The Tragedy of Kevin Love

The following post is from Andres Alvarez. And since Dre gave us Automated Wins Produced, it is these numbers he uses to tell the story of Kevin Love and the Minnesota Timberwolves.

For over half the league the season is still going on. For a select few the season is over. For an even more select few the season never really began. The Minnesota Timberwolves season can be summed up as terrible. To put a positive spin on this I can say they “exceeded” my expectations. Before the season started I recall nodding my head to an article by Ty over at Courtside Analyst about how this season had to be better for the Wolves than last because it couldn’t possible get any worse. I also recall knowing Kevin Love was a very good player and somehow he blew my expectations away. Let’s take a look back at the Wolves.

Kevin Love was good but his team was terrible.

Maybe you’ve had some time to play around with the new Gadget Arturo Galletti came up with. I used it to make a pretty nice picture of how the roster of the Wolves looked this season.

2011 Minnesota Timberwolves Wins Produced vs. WP48

That’s a player’s Wins Produced vs. their Wins Produced per 48 minutes. If a player is in the top right it means they’re at least as productive as an average starter that played 2000 minutes. We notice Kevin Love is amazing by this standard, Ridnour was about spot on average and the rest of the team was terrible. Here’s the same data in table format

Table 1: 2011 Minnesota Timberwolves Wins Produced

Name	Pos	G	MP	WP48	WP
Kevin Love	4.3	73	2611	0.465	25.3
Luke Ridnour	1.0	71	2159	0.109	4.9
Anthony Tolliver	4.1	65	1362	0.081	2.3
Martell Webster	2.1	46	1094	0.077	1.8
Anthony Randolph	4.3	23	463	0.057	0.5
Lazar Hayward	2.5	42	419	-0.002	0.0
Maurice Ager	1.4	4	29	-0.016	0.0
Michael Beasley	2.9	73	2361	-0.004	-0.2
Kosta Koufos	4.4	39	336	-0.045	-0.3
Sundiata Gaines	1.1	8	65	-0.190	-0.3
Corey Brewer	2.2	56	1362	-0.018	-0.5
Sebastian Telfair	1.0	37	711	-0.055	-0.8
Wayne Ellington	2.0	62	1181	-0.042	-1.0
Wesley Johnson	3.1	79	2069	-0.034	-1.5
Jonny Flynn	1.0	53	983	-0.113	-2.3
Nikola Pekovic	5.0	65	887	-0.131	-2.4
Darko Milicic	5.0	69	1686	-0.074	-2.6
Grand Total		82	19780	0.055	22.8

They were Unclutch or Unlucky

Wins Produced is built on point differential (more specifically, efficiency differential). Essentially when we look at how much they were outscored by night in and night out we’d expect them to be a 23 win team. Yet somehow they only won 17 games. A fun stat to look at is how did the Wolves do in close games?

Table 2: 2011 Minnesota Timberwolves record in close games

Margin	Wins	Losses
Over Time	0	4
5 Points	0	3
4 Points	1	1
3 Points	1	5
2 Points	1	2
1 Point	1	1
Total	4	16

The Wolves managed to have an abysmal 4-16 record in games decided by 5 points or fewer (or OT). It’s not as if winning a few of those close games would have changed the Wolves season, and in fact it may be the opposite of what a lottery bound team wants. Still I hope it makes a few Wolves fans happy to know their team wasn’t that bad.

Employing — and then playing — the wrong players

When Love went down a funny thing happened. The team didn’t play amazingly worse. We can point out after Love’s injury on the 20^th of March, the team did not win a game for the rest of the season. This actually helps illustrate two great problems with the Wolves

Table 3: 2011 Minnesota Timberwolves record with Kevin Love

Name	Pos	G	MP	WP48	WP
Kevin Love	4.3	73	2611	0.465	25.3
Luke Ridnour	1.0	62	1853	0.116	4.5
Anthony Tolliver	4.1	56	1063	0.070	1.6
Martell Webster	2.2	37	871	0.063	1.1
Anthony Randolph	4.2	14	218	0.001	0.0
Maurice Ager	1.4	4	29	-0.016	0.0
Lazar Hayward	2.5	37	378	-0.006	0.0
Sundiata Gaines	1.1	8	65	-0.190	-0.3
Kosta Koufos	4.4	39	336	-0.045	-0.3
Wayne Ellington	2.0	54	1042	-0.017	-0.4
Corey Brewer	2.2	56	1362	-0.018	-0.5
Michael Beasley	2.9	64	2040	-0.014	-0.6
Sebastian Telfair	1.0	37	711	-0.055	-0.8
Wesley Johnson	3.1	70	1822	-0.036	-1.4
Nikola Pekovic	5.0	57	746	-0.122	-1.9
Jonny Flynn	1.0	45	859	-0.123	-2.2
Darko Milicic	5.0	65	1590	-0.075	-2.5
Grand Total		82	17595	0.059	21.6

Table 4: 2011 Minnesota Timberwolves record with out Kevin Love

Name	Pos	G	MP	WP48	WP
Anthony Tolliver	4.3	9	300	0.118	0.7
Martell Webster	2.1	9	223	0.133	0.6
Anthony Randolph	4.4	9	245	0.106	0.5
Michael Beasley	2.9	9	321	0.062	0.4
Luke Ridnour	1.0	9	306	0.065	0.4
Lazar Hayward	2.5	5	41	0.039	0.0
Wesley Johnson	3.3	9	247	-0.019	-0.1
Jonny Flynn	1.0	8	124	-0.042	-0.1
Darko Milicic	5.0	4	96	-0.058	-0.1
Nikola Pekovic	5.0	8	142	-0.179	-0.5
Wayne Ellington	2.0	8	139	-0.227	-0.7
Grand Total	3.0	87	2185	0.027	1.2

Much of the time Love was out, so was Darko. With their best and worst players off the floor, Rambis had to play Tolliver and Randolph more minutes. These two players managed to play average in their increased time. Even with that in mind, the Wolves played less than half as well without Love than with him. With David Kahn — the team’s general manager — emphasizing players like Johnny Flynn, Michael Beasley and Darko as the future of the team, it is unlikely Love will get the support he needs. In other words, just like we saw with Kevin Garnett, it looks like we once again see a dominant player fail to win because his supporting cast is so poor (in other words, just like the Tragedy of Kevin Garnett we now have the Tragedy of Kevin Love).

David Kahn is also not the only problem. We can see that Rambis insisted on playing many of the poor player he was given, and left more productive players on the bench.

Rambis may be gone, and we can hope that maybe a new coach will play the right players and make the Wolves semi-respectable. If I was a Wolves fans, though, I would probably not be optimistic. After all, Kahn would have to choose the next coach (and he hasn’t done so well in the past when he had to make a choice).

-Dre

32 Responses to "The Tragedy of Kevin Love"

diehardNFFLbarnone says:

April 22, 2011 at 7:33 pm

It might be time to do another post about NBA coaches since the Rockets let Rick Adelman go on Tuesday.
Mike says:

April 23, 2011 at 12:39 pm

So according to wins produced:
- Kevin Love has won 25 games when his team has won 17.
- Anthony Randolph is useless when Love is playing (.001 WP48) but average (.106 WP48) when Love is out of the game.
- Anthony Tolliver becomes half a win better per 48 minutes (.070 WP48 vs. .118) in Love’s absence.

Isn’t this all persuasive evidence that the impact of Love’s rebounds might be overstated – that at least some of Love’s rebounds are taken from other player on his own team?
Crow says:

April 23, 2011 at 12:44 pm

Need to play the right players in the better combinations.

Team +/- this season:
Love without Tolliver -7 per 48 minutes.
Tolliver without Love -7 per 48 minutes.
Love with Tolliver +6 per 48 minutes.

Love with Tolliver only got 366 minutes, or about 6 minutes per game. There may be some fortuitous randomness in this small sample result but the difference between together and not is so huge I would try to ride it a lot more often, say at least 15-20 minutes per game. If it doesn’t work, well, little if anything lost. If it does work, great. And pretty easy to come up with that strategy.
Thomas says:

April 23, 2011 at 1:10 pm

Mike:

You are undeniably right but that goes against message. The message on this site for years has been “scoring is overrated, scoring is fungible, GMs/media/fans only care about scoring, watch as I shred this overrated scorer for not really contributing to his team’s win, my stat wins again!” This year’s big losers, Rudy Gay, Carmelo Anthony.

Diminishing returns exists for ALL things in basketball. Not only is shot creation mostly a myth–with most shots being taken away from your teammates instead of truly being created, except in the case of guys like Kobe/Lebron–defensive rebound creation is mostly a myth also. The dirty secret of the NBA is that very many defensive rebounds are uncontested and that gathering them in doesn’t really contribute to winning in any sense. Any member of your team could get that one, in fact if I suited up I could have grabbed that one despite not having a single noticeable basketball skill and being 6’1″” in shoes. In fact, even if nobody on your team gets it, possession (the actual value of the defensive rebound) would still be gained since the other team was completely disinterested and would have let it go out of bounds.

A lot of Kevin Love’s “value” would instantly go down if he was paired in the frontcourt with another strong player instead of with arguably the worst player logging serious minutes in Darko. It’s similar to what happened to Landry Fields after Carmelo came (Melo took a lot of the rebounds in the spots Fields was skilled in getting them, hence Fields’s productivity took a dump) or what happened to David Lee playing near Biedrins, another skilled rebounder. It would be nice to have this acknowledged by Berri at some point, but don’t hold your breath.
dberri says:

April 23, 2011 at 1:16 pm

Thomas and Mike,
Haven’t figured out how to install the filter that forces people to read the FAQ page before commenting.

Just to note again (something I have said for five years)…
- diminishing returns exists
- I have estimated the size of the diminishing returns effect
- the estimate involved doing more than citing anecdotal information (I actually looked at a 30 year sample of data
- the effect is not “large”
- I have re-estimated WP with this effect for defensive rebounds included and it didn’t change the basic story

Again, all this was said before.

We have seen that each of you can type. Now let’s see if you can read.
Thomas says:

April 23, 2011 at 1:22 pm

dberri:
So do you confirm or deny that some of Love’s amazing productivity would decrease playing alongside a center active in gathering in defensive rebounds such as Marcus Camby or Reggie Evans (the only two players in the league with a higher DRR% than Love at 35.2 and 34.4) instead of Darko, #50 among qualified centers.

Indeed it isn’t even a question that his WP WOULD take a hit, it’s merely a question of how much of a hit would it take. I’m suggesting that the hit would be quite large indeed meaning that a lot of his productivity comes from the startling lack of productivity his teammates are offering with him on the court which is exactly what Mike is saying. That his actual worth is grossly overstated.
dberri says:

April 23, 2011 at 1:30 pm

Thomas,
Read the FAQ! Stop typing and go read.
Italian Stallion says:

April 23, 2011 at 2:05 pm

Would anyone be willing to recalculate WP using .5 as the value of a defensive rebound and 1 as the value of an offensive rebound so we could see a comparison for this year?

I do a calculation similar to Win Score for every player using those values but I don’t have defensive and other adjustments programmed.

I also don’t do positional adjustments (for other reasons)
dberri says:

April 23, 2011 at 3:13 pm

IS,
Did it for last year (see FAQ). Will do it for this year (but not today).
drryanpepper says:

April 23, 2011 at 3:47 pm

All of the haters who post on this forum act like everyone who “buys” wages of wins thinking (Dr. Berri being the greatest offender of this) is so unwilling to listen to other information or consider the possibility that wages could be inaccurate.

What inevitably happens is that Dr. Berri (repeatedly and with great patience) answers those objections with more hard data (ex. when you discount defensive rebounds at FIFTY PERCENT the story doesn’t change). It is amazing to watch over and over as these “objective basketball purists” who are so much wiser than the rest of us completely ignore the facts and keep spouting the SAME arguments.

I mean seriously guys, aren’t there other criticisms or concerns you could bring? Or is “defensive rebounds are overrated” the only thing you can think of? That one has been answered…repeatedly.
nerdnumbers says:

April 23, 2011 at 5:35 pm

A team doesn’t need to try for rebounds. Proven by the Timberwolves. (This was without Love in the game by the way)
Mike says:

April 24, 2011 at 10:14 am

@Berri

In your FAQ, you say that valuing DRB at 50% does not have much of an effect.

However, among the top 20 players in wins produced for the 2010 season, five of the players (Dwight Howard, Gerald Wallace, Marcus Camby, Carlos Boozer, and Lamar Odom) lose at least 3 wins using the new WP48. 10 of the 20 players listed lose at least 2 wins. And that does not include Troy Murphy who drops out of the top 20 under the new metric.

How many wins would Love be credited with with DRB discounted by 50%?
Crow says:

April 24, 2011 at 10:21 am

WP using .5 for defensive rebounds would probably affect the rankings of really strong and weak rebounders nearer the middle of the WP distribution by the traditional formula more than it does the very top guys who rebound well, often there for more than one reason (i.e. very efficient scoring, or limited shooting or high blocks or low turnovers or other stuff).

Spreading the .5 for defensive rebounds to team players based on minutes can be seen as effectively rewarding team shot defense equally. I’d use a version of WP with .5 for defensive rebounds to the individual and .5 split out to the team more.

There are different schools of thought regarding shooting efficiency and there probably always will be. Those who have rejected WP entirely tend to do so based on both the defensive rebounding credit issue and the interaction of the rebounding crediting issue with the shooting efficiency standard. WP with .5 for defensive rebounds would focus more of the debate on the proper standards for crediting and charging for high and low shooting efficiency.
Go Bruins! says:

April 24, 2011 at 11:08 am

I hope Love can play out his contract and sign as a free agent with a winner. Ditto for Blake Griffon.
marparker says:

April 24, 2011 at 8:58 pm

Instead of giving credit to the 90+% of players the model correctly rates some guys really feel the need to harp on the 10% or so of players who are under/over rated.
Philip says:

April 25, 2011 at 4:43 am

Thomas,

If there were “many uncontested” defensive rebounds, why are there so many players that have played on many different rosters, the Bargnanis and Songailias or the league, consistently post poor rebounding numbers? Shouldn’t they be getting all of those free boards?

Subjectively, I agree that it can sure like a lot of boards just fall into Howard’s hands without even having to fight for them. But it doesn’t happen by magic or accident – it’s because Howard is such a good rebounding etc is so good, that opposing players know they don’t have a chance of grabbing a rebound anywhere in his vicinity. It’s for the same reason virtually no one applies man-to-man full-court pressure to Chris Paul – there’s almost no chance of success.

But if journeyman Darius Songailia, he of the 4.4 DRB/36 min, is guarding you? Well, rebounding becomes a different prospect. Songailia hasn’t proven that he can prevent his opposing man from grabbing rebounds (quite the opposite in fact), so he doesn’t get those “many uncontested” rebounds you tout. Howard does. But again, they’re uncontested boards for a reason; they’re earned by virtue of Howard being an excellent rebounder.
Mike says:

April 25, 2011 at 11:38 am

Philip,

Are you saying that none of Howard’s rebounds are taken from his own teammates?
nerdnumbers says:

April 25, 2011 at 1:59 pm

@Mike,
Are you saying Dwight Howard doesn’t lose some of his rebounds because his own team mates take them?

Let’s just admit stats are dumb. We should stop using them and just watch every single game instead. Dave just put up this year’s numbers but clearly we need to fix that. At 1230 games at around 48 minutes each you can be done watching them all in about 41 days. Have fun!
Philip says:

April 25, 2011 at 2:06 pm

Mike,

Dwight’s teammates would probably rebound a bit better if they didn’t play alongside such a prolific rebounder, but the effect would be minor. Diminishing returns are minor for defensive rebounds and nonexistent for offensive rebounds.

FWIW, Hedo, Lewis, Carter, both Q and J Richardson, and Pietrus have all been the essentially the same defensive rebounders/min for their entire careers, regardless of their team. The only players I could find who’ve played significant (500+) minutes for both the Magic and other teams and who were significantly worse playing with the Magic were Chris Duhon and Gilbert Arenas. They both lost about 1 rebound/36 minute.

So perhaps Howard is taking rebounds from his teammates, but only from his backup PGs. I think Magic fans can live with that.
Mike says:

April 25, 2011 at 9:07 pm

@nerdnumbers
When did I say we should throw away statistics?

@Philip
Actually diminishing returns are pretty big for rebounds. Phil Birnbaum showed that for every rebound that a player gets about 2/3 of a rebound are taken away from his teammates (http://sabermetricresearch.blogspot.com/2011/01/do-players-steal-rebounding.html). Also, the variance in team rebounds per game is less than the variance in rebounds per game by the center alone.

Howard taking away teammate’s rebounds will not have much of a noticeable effect on any one particular player. If Howard “steals” one rebound per game from his teammates, it’s not like he’s stealing exactly one rebound from Arenas. He might be stealing .2 rebounds from Arenas and .8 rebounds from all his other teammates. No one particular teammate will look significantly different, but the cumulative effect is one stolen rebound.
dberri says:

April 25, 2011 at 10:26 pm

Mike,
That is just an awful study by Birnbaum. He then cites another awful study by Eli Witus (which I already refuted in the comments on a post on Aaron Brooks — I think). You have to do more than just look at one season and look at a correlation. My study — which looked at thirty years of data and controlled for a variety of other factors — doesn’t find that large of an effect (although there is a diminishing returns effect).

Here is the problem. So many on-line studies are done by people like Birnbaum and Witus, who simply do awful work. And then people like you go around citing these studies as if they said something. Unfortunately, I can’t teach Birnbuam, Witus, and you econometrics in an blog post. So I guess we just have to live with nonsense.
Philip says:

April 26, 2011 at 2:14 am

Mike,

If Howard’s rebounding doesn’t have any noticeable effect on any of the players he plays with (and all evidence points to this), doesn’t that seem to indicate that there diminishing returns when it comes to rebounding is small?

You are correct that the effect is cumulative, but again, it’s quite small. One rebound/game for Howard sounds about right.

Some players are more notorious for stealing rebounds from teammates. Camby is an oft-cited example (and I’ve seen evidence of this – on/off stats showing that his teammates’ rebounding improves significantly with him off of the court, though like all on/off stats I’m skeptical of their reliability). But judging by the strong consistency year-on-year of rebounding, players that have a significant negative effect on their teammates’ rebounding are outliers (if they exist at all). Rebounding struggles usually are a result of age and injury more than anything else.
Mike says:

April 26, 2011 at 5:55 am

Are there are any studies other than yours that show only a slight diminishing returns effect on rebounding?

If diminishing returns is so small then why is the variance in team rebounds per game less than the variance in rebounds per game by the center alone (at least for the season which Birnbaum studied)?
Mike says:

April 26, 2011 at 7:46 am

Philip,

The one rebound per game I cited was just a hypothetical.

If some players actually do steal more rebounds from their teammates than you would expect, would there be any way to tell who those players are?
Philip says:

April 27, 2011 at 5:15 am

Mike,

You could look at on/off stats for a player’s teammates, but I’d expect that those fluctuate a lot so don’t have much predictive value.

Alternatively, you could look outright replacement, either due to trade or injury. If a team’s rebounding doesn’t change with the replacement of a great rebounder by a poor one (or vice versa), that would point to the great rebounder “stealing” rebounds.

There are a lot of other factors that could account for this lack of change, however – age, injury, that player rebounding isn’t perfectly consistent, other roster or coaching changes, etc.

The best case scenario would be a great rebounder replacing a terrible (or vice versa) due to trade or injury, and the team’s roster remaining identical otherwise. The discrepancy would have to be huge to get any significant results.

Does anyone know a way to look at the difference for teams before and after the trades? Gortat replacing Lopez and Frye, or OKC drastically upgrading their frontcourt, are both suitable. Even then, it’s tough to draw anything conclusive. Gortat is about 2.5 boards/game better than Lopez or Frye, but even that could simply be other players incidentally slightly improving their own rebounding.
Mike says:

April 28, 2011 at 6:20 am

@Philip

You cannot simply look at one single example to show diminishing returns because any change could be attributed to small sample size, change in ability, etc.

The best way to measure the impact is to take lineup information from 82games.com and predict how each lineup would rebound compared to how they actually rebounded. If one lineup with two dominant rebounders rebounds less than expected, that could just be sample size. If every lineup with two dominant rebounders rebounds less than expected, that has to be evidence for diminshing returns.
dberri says:

April 28, 2011 at 7:31 am

Mike,
Why is the way I analyzed this not the “best” way?
Mike says:

April 28, 2011 at 2:02 pm

Where can I find your analysis?
Philip says:

April 30, 2011 at 2:44 am

Mike,

I agree that such a small sample isn’t reliable, but I’m also extremely skeptical of on/off stats. +/- just isn’t a reliable stat because of the high variation from season to season. I suspect the same may be true for rebounding, though I haven’t seen a study which addresses this.
Links: This Week Never Happened | Plexiglass Principles says:

May 1, 2011 at 6:30 pm

[...] Possibly more depressing than the Twins: Andres Alvarez at Wages of Wins on the Timberwolves. [...]
A Wolf’s Tale « says:

July 4, 2011 at 6:10 am

[...] Despite having an amazing increase in performance somehow Kevin Love’s team got much much worse. [...]
Rubio vs. Irving? Why not Pekovic vs. Varejao? | The Wages of Wins Journal says:

March 12, 2012 at 11:33 pm

[...] was a great second option behind Kevin Love. Last season the Wolves were Kevin Love, an average Luke Ridnour and not much else. This season the great Kevin Love and average Luke Ridnour returned. Luckily, they got help. When [...]