Editor’s Note: The following is from Andres (Dre) Alvarez of NerdNumbers
I decided to look at Chicago’s Win Produced numbers after I heard for the millionth time that
- Rose was MVP (Thanks to the Stat Heads who dispute this!)
- There was no way Kevin Love could be considered MVP
An amazing thing stood out to me – the Bulls don’t have bad players! An average player has a WP48 [Wins Produced per 48 minutes] of 0.100. Giving us a little wiggle room, let’s define a “bad” player as one that is ½ the value of a good player, or a WP48 of 0.050 or less. Keeping that in mind, let’s take a look at Chicago this season.
As one can see, Chicago has had the fewest “bad” players suit up for them. Furthermore, the Bulls have had a remarkably low 1,215 minutes allocated to those “bad” players. This is a whopping 700 fewer than the next closest team (the San Antonio Spurs).
To put this in perspective the Nuggets have given Al Harrington over 1,500 minutes, and received about -2.0 wins as a result. All of the Bulls “bad” players put together aren’t even as bad as many teams’ worst player.
Now let’s also take a quick look at Minnesota. The Timberwolves have had 13 “bad” players. These players have played over 12,000 minutes and combined to offer -10.0 Wins Produced.
What does all this mean? Not only does Derrick Rose have some very productive teammates (think Carlos Boozer, Joakim Noah, Ronnie Brewer, etc. . .), unlike other teams he hasn’t had to play with many “bad” players. In contrast, when Kevin Love looks around the locker room, “bad” players are everywhere.
Just to hammer this point home, let’s put how good Chicago is playing in historical perspective.
Chicago, despite injury problems, has been able to get away with having the 9th lowest total of “bad” minutes played in the last 33 years. Of all the teams in front of them, only the Kansas City Kings failed to make the playoffs*.
And except the 1983 and 2001 Kings’ teams, every team in front of Chicago made it all the way to the Conference Finals. In short, Chicago is in historic company.
How did Chicago do this? The front office essentially followed a very good formula: keep all of the team’s good players from 2009-10 and get most of the good players from Utah. This formula has produced a very good team.
Unfortunately in the NBA, all of the regular season awards are individual based. This means Derrick Rose will get the credit for what was essentially a team effort. While it is worth virtually nothing, I would like to award the Chicago Bulls with the Most Valuable Team award this season and reiterate that Rose is not the MVP.
One last note…
Some might argue that Chicago’s players are essentially “not bad” because of Rose and/or Coach Thibodeau. But if we look at the productivity of these players this season and what these veterans did in 2009-10, we see that this team’s results are not surprising. As the following table indicates (numbers from NerdNumbers), most players on this team are offering essentially the same production seen last year.
The primary exceptions – Derrick Rose, Ronnie Brewer, and Carlos Boozer – are easy to explain.
- Rose is young, and young players get better.
- Brewer is essentially offering what he did prior to 2009-10
- And Boozer – who has been hurt – is offering less this year
In sum, we don’t need to argue that this team is succeeding because of its dynamic point guard or amazing coach. The Bulls story is really about choosing productive players. And the Bulls – as the study into “bad” players indicates – have been very good at making these choices.
-Dre
*An explanation for Kansas City. First, they did share the same record as the Nuggets and were thus the 6th Seed in the playoffs. But unfortunately they lost the tie-breaker with Denver. Second, unlike all of the other teams on this list, Kansas City was the only one without a Star (WP48 > 0.200) or Superstar (WP48 >0.250) on their team.
Wages of Wins Journal 
Gil Meriken
April 1, 2011
I was hoping for more about Love, and not Love vs Rose, which can be a semi-plausible argument even outside the world of Wins Produced, but rather, I wanted to read about Love vs Lebron, Dwight Howard, Chris Paul, and Dwyane Wade.
I would like to hear about how he is clearly a more productive player on a WP48 basis, and how a GM who values winning (and not about box office, ratings, and other non X and O issues) should take advantage of popular perceptions and create a trade as soon as possible to include any of those lesser players for Love, especially given his age and contract.
Minnesota’s David Kahn might just be dumb enough to accept an offer of Lebron for Love, Ridnour, Tolliver, and Webster, and Miami would be on its way to more wins! Better yet, Miami could offer only Wade, and Minnesota would still probably take the bait.
But Minnesota’s bid for James might be thwarted if the Magic wised up and sweetened the pot by offering Howard for Love, and then we would have a real bidding war for the productive services of Love.
Chicago Tim
April 1, 2011
Only the most diehard Bulls fans think Rose is the best player in the league. But that has never been the criteria for MVP, so I’m okay with Rose winning it this year.
Daniel
April 1, 2011
There have probably been 275k-300k total NBA minutes played this season so far. If 167k are played by players with a WP48 of .050, shouldn’t the production of an “average player” be similar? The average NBA player is probably close to .040 production with a .160 standard deviation (since “stars” in this metric have a .200 WP48), which tells a vastly different story than the weighted average production of .100– a number which is highly skewed because of the propensity of coaches to play their stars more than their scrubs (this is kind of a joke… ha).
If you apply the Pareto principle and create a team with players right around the 50th percentile (which should be the “average players”) with NO “Big 3” players who can produce ~80% of your wins; you should have a team that wins about 20% of its games (this is where I extrapolated the .040 from– 5 players on the court averaging .040= .200 wins per game, which is 16.4 wins over a season– I wouldn’t be surprised if the real numbers are almost exactly the same– statistics are awesome!).
It should pass any casual NBA fan’s “smell test” as well, since creating a team of players who fall somewhere between any given team’s 5th and 10th best WP48 man would give you a terrible team. It’s simply a view of a different facet of “average” that may help reinforce the idea that the NBA is truly a superstar’s league.
Daniel
April 1, 2011
Toronto has 6 players on their roster with a WP48 of .163 or above (a quick skim tells me this is more than ANY other NBA team), and yet they have managed to give 8500 minutes to players who produce in the NEGATIVE range!
Bargnani is clearly the worst #1 pick of all time– at least Olowokandi was kind enough to the Clippers to be bad at taking lots of shots so he wasn’t seen as a valuable “scoring” big man and kept on for an entire decade.
Fire Jay Triano!
marparker
April 1, 2011
Gil,
The brilliance of your sarcasm has caused me to see the light. My, how I have been so blinded by the temptation of regression analysis. I can only hope that my ill-fated devotion to the fake genius of David Berri will be forgiven by you who has proven to be the true genius. I kneel before thee and humbly ask for you to illuminate my path as I prepare to make my journey to true salvation.
nerdnumbers
April 1, 2011
Daniel,
Just some quick math, there have been around 75 games of 48 minutes in length for 30 teams, each of which has to field 5 players at every minute. 75*48*30*5 = 540000, so around 30% of available minutes have gone to below average players. The average for each player is 0.100, as that is how the metric is constructed. You are absolutely right though that the standard deviation can be huge.
Tommy_Grand
April 1, 2011
I look forward to the Lakers acquiring Kevin Love during the off season for like 2 draft good picks and Aaron Mckie.
Mike
April 1, 2011
This blog is funny…your method of evaluating players is clearly wrong, but I end up agreeing with you more often than not.
Italian Stallion
April 1, 2011
Who is this Derrick Rose character everyone seems to be talking about?
Adam C. Madison
April 1, 2011
Interesting article. Keep it up. I look forward to similar posts in the future.
Thomas
April 1, 2011
It’s funny how the Lakers give out so many “bad player minutes” (which I suspect are filled 48 minutes a game at the point guard position), will probably lead the playoffs in “bad player minutes” (again, due to the fact that somebody will be playing point and will be contributing nothing) and yet will likely win another championship.
That by itself should send up a red flag on analyzing a situation like this, shouldn’t it?
Gil Meriken
April 1, 2011
@marparker – Do you disagree with what I wrote? Is it not true that based on the results of this season, and assuming a team just wants to win, that a team should be willing to trade any one of those players (LBJ, Wade, Howard, CP3) for Love?
I have never declared Dr. Berri to be a fake genius, nor would I ever do so. He is quite clearly an intelligent person. Because I disagree with the conclusions made from the results of his system does not mean I think any less of him as human being. Or you, for that matter.
Zack
April 1, 2011
Would the WoW network really give Love the MVP? I mean, I love the guy (no pun intended), but I’d say D-Howard is the clear MVP this season. You’ve shown that production of and individual drops if his teammates are more productive, wouldn’t this be applicable in the case of Love v. Dwight?
Also, I could see someone arguing that if Lebron James took the place of Kevin Love on the Timberwolves right now, they would win more games because James demands more attention on offense. He’d draw double teams, his teammates would have more open looks, etc. As a result, the other players on the T-Wolves wouldn’t be quite as negative as they are now. How would you respond to this
MKSE
April 1, 2011
I agree with Gil. There have been numerous articles on the WoW network extolling the value of Kevin Love, and showing how he has been more productive than anyone else in the NBA this season.
So for once, I’d just like Dr. Berri to come out and say two things: one, Kevin Love is the MVP; and two, he would be willing to trade anyone in the league for Love, including James and Howard. Seems like an obvious leap, so why the silence?
nerdnumbers
April 1, 2011
Zack,
Let’s be clear WP48 isn’t the end all be all. Dwight Howard has been a top 5 WP player for the last 6 years, including this one. He’s a defensive nightmare and remained healthy. Kevin Love has played at superstar level for the last two seasons but has had some injury concerns. The Salary and age difference would make me want Love over both Lebron and Howard at this point.
There’s this weird notion that good players on bad teams are inflated. What confuses me about this is. . . . hasn’t anyone been paying attention? The Celtics fleeced a great Garnett off of a terrible Boston and BAM contenders. The Lakers fleeced a great Gasol off a terrible Memphis and BAM contenders. When you bring up Love somehow he can’t be that good, his team sucks!
Also I gotta call party foul. You list diminishing returns in paragraph 1 (by being good you make your team mates worse) and then list improving team mates in paragraph 2. Just to clarify your point is good players make other good players worse, but good players make bad players better?
Adam C. Madison
April 1, 2011
One thing about the WoW network is that it hedges. It won’t come out and stand firm ground, which is one of the reasons it (and the sabermetric community as a whole) remains a niche.
Look at baseball and how far it came. And really, fantasy sports was a major propeller for that. I draw on my own fantasy playing/writing experience. Take a look at Matthew Berry. He purports a lot of “out there” things, but he does two things: One, he takes accountability, standing on one side of the ledger. Secondly, he acknowledges vulnerability, that he may, in fact, not have it all figured out.
It’s interesting, and reminds me of Malcolm Gladwell’s The Tipping Point. Sabermetrics in basketball have passed the Tipping Point. TrueHoop has a stranglehold, yet not quite a lock. And this site gets a ton of hits from Henry Abbott love.
Yet it’s almost like they prefer being in that niche, because when you come from out of that niche, you get shots fired. You get wise-ass mother fuckers like me reading your every move, waiting to abandon ship at the nearest sign that, hey, the opportunity cost of reading this site is quite large, and a ton of other competitors will rush to take your place.
Anyone see the last season of Dexter? TAKE IT!
Gil Meriken
April 1, 2011
@nerdnumbers two points 1) I think Love is a very good player, it’s the degree of how good he is relative to Lebron et al that I disagree with 2) Love’s team’s poor performance carries little weight to my estimation of his impact on winning. In fact, if the Wolves were to win the championship, it would carry more weight with me, which may seem to be inconsistent, but I don’t have the words to describe how it’s not equal, I’m sure there is a concept name for this. For example, in the stock market, it’s said that Insider selling can mean many things, but Insider buying almost always represents confidence in the company. In the same way, poor team performance is not always a negative reflection on a player, but excellent team performance (especially winning it all) always bears some positive points for a player (we’re talking players who play the most or second most minutes for said team).
MikeMother
April 1, 2011
What you are attempting to do is construct a Reductio ad absurdum argument:
The problem is, the idea that a trade should be made based upon WP48 numbers alone is silly when WP48 is a heavily position based metric. As a very basic rebuttal, there is no way a team with 8 PGs wins many games. Ditto a team of Kevin Love, Howard, Chandler, Gortat, Shaq, Ben Wallace, Zach Randolph, Tim Duncan, Joakim Noah and Jeff Foster, all players over .200 WP48, would struggle to win even a single game. As a Suns fan, trust me, a team of all wings is just as bad!
The key to winning in the NBA is to field a rotation of 8-10 players who collectively produce at each of the 5 spots on the court. You can’t do that with 8-10 PGs or 8-10 centers. No one has ever argued you can, and no one ever will, and any argument that disregards the impact of position, and the likely allocation of minutes after a trade (e.g. who replaces the traded players) is flawed.
As an example, Kevin Johnson for Larry Nance was a great trade for Cleveland as they already had Mark Price. Without Price, a PG for a 4 is a very odd trade, for both teams.
The best team I know of, the Bulls circa 1993, the one that won 71(?) games, fielded centers who were AWFUL. So not really that big a surprise :)
Of course when you have MJ, Pippen and Rodman…
The article is about the quality of team mates, not the value of a team with no bad players. In fact, WP48 states it is as better to have one Lebron (.354 WP48) and D Wade (0.297) and 3 bad, zero production players (.000 WP48) than 5 average players (.100 WP 48).
I think that is pretty much what everyone thinks, no matter what system they subscribe to, don;t you agree?
However, when it comes to MVP, if the player’s support helps the win column more than most teams, how is the “star” the MVP?
Thomas
April 1, 2011
I can’t speak for the Celtics case (no way to know how good they would have been with Pierce, Allen, Al Jefferson + spare parts), but in the Lakers case history always forgets where the Lakers were pre-Gasol.
That year’s Laker team spent a day in 1st place in the West right after Bynum got hurt. If memory serves, they won in Seattle in OT, Phoenix lost one along the way and the Lakers vulture’d 1st place before they went on the road and started (predictably) dropping games.
That year wasn’t “fleeced Gasol and BAM contenders”. They became contenders already with Bynum having grown up and Fisher replacing Smush Parker. Obviously for later years it’s better that it worked out this way, but if Bynum never gets hurt and the Lakers never get Gasol they’re still a contending team. Probably would have put up more of a fight vs. Boston that year in that scenario also.
Italian Stallion
April 1, 2011
I suspect that if Love has another couple of seasons like this one and the T-Wolves get into the playoffs a couple of times a conversation about him getting the MVP won’t seem as that far out of line.
People’s perception about the quality of a player is almost certainly somewhat linked to the longevity of his accomplishments and his ability to perform on the biggest stage.
At this point, Love is one of the new kids on the block, his team is horrible, and he has never played on the big stage.
arturogalletti
April 1, 2011
Ah the MVP debate, If only someone on the WOW network was taking a stand on it.
Repeatedly,with numbers.
And trying to account for D and Dim returns.
:-)
Sam Cohen
April 1, 2011
Does anyone know how many standard deviations above the mean Kevin Love and Lebron James are for their respective positions? That seems to be an important consideration (and one that I know has been part of the conversation in the past, but is seemingly being ignored this time around).
From my quick look through the automated numbers, Paul Pierce, Matt Barnes, Kevin Durant, and Andre Igoudala are the only other small forwards that are even above .200 WP48 (excluding Peja’s 22 minutes in Toronto and Jeremy Evans whose position is listed as 3.5), and none of them are in Lebron’s range. Then again, there aren’t really any other 4/5’s in Love’s WP48 range other than Howard, even if there are a lot more 4/5’s who have more than a .200 WP48.
Zack
April 1, 2011
@nerdnumbers
Just to clarify: I’m not really sure whether I think Love or Dwight should be the MVP; I’m throwing these questions at you to see what your response would be.
A person might, for example, point to the Cavs this year. We all expected them to win around 30 games even with LeBron leaving, but they’ve been awful. Yes, they lost Varejao, but many other players on that team have seen their WP48 drop dramatically. For example, Hickson has seen his production drop as he’s been forced into a more offensive role. This seems to support the idea that dominant offensive players like LeBron make their teammates more productive, whereas players who derive most of their production from rebounding and other non-scoring stats don’t make their teammates more productive. How would you rebut that?
nerdnumbers
April 1, 2011
Zack,
So Arturo was such a nice person he has the retort to your question (that he amazingly thought of 4 months ago!):
http://arturogalletti.wordpress.com/2010/11/24/re-examining-myths-and-explaining-how-regression-works/
By the way I like arguing as much as the next person but come on man!
Post 1: “I’d say D-Howard is the clear MVP this season. ”
Post 2: “Just to clarify: I’m not really sure whether I think Love or Dwight should be the MVP; I’m throwing these questions at you to see what your response would be.”
So you’re basically just talking to see what we’ll say. Interesting. My advice, just read Stumbling on Wins, Wages of Wins and Arturo’s blog instead. Have fun.
marparker
April 2, 2011
Gil Meriken,
I don’t agree with what you wrote. At what point do we consider standard deviation(already covered)? At what point do we consider injury history? At what point do we consider consistency?
I wouldn’t trade Wade for anyone in the league. Shooting Guard is almost certainly the key to winning a championship. Here’s a list of the last 10 shooting guards to win an NBA championship and their career offensive rating. Bryant 112, Allen 117, Ginobili 115, wade 111, Hamilton 106, Ellie 116, Jordan 118, drexler 114, Dumars 113, Scott 112. All but Wade and Hamilton are top 15 all time for that position and Wade is number 16. Offensive Rating is not a Wins Produced number but has the same weights of the components that they do share.
All of the other players(Paul, Lebron, and Howard) have been above .4+wp48 players before and have done so without so many negative teammates. The negative numbers are a problem for the system as obviously it is impossible to have negative wins and a team consisting of 100% negative players would be predicted to have negitive wins by the metric. One day someone somewhere is going to realize the elegant solution for that and have the same eureka moment that I did. Getting back on topic, I’m not sure if there is a “huge” difference between players who both post .4wp48 marks.
marparker
April 2, 2011
Lastly, Rose is still on his rookie contract for the no. 1 team in the conference. Who else can say they are leading a really good team in wins produced with so little effect on the bottom line? The problem for the Bulls will arise when/ they pay Rose a max contract for less than .2wp48 production. That is probably not going to work. For now, I’m o.k with Rose winning the MVP as he is in-line with the tradition of the award and hypocritical NCAA has already erased one of his great accomplishments.
Last lastly, for all of the commentary on race and Jimmer mania and the race questions being brought up, aren’t we going through Derrick mania right now? Isn’t Derrick Rose black? Isn’t he probably overrated for shooting too much as well? People like small basketball players who are player for winners, are polite and ballsy/tough. Can we please stop injecting race into everything?
Italian Stallion
April 2, 2011
marparker,
I thought I saw what I believed was a satisfactory solution to “negative wins” idea not so long ago. It was a “eureka” moment for me even if it doesn’t have the same impact on you. I can’t recall who posted it or I’d
give him the credit.
When we are estimating Wins Produced we are actually estimating a probability.
A team that totals 2 wins will not necessarily get 2 wins. It can get zero through “X” wins (where “X” will be a small number) .
A team that totals 0 wins will not necessarily get 0 wins. It can get zero through “X” wins . However the probability of zero rises and the value of “X” diminishes slightly.
A team that totals -10 wins can’t get negative wins. It can get zero through “X” wins . However the probability of zero rises dramatically and the value of “X” diminishes significantly.
A team that totals -30 wins can’t get negative wins. It can get zero through “X” wins . However the probability of zero is almost certain and the value of “X” is probably close to 1.
As long as you think of WP as a measurement of probabilities instead of literal expected wins, the whole concept of “negative wins” makes more sense.
Gil Meriken
April 2, 2011
@marparker The % differences between WP48 scores above .400 are not as meaningful as the differences between those below .400? I did not know that, no sarcasm.
And again, I don’t downgrade Love because of his teammates. I’m simply evaluating his skill set and impact on winning, using purely visual methods, and it seems quite clear he is not as good as Lebron, Wade, Howard. And yes, I have been wrong, but no, I did not rate Carmelo highly, nor Iverson. There have been times I have agreed with WP48, and times I haven’t, but it’s a crapshoot. You could get to similarly useful results just by looking at minutes played and a few other statistics. But they’re all going to be flawed, not because it’s impossible to describe basketball using statistics, but because the base statistics themselves today (points, reb, assists) don’t accurately describe winning processes, at least on an individual level. Doing regression on data that is “sort of” related to basketball (again, on an indivdual basis) is not going to bring you very meaningful results, and may help you separate the wheat from the chaff, but it’s going to get murky in trying to distinguish the better players from each other. Anyway, I have a feeling we’re not going to get very far in this conversation, as I’m talking about the basics of source basketball stats, and I’ve said these things about a thousand times, so there it is again. Maybe someday Dr Berri will build a model based on the data we get from one of those databases that tracks the position of the players in relation to the ball, his teammates, the defenders, and the basket, in concert with each point, rebound, assists, in order to better qualify each stat. I am aware that more data does not always bring more accuracy, but I don’t see how better quality data can hurt.
As for the race issues, I think you need to respond to Bill Simmons for that. I have not seen anyone on this particular thread speak of a player’s skin color.
Tommy_Grand
April 2, 2011
One thing that tilts the MVP debate is the irrational crowd who value the regular season’s final 8 weeks more that they value its first 8 weeks. Say two players rack up similar stats, and their teams finish with the same final W/L record, but one of them owned December and the other dominated the league in March. The second player invariably gets more votes for MVP. This is crazy. A win is a win and rebound is a rebound. I can see giving more respect for playoff wins, but the mvp is a regular season award: all regular season games should matter equally.
reservoirgod
April 3, 2011
I want to make two points on negative wins & Kevin Love.
First – not everyone on the WoW Network endorses Kevin Love for MVP. I believe the only articles written stating that have come from two writers – Arturo & Nerdnumbers (Dre). Kevin Love is on my WoW fantasy team, and extremely productive, but I’ve stated in the comments of Arturo’s and the Nerdnumbers blog that I don’t believe he’s the best player in the league. I don’t think they do, either. Productiontalent. Being the most productive doesn’t mean being the most talented or “best.” I do believe Love benefits from playing a lot of garbage time minutes that may make it easier for him to be so productive and there’s no way I’d choose him over Blake Griffin if I was starting a real NBA team (fantasy team’s a different story!).
Second (regarding negative wins) – I think there’s a misunderstanding of what WP48 says. WP48 measures how productive a player is in comparison to the average for their position(s) played. Nerdnumbers and Dr. Berri calculate the average for the league while I calculate the average for the game played and the Courtside Analyst calculates the average for the team and its opponents. Players that perform way below these averages will have a negative impact on their team’s production. However, the likelihood of a team being composed completely of negative win producers is very low (see how I avoided the use of the word “never” there, Nerdnumbers). Italian Stallion touched on this a little bit with his comment. The reason a team composed completely of negative win producers couldn’t exist is because their poor play would drag down the average performance at their position in the games they play and sufficiently lower the bar to the point that their production wouldn’t look negative. In a game featuring Bargnani & Darko at center, guess what? The bar for average performance is going to be really low and one of those players is going to end up looking productive for one night. So harping on “negative wins” is really a waste of time and non-issue. Those players have a negative impact in a relative sense, not an absolute one. If you change the context, e.g. compose a team entirely of negative players, the context changes and that changes their production, relatively-speaking.
marparker
April 3, 2011
Italian Stallion,
I guess I should just say that a data set was used for this regression and once you get outside of the data set predictions are harder to make(the data set being 9 win team all the way up to 72 win team). I take that to mean we have to take numbers outside of the range of .03(9/240)-.30(72/240) with extra caution. No human has figured out how to build a team that can be better or worse than those extremes but we have a model that asserts that process is simple.
marparker
April 3, 2011
Secondly,
I think the 0 line could be redefined. We know that 20% of the league is pulling 80% of the weight. If this is true then we don’t need to run a regression on every team. We might be able to run a regression on only the data for teams with a positive efficiency differential. Then the negatives wouldn’t subtract from a baseline of 0 wins. The negative would subtract from a baseline of 41 wins. The negatives are inherent to a model but they need to reflect situations which are possible. I hypothesize this could be accomplished by moving the end point of the data set further into the realm of possibility instead of having it teeter at possibilities edge.
marparker
April 3, 2011
Gil,
Thats exactly what I’m saying. After all a team consisting of the top 5 WoW mvp candidates right now would have already racked up 105 wins, which is not possible. Naturally, someone’s numbers would have to suffer and probably greatly. If we consider a team of all .2 players that impossibility disappears and the change in numbers wouldn’t be so drastic.
As for the race isssue, I wasn’t addressing anyone on this forum. I just had some basketball ranting to get off my chest. I get sick of hyperbole and then the eventual backlash against the hyperbole. Should have left it out of this forum.
marparker
April 3, 2011
Reservoirgod,
My “harp” with negative wins comes from trying to make predictions using wins produced. Using the Morrison Bobcats as an example Wins Produced predicted that Charlotte could pick up 15 wins just by swapping Morrison’s minutes with someone who was average. They added an above average player and moved back a win. It could be that in this example all the team negatives got lumped on one player. If that’s possible then maybe sometimes all of a teams positives get lumped on to one guy as well. Maybe, if finding a way to get rid of the negatives would eliminate the lumping that I believe goes on at both ends of the spectrum.
Italian Stallion
April 3, 2011
reservoir,
“The reason a team composed completely of negative win producers couldn’t exist is because their poor play would drag down the average performance at their position in the games they play and sufficiently lower the bar to the point that their production wouldn’t look negative. In a game featuring Bargnani & Darko at center, guess what? The bar for average performance is going to be really low and one of those players is going to end up looking productive for one night. ”
Not sure I understand what you are saying.
I understand that guys like Milicic and Bargnani would tend to drag down the average for the position, but I don’t see how a single game means much to their rating. For every Milicic, there’s a Howard dragging it back up.
Wouldn’t they derive their value compared to the average for the league at their position across all the games?
Theoretically, it would be possible to create a team with an expectation of negative wins as in my example above and that expectation would simply be saying something about the probability of 0 wins. Correct?
Italian Stallion
April 3, 2011
marparker
You have some interesting ideas.
There is some recognition of diminishing returns in Wins Produced. That’s especially true when we are talking about scoring.
Most of the MVP type players generate at least some of their value from efficient scoring. There are only so many shots to go around. So they would all likely suffer somewhat in terms of wins produced via lower scoring usage. I think there would be impacts also.
IMO this issue is similar to not being able to build a team with only great PGs or only great Cs.
I think no matter what model you use you have to recognize that player value and team construction are two totally different things. When teams are constructed especially well or poorly (too much or too little of anything) it will have an impact on the productivity of some players and the success of the team.
In my own handicapping, when players are traded, injured, there’s a huge lineup change etc… that part of analysis is a separate process from when I am trying to simply isolate the value of individual players.
marparker
April 3, 2011
Stallion,
You make a great point about evaluation vs. building. You have to maximize efficiency differential while also maximizing role coverage and minimzie overlap. You also have to keep in mind that something you don’t need could very well put another team over the top and practice some good hoarding. I always point to the example of Portland giving the Bulls 6 championships(1 at their own expense) by not drafting Jordan whom they had ranked higher than Bowie.
Phil
April 3, 2011
Thibs actually delegates minutes played by each player, so even using your metric, he is adding value by not playing guys like James Johnson. Additionally, even through the lens of your metric, you could argue that Rose is not only the most produce player on the Bulls, but he is the most valuable on a replacement level. If CJ Watson took all of Rose’s minutes, and John Lucas took Watson’s back up minutes(ie if Rose was hurt for the year), the pg position would produce (.035*2732+983*-.351)/48 = -5.34 wins versus 12 wins. That difference takes the Bulls from a 56 win team to a 39ish win team. Meanwhile, the loss of Noah and Boozer can be replaced by above average players: Asik, Thomas, and Gibson. This 17 win differential puts him in terms of value as one of the most important pieces to a team given the Bulls personal.
Adam C. Madison
April 3, 2011
My MVP ballot:
1. D-Ho
2. Love
3. LBJ
4. Wade
5. Rose
Look at the Four Factors:
Shooting (40%)
Turnovers (25%)
Rebounding (20%)
Free throws (15%)
If it weren’t for Love’s middling defense, we’d be looking at one of, if not the, greatest season of all-time!
Italian Stallion
April 4, 2011
Phil,
That’s an excellent point.
That’s the kind of analysis that becomes very relevant when you gamble on games and have to adjust an odds line due to injuries. ;-) Sometimes when the best player goes down it’s not as big a blow as when the 2nd or 3rd best player goes down. It depends who is behind them. It can also change your view on “valuable”.
Matt
April 4, 2011
When Garnett was in Minny and Pau was in Memphis their teams weren’t nearly as awful as KLove’s is right now. Both of them led teams to the playoffs as well. Not to say that Love will never make the playoffs, but his season doesn’t directly correlate to when Garnett/Gasol were on bads teams then moved to contenders.
KLove reminds me more of David Lee than he does of Garnett or Gasol. Certainly he’s having a great season and I expect him to continute to be a “good” player as his team improves, but he has no business being in an MVP debate. Other teams only have to play him for the first quarter and then the rest of the game is garbage time. That isn’t comparable to star players leading teams into the upper echelon of the playoffs.
Adam C. Madison
April 5, 2011
Because he’s white and stout and not black/foreign and long?
Kelly
April 8, 2011
So now the MVP goes to the player who has to play with the most bad players? The arguments against Rose grow increasingly implausible and desperate. Last night I watch the MVP of the NBA lead his team to a thrashing of the Boston Celtics.
Why is there this incessant need to try and take away from what Rose has done? I mean seriously, arguing that because the Bulls have fewer bad players than Orlando? Really? Over 20 years no less? Rose has been there for three and the MVP is for this year.
I mean really, it’s just sophistry.
John Hollinger's Eyebrows
April 8, 2011
This would make sense, but too bad the bucks and pacers also have only 5 “bad players” while the lakers are near the bottom with 9 “bad players”…..yeah the lakers are clearly worse than the bucks and pacers. The bucks and pacer would also be better than the spurs/celtics/heat and other contenders. You’re reaching with this.
Silver
April 9, 2011
Wow, I don’t think I’ve ever seen so much nonsense in all my days.
God, you stats guys are ridiculous.
Head Screwed On
April 9, 2011
Absolute bull-s***! Stop playing with the meaningless numbers and start watching the games. Rose dominates day in, day out and remains humble and motivated. No one in their right mind would say that the Bulls would be where they are this season without Rose! That in no way takes away anything from the other “good” players in the team, last time I checked, basketball was a team game, one player can only do so much!
Lance Russell
April 9, 2011
Anyone who has watched most of the Bulls games this year could tell you that their second 5 routinely outperforms their opponents’ second 5. Simple plus/minus or just an eye test makes that conclusion obvious. The rest of your analysis is noise. Further, any analysis that shows Luol Deng is playing at a worse level this year than last is clearly faulty. This is his best year as a pro, by far, when taking offense, defense, minutes and leadership into consideration. Basketball people get this.
kidejo
April 9, 2011
“An amazing thing stood out to me – the Bulls don’t have bad players!”
Well, I have to say that your article and your analysis actually support the idea that Rose is the MVP on the league.
Almost none of the Bulls players on the roster (except Rose) can create scoring opportunities for themselves… Yet, Rose with his aggressiveness and play-making abilities make all those players around him productive or “good/not bad”.
Statistics means nothing without proper context…
Keith
April 10, 2011
A fundamental assumption of the stats you are using is that each minute of playing time contributes equivalently to a given win/loss. Anyone who’s ever played a minute of crunch time knows this isn’t a valid assumption. Rose’s true impact is not measured in the first 33 minutes of his playing time each game, but in the last 5.
That is part of the reason for the 20-22% error in using the wins produced numbers for the Bulls players from 2009-10 (which sum to 51) to predict how they would fare as a group this year – when they are going to win at least 61 games. The predictions made by Wins Produced statisticians that Miami would clearly be the #1 seed and Cleveland would be in the playoffs prove the lie of this form of statistics.
reservoirgod
April 10, 2011
Italian Stallion:
There are two ways to look at wins produced. You can view it at the season level (as Dave does) or at the game level (which I do, see http://miami-heat-index.blogspot.com/p/heat-produced.html). What happens at the game-level is that in order for a player to be productive in a game, they must perform better than the average at their position(s) played in that game.
For example, LeBron’s Win Score per 48 mins is 14.7 this season and he’s played ~39% of the available minutes at SF in Heat games (there are 96 minutes available at each position in a game that doesn’t go into OT). The avg Win Score at SF in Heat games is 9.4 because opposing SFs have only managed to produce a Win Score of 5.8 per 48 minutes. The avg Win Score at SF is around 7 or 8 (see http://wagesofwins.com/posavg.html) but it’s much higher in Heat games due to LeBron.
I believe the opposite would happen w/ a team composed completely of players that were “negative win producers” in the previous season. Their poor productivity would lower the average at their positions to a point that their production wouldn’t be negative. It might not be average, but I don’t think it would be below zero for the season.
christian h.
April 10, 2011
I find it very amusing how worked up the people who religiously believe in stats’ power to compare apples to oranges get about the MVP issue. (In fairness I also find it amusing how worked up fans get about it.) Here’s my take as someone who has the professional mathematician’s healthy scepticism of statistics: Statistics CAN be used to, for example, prepare your team how to defend a given player (should you try to force him to go right or left when attacking the basket, that kind of thing). They may be used to decide which team to place a bet on or where to put the line if you are a bookie. Statistics SHOULD NOT be used to rank players (talking about basketball here, but of course nobody in the MLB stats community would be stupid enough to try and use stats to compare Barry Bonds and Randy Johnson – or so one hopes), current or historical. The whole idea of a “best player” is utter nonsense. So that means something like MVP will always be a popularity contest. Basing it on story line is no less valid than basing it on complicated-looking formulas or whatever other approach you might come up with.
art
April 14, 2011
Insightful and intelligent article.
The HC Thibs is the real MVP. The Bulls have the best bench in the L, which saved them a lot of games. They are also one of the most health teams in the L (Noah and Boozer never injured at the same time. Taj, KT and Asik can start on many teams, eg, the Hawks, Wiz, Nets, etc.). The Heat starting C and 6th man missed almost the entire season; The C’s has a lot of injury; the Blazers’ starting C and SG are severely injured, etc. And this is the first year that Deng has not missed a single game (in the past Deng often missed significant numbers of games due to injury).
They benefit by playing in the weakest division in the entire L (had they play in the Spurs/Mavs division or Heat/Magic division…).
They also benefit from some favorable schedule; e.g., they got playing quite a few teams that played on the second game of a back-to-back that needed to travel from West to East (or East to West) (I’m not sure if people who made the schedule did on purpose). Not all second games of a back-to-back are the same. It seems that teams can win the second game of a back-to-back if they travel within the conference (more than 50% of time). However, if a team needed to travel to another conference on the second game of a back-to-back, they always lost (at least in this year).
Mike
April 15, 2011
Joakim Noah is credited with a .279 WP48 this year (far better than anyone else on Chicago’s team), but Chicago went 36-12 (.750 win%) with him in the lineup and 26-8 (.765 win%) without him. interesting…
JJ Smith
May 14, 2011
This article is a waste of time…and whoever thinks Love is more of an MVP than Rose is too. Is Love in the playoffs? Isn’t it easier to score more and get more rebounds on a shitty team? you guys are just funny