – per the official site, BL has 28 minutes in conference play and Bobo has 13.
http://www.uclabruins.com/auto_pdf/p_hotos/s_chools/ucla/sports/m-baskbl/auto_pdf/cumulative-stats
Is Bobo healthy yet?
I’d hate to see such a highly-touted recruit go to waste two years in a row.
Then again, given his defensive shortcomings, I suppose he doesn’t really deserve more playing time. But still. It would be nice to have him back to give RN a rest.
by longbordr52 on Jan 27, 2010 11:26 AM PST up reply actions
Didn't quite mean to post yet...
I’m thinking if an iron man who has played every second of all seven games would be at minus 32, perhaps that might lead to some comparison. Again, an open invitation to prove me laughably wrong if I am, which wouldn’t surprise me.
Regardless, P, this is quite interesting. I appreciate your taking the time to put it together and am eager to see the discussion that follows.
Bobo and Lane are composed of a very small sample
Certainly not a sample that can be scaled up to say that they deserve a rotation spot, for instance. But, given the quick hook they are subjected to and the thoughts surrounding their play (at least from my recollection of game threads), I was expecting something closer to a consistent – from those two on a game-by-game basis.
formerly bruinhoo
I'm South Campus, but it doesn't help a whole lot in this discussion
I think +/- stats are a lot more useful for a sport like hockey, when most of the best players are on the ice 30-40% of the time, not 90% of the time. I was going to comment that Roll’s and Lee’s +/- stats are almost perfectly in line with the team’s performance, but who’s to say that the team’s performance isn’t influencing Lee and Roll (and not the other way around)? Either way, interesting stuff, especially for the people who rotate in and out like Anderson, Keefe, Dragovic, Honeycutt, and Reeves.
OT
I know you’ve been on here before, HailRover, but your username is great. I’d put 95% odds or better that you are/were in the band!
Why, thank you
and yes… I am in the band. Last weekend was a big relief, we hadn’t been playing that combo enough. You were in the band, right?
by HailRover on Jan 27, 2010 12:30 PM PST via mobile up reply actions
Fall '90 through spring '94
and loved every minute of it (well, that senior year tournament trip to OKC sucked…).
Don't forget
The user “solidgoldsound” who posted below!
by KSBruin on Jan 27, 2010 6:12 PM PST via mobile up reply actions
Ya
These are skewed towards minutes played per game.
UCLA '08
by BruinTrouble on Jan 27, 2010 11:50 AM PST up reply actions
They are, to a degree
Lee and Roll have played about the same number of minutes over this span of games (Roll has played 2 more), while Drago isn’t that far behind in the % of the team’s minutes that he had played. TH is a bit behind them, but has accounted for a similar % of minutes as RN (roughly 2/3’s of the total minutes available over that span). Surprising that JA’s large negative figure has come while appearing in fewer than half of the team’s minutes played (44% of total minutes, or 61% of minutes in games that he appeared in).
formerly bruinhoo
quick request
if you continue to post these, could use add a column with total minutes, for context’s sake? Obviously it doesn’t solve all problems of interpretation, but might be useful.
by britishbruin on Jan 27, 2010 1:09 PM PST up reply actions
Yeah, I think I'll be updating this post later in the afternoon with minues
I have Total Minutes, share of total minutes and share of minutes in games that each player played in (for Anderson/Bobo/Lane… that have not appeared in every game). I’ll put up another table with those figures.
formerly bruinhoo
Very good stuff
Appreciate it.
UCLA '08
by BruinTrouble on Jan 27, 2010 3:28 PM PST up reply actions
Another South Campus guy in agreement
In hockey, entire lines are substituted instead of single players. The +/- stat is usually an indication of how the player performs with the line and how the line performs with with player, especially when they’re tinkering with lines. +/- is an interesting stat, but I suspect the real reason why it’s there (in hockey) is to credit non-scoring non-assisting players for good play in a low-scoring game. :)
I haven’t really thought about what the stat means in basketball (it’s probably just there to stir up debate amongst stat junkies), but I think it should also be applied to how players play with other players. Maybe P can put together a Taguchi matrix or something and come up with the line that maximizes everyone’s +/- stat. :)
by solidgoldsound on Jan 27, 2010 1:06 PM PST up reply actions
There are plenty of Laker games where Derek Fisher adds very little and winds up with a huge + number while Farmar plays his brains out and ends up in the negative. That has a LOT do to with the Lakers starters vs bench quality. Still an interesting stat that can be mixes in as a part of the overall analysis…
Careful, North Campus guy here...
Although I did do reasonably well in Stats, several years ago, that is why I am keeping this relatively basic. I am thinking of working with this more in upcoming days, going back to cover the non-conf games, as well as looking at the +/- of some different player combinations (don’t think I want to go through every lineup permutation, or every possible pairing, but anyone with suggestions for player pairings that might be of interest – ie. Anderson/Lee – that might be of interest, feel free to post them). If I continue on with this, I think I’ll also separate out the points scored/allowed for each player; adjusted for playing time (points-40 minutes), that might also be of some interest.
BTW, I am sure there are punchlines galore for the idea of a Lawyer fooling around with Statistics.
formerly bruinhoo
Some apples and some oranges
Thanks very much for doing this, P, as it definitely adds something to the analysis.
Interpretation remains difficult. M mentions that it depends on who one is playing with; an equally valid question is, who are these people playing against?
Our starters get to start the game against the opponent’s best 5 when both sets are fresh, and typically finish each game against the opponent’s best 5 when both sets are pretty tired. Therefore, it makes a lot of sense to compare starters +/- to each other, as they mostly play against the same caliber of player, and have enough minutes to make the statistics meaningful.
It becomes a lot less easy to compare starters to bench players, or make any case for JMM or BL getting more playing time based on the +/-. Both those guys tend to come in either when the opponent’s bench players are in the game, or when the opponents’ starters are tired.
Another thing too
is that in hockey, a player usually is part of a line and will play with many of the same players on a consistent basis and thus makes it easier to compare. If there was a way to break out the stats within player groupings, it would be more telling.
Some interesting things do jump out at you from these numbers.
Every game that we have won, TH is in the + and other than the ASU game, MR would have been as well.
On the night that MAH sunk the winning baskets and scored 8, he was a -8. I am pretty sure that it wasn’t him making the rest of the team worse during his minutes.
Thanks Patroclus for your efforts. I am not sure if it means much, but is none the less, interesting for discussion and comparison sake.
hehe np :)
But you brought up more interesting points that I won’t have confidence posting!
by solidgoldsound on Jan 28, 2010 12:21 AM PST up reply actions
Good work P but don't waste your time
I really appreciate the hard work you did in compiling all of this on your own, but my friendly advice would be for you to save your time and not worry about this anymore.
As has been mentioned above, +/- really isn’t a good concept for basketball. It is the baseball moneyball concept trying to invade basketball and just doesn’t work because there are too many intangibles in basketball to statistically capture. Its easier to stick to objective things like rebounds and points (even assists and turnovers are rather subjective) and your eye.
The other problem with +/- particularly with this team is that we don’t have much depth. Last game CBH only played 7 guys and we had one of our best games of the year. The more minutes players get on the court and with certain players (again you cant calculate chemistry and confidence) they better they generally play which is why now that CBH has shortened the rotation and people have more defined roles, they are playing better. If the only reason to look at plus/minus here is to say that BL or JMM should get more minutes or that someone is our fourth most efficient player, that doesn’t seem like something you should be devoting a ton of time too. Of course, if you really like doing it, by all means go for it.
I think it supports the argument pretty well
that Bobo should be running the point
by britishbruin on Jan 27, 2010 2:35 PM PST up reply actions
Thanks so much for looking into this P
I understand the issues with these stats as others have already pointed out. However, I still found the info. very interesting. To me this info is still helpful for at least proving some context while having big picture discussion.
+/- not irrelevant, but difficult in b-ball
First of all, thanks P. This clearly took a lot of work. I’m in general agreement with bruinponcho that working more on this probably isn’t the way to go. Unless you are really good working the numbers and have them in an easily manipulable format, making a +/- make sense in b-ball is tough.
It’s my understanding that some places add in rebounds, assists, and subtract turnovers to come up with a more well-rounded number indicating impact on the court.
As for pairings, it would be a good idea to look at A) Howlands rotations (are there natural pairings HE is searching for?), and B) guard vs big men pairings. So, take a look at the difference between ML and MR versus MR and JA. This will give a more natural look at how different elements of the team are working together.
Another way to think about player impact is to measure how well the team does with them OFF the court. This would be the negative image of the +/- and would measure the the impact of players by demonstrating what the team is missing when they are not on the court.
You also might want to just throw out the 2-3 big losses. They might be skewing the sample, and if you take them out other factors might come to the fore.
at least points are objective
as bp suggest above, the following logic may be incorrect
add in rebounds, assists, and subtract turnovers to come up with a more well-rounded number indicating impact on the court
precisely because assists and turnovers are somewhat arbitrary statistics; and even rebounds are misleading – I’m sure some people here can point to the number of rebounds credited to Nikola “battle on the glass or through it” Dragovic which are the result of 4 players boxing out effectively and the ball falling into his hands. And there is no quantification of how a player draws double teams on offense or effectively helps on defense. So, while +/- is certainly flawed, I would strongly suggest not spending more time calculating more convoluted (and perhaps more arbitary) statistics.
by britishbruin on Jan 27, 2010 3:52 PM PST up reply actions
points are only marginally more objective than rebounds...
Some people make lucky shots, just like ND may get lucky rebounds. Some shots are taken from half court and still count against a players percentage. The points of statistics is that through time luck evens out and you can test for skill. Points are only ‘more objective’ because the ball goes through the hoop. To keep the Hockey analogy, are assists in hockey subjective? I don’t think so.
Anyways, BP is right that this is a tough measurement and of questionable utility in b-ball.
Sure
But in addition to the objectivity of points, they are also more predictive of victory than any other statistic. You win every game in which your overall team +/- is more than zero… Rebounds, blocks, assists, turnovers etc are one step removed from your actual goal; their value comes in how one imagines they contribute to games, but composites of these give the false impression of objective comparability of one player who has 10 rebounds with another who has 10 assists.
Not to defend +/-, just to suggest that arbitrary combinations of mildly subjective statistics whose relative value in contributing to victory is unknown is “unfair, wrongheaded”. Or not that big a deal…
by britishbruin on Jan 27, 2010 7:49 PM PST via mobile up reply actions
and why doesn't the athletics department have somebody doing this?
Get an Econ. grad to punch all the numbers into STATA and you’re good to go!
ahh STATA
good program. I wonder if any students/recent alum on BN have taken a class with John Zaller.
ND, Nestor and other thoughts
Thanks for doing this. Agree with the caveats but let me just add:
1. Lane and others haven’t played enough to judge. Our best team game (WSU) they did not even play.
2. The ND effect is interesting. An example of how it matters and is relevant is the Washington game. ND stunk up the place BUT UW was keying on him, even double teaming him when he put the ball on the floor in the first half. I heard that Romar thought shutting down ND was a key to the game on defense. (As an aside to British Bruin, it is one of the reasons I said UW was badly coached in that game.)
Other teams not only don’t read Nestor’s postings but may read what ND did to ASU or how he is the leading returning scorer. Thus even though ND has not played well much of the time, his presence on the floor makes us a better team because of the way other teams react. Thus his better than expected +/- and the best of the four long time starters.
3. JA being so bad makes sense. Watching him in the man defense in the AZ game says it all. (Caveat, I hope his injury was the problem.)
4. RN being so bad is a bit surprising. He has had major defensive lapses and seems to support CBH’s decision to start JK.
5. I have been a big fan of TH for a while and this seems to prove what many have been saying, he is this team’s best player.
don't you mean
ND stunk up the place because UW was keying on him? :-)
by britishbruin on Jan 27, 2010 4:01 PM PST up reply actions
I know you are joking
But you’re right that’s what the stats show. I did not look at the game by game until after I wrote it. But it shows EVERY game we won but that one, ND was a big + and 2 of the 4 he was the biggest +.
Let me stop before I go to the unpleasant place that line of thinking leads.
it's because he gets the quick hook
when he is playing badly, and CBH rides his hot hand when he is doing well… that’s why he is often the biggest plus, and rarely the biggest minus… right, Nestor?
by britishbruin on Jan 27, 2010 4:27 PM PST up reply actions
Even objective stats aren't objective
Or really accurate. Particularly rebounds, blocks, steals. There was a piece about the Houston GM about a year ago that addressed this.
What if somebody is boxing out for others to get the rebound? Perhaps I’ve been underrating James Keefe because of this. What if 4-5 rebounds fall into somebody’s lap, as opposed to rebounds that are truly fought for? Perhaps that’s why Drago had so many rebounds in some games and barely any in others — just chance. I think we can see Nelson and Honeycutt really fighting for rebounds, whereas other just pick up the trash.
Most blocks go straight back to the other team, especially those wild rejections everyone gets all excited about. As for steals, a lot of players become overrated defensively because they get a few steals, while missing out on many more and giving up layups.
Some GMs have apparently devised very precise, secret stats to measure these things more accurately. We’ll just have to rely on our eyes to capture the context.
We had this discussion about Drew Gordon last year
Whose blocked shots were disguising his actually lousy defense.
hate to say it
but I think RN’s defense gets a little overrated for the same reason
by britishbruin on Jan 28, 2010 7:02 AM PST up reply actions
Can someone make this a fanpost, I can't firgue out how to do it, From ESPN Blog
There haven’t been any documented violations yet by former Tennessee coach Lane Kiffin at Southern California.
But did you catch the comments of Frostproof (Fla.) High School athletic director Chuck Loveless earlier this week?
Frostproof cornerback Nickell Robey was previously committed to Georgia, but he’s looking around again. And one of the places he’s looking is USC.
Loveless told the Atlanta Journal-Constitution: “He’s planning to visit Southern Cal next week. Monte Kiffin has been very persistent. He’s probably been in here more than is legal. Of course, they don’t pay attention to the rules.”
Imagine that.
espn.go.com/blog/sec/post/_/id/8483/kiffin-not-paying-attention-to-the-rules
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
