“The Black Swan Theory or Theory of Black Swan Events is a metaphor that encapsulates the concept that The event is a surprise (to the observer) and has a major impact. After the fact, the event is rationalized by hindsight.“- Wikipedia definition
Yesterday was all about perspective.
Today it’s all about context.
Really it all begins with the following questions: How good are every players teammates? What would happen if I removed each player from their team?
It would look something like this for the current season:
See I wasn’t looking for what I found. I was looking for the quality of every teams roster and the way I was doing it was by removing their best player.
Teams | Win Projection Team without best player |
CHI | 51.1 |
SAS | 50.9 |
BOS | 47.2 |
LAL | 47.2 |
DEN | 44.0 |
MIA | 41.7 |
DAL | 41.3 |
OKC | 39.5 |
HOU | 39.3 |
ORL | 36.4 |
PHI | 36.0 |
POR | 33.6 |
MEM | 32.5 |
NYK | 32.0 |
IND | 30.8 |
MIL | 30.7 |
UTA | 30.4 |
NOH | 28.2 |
PHO | 28.0 |
ATL | 27.8 |
GSW | 27.0 |
CHA | 26.9 |
DET | 24.5 |
SAC | 19.7 |
LAC | 18.7 |
TOR | 16.5 |
WAS | 14.6 |
NJN | 11.6 |
CLE | 7.0 |
MIN | -4.9 |
The answer was that Chicago and San Antonio have the best rosters followed by Boston and LA. But what caught my eye was the number on the bottom. Minnesota had a negative Win total for their supporting cast? Kevin Love’s team without him produces negative wins? That’s really strange.
How strange? One of my favorite posts dealt with the worst teams of all time (see here). Negative win production is exceedingly rare.
In fact it’s only happened seven times since 1978:
Name | Team ID | WINS | Rest of Team Wins | WP48 |
Michael Cage | 1987LAC | 14.9 | -5.5 | 0.245 |
Michael Cage | 1988LAC | 16.1 | -2.8 | 0.290 |
Mike Iuzzolino | 1993DAL | 2.9 | -2.3 | 0.078 |
Randy White | 1993DAL | 2.7 | -2.1 | 0.091 |
Sean Rooks | 1993DAL | 2.1 | -1.5 | 0.049 |
Terry Davis | 1993DAL | 1.4 | -0.7 | 0.027 |
Derek Harper | 1993DAL | 0.9 | -0.2 | 0.021 |
Michael Cage twice on the Clipps and our old friends the 1993 Mavs. Kevin Love is number 8 out of 14600 but really he would rank as the player with the 2nd worst teammates (or about 2/14600).
Teams | Win Projection Team without best player |
1987LAC | -5.5 |
2011MIN | -4.9 |
1988LAC | -2.8 |
1993DAL | -2.3 |
2000LAC | 1.3 |
1989MIA | 2.6 |
2000ATL | 2.7 |
1996PHI | 3.1 |
2010MIN | 3.5 |
1991DEN | 4.4 |
Yesterday we also said that he was in very select company in terms of Wins Produced.
In fact if we project him at 30 Wins Produced for the season, he’d be in the top 9. That’s 9 in 14600.
The likelyhood of a player having one of the top nine seasons while playing with the 2nd worst teammates? That works out to about 1 in ten million or once in every 20,000 NBA seasons.
Kevin Love’s 2010 season is a Basketball Black Swan Event.
What does it all mean?
Actually what it means is that it’s extremely unlikely that we will see something like this again. Kevin Love’s situation is thus very hard to read. I’ve done some work on Diminishing Returns (see here):
And this is good. By these numbers Kevin Love’s .480 WP48 would still be over .400 on any team. That would be enough under normal circumstances but we are dealing with a double outlier here so let’s break out a final table:
Note: I totally screwed up this table when I first put it up. In my defense it was really late when I made it. Fixed now.
Michael Cage is the best comparison we can make and we can see that as a good player on the Clippers he of course got traded to a contender and his numbers went down from his peak but stayed in the ballpark.
So Kevin Love is really good, regardless of the weird circumstance.
Some Dude
03/15/2011
So on one hand we can believe a once in a 20,000 season even is happening OR we can assume the player is being overrated by the metric.
Seeing as how it’s impossible for a team to win in the negative range, I find it odd you pick the first option over the second. Nevermind that you don’t throw in Berri’s own admission of DR from bad teams to good team, which is significant, into the calculations.
BTW, the above chart shows quite a large discrepancy between team WP and actual wins. In fact, only the 1st Clippers team and the Philly team are even in the right ballpark. It’s off by 10 wins at times.
arturogalletti
03/15/2011
SD,
I screwed up the final table. Fixed it now. In my defense I did it just before bedtime 🙂
Did you miss the handy dandy DR table I made by any chance (with a link to the post explaining the DR effect)?
K-Love numbers are inflated by being on a lousy team, just not enough to take him out of the .400 range so he still is crazy good.
Some Dude
03/15/2011
good to see the table fixed.
Yeah, I saw it, I don’t agree that’s what Berri found. It was an average, but i am certain the distribution of the data matters. Love is at the extreme margin here and thus would be affected a lot more. Look at Al Jefferson. You could say his WP48 went up by .046 or you could say it went up by about 50%. Using that figure, Love would be around .31 (33% going down). Adjusting for him playing more defense, I’d say he’d setting at the .25-.3 range for WP on an average paced good team. You might think this is crazy, but I bet I’d be more accurate in the end.
Also, they play at a very fast pace, which RB48 does not realize. I mentioned before that Love’s rebounding is inflated by ( I think) 2 rebounds compared to the average pace alone.
College Wolf
03/15/2011
This is preposterous. Clearly K-Lav is being vastly overrated.
I know we are bad and all, but are the Wolves without K-Lav that bad??? No. The rest of our team is better than the teams we have had the past two seasons, I am quite sure of that. So how is this year’s team rank as the second worst of all time? That makes no sense.
All I can guess is funky stats and/or analysis.
College Wolf
03/15/2011
P.S. “K-Lav” is how they spell K-Love in Serbia. In case you were wondering and don’t approve my comment because of it.
Chris
03/15/2011
Michael Beasley came to a worse team (MIN) and his production went down. Martell Webster also had the same results coming to MIN. Any thoughts on this?
In any case, the odds you’re quoting against this happening again is good news for the Twolves. This means the non-K-Love players on the team can’t possibly be as horrible next year. If K-Love still performs near his mark this year the wolves just might suck less.
Go Twolves!
arturogalletti
03/15/2011
Oh Chris. “They might suck less”. I feel for you. Khan is still in charge.
College Wolf
03/15/2011
Nice response Chris , lol. Let’s hope for that!
College Wolf
03/15/2011
P.S. arturogalletti – While I think these particular results are wrong or weird or fishy (or something?), it’s an interesting read none the less. I also enjoy your site thoroughly.
Nerd Numbers
03/15/2011
Arturo,
I have that Batman and now have to look up the context.
College Wolf,
The T-Wolves need to .500 to have a better record than 2 years ago. This team IS better than last year. The change? Kevin Love is playing better than last year and more minutes (8 more per game) and more games (don’t forget he was injured 22 games last year). As Arturo brought up in this week’s podcast. Love’s numbers are being compared to Moses Malone’s and Wilt’s! That’s insane. Moses Malone played on a bad team and won MVP on a “bad” Houston Rockets team that won 46 games. Keep that in perspective.
Some Dude
03/15/2011
Neil just did a post on this for Win Shares.
http://www.basketball-reference.com/blog/?p=9059
Love comes in at #4. And the numbers look, gasp, reasonable.
arturogalletti
03/15/2011
It’s in the difference in the calculation. WS has a significant team adjustment which tends to flatten out the players to the teams average performance. A player’s WS48 is going to look kinda like WP48 player times x/100 + WP48 times (100-x)/100 where x is some number between 50 and a 100. So KLove takes a significant hit because of where he plays.
Again he’s still really,really good.
Some Dude
03/15/2011
Yeah, I know the differences are in the calculation. What else could account for the differences? That’s like saying water is wet, Arturo.
I was just citing Occam’s Razor again.
Anyway, Love is all-star quality but not super-elite according to every metric I’ve seen but WP48. This is also what pretty much any NBA fan and GM and coach will tell you. Now, it’s possible WP48 is right and everyone else is wrong. But the odds of that being the case, well…let’s just say I wouldn’t put anyone’s house on it.
alvy
03/15/2011
EntityAbyss
03/17/2011
I always hear about eye test, and from watching the wolves, they are horrible outside of kevin love. Luke Ridnour is decent but everyone else is hard to watch. Inefficient shooters, bad rebounders (to the point that it seems like if Kevin Love doesn’t get the board, it won’t get picked up by Minnesota.
Also, as to why their point differential is similar when Kevin Love is on and off can be simply attributed to runs. From what I’ve watched, when teams go on runs, Kevin Love gets put back into the game immediately, but the other team is already on a run. In a game of runs, plus/minus doesn’t really show what’s going on. If Minnesota’s on a run while Love is off, the coach let’s the team go, but when the other team goes on a run, Kevin Love gets put in, and has to deal with that. Same with rebounding. If Kevin Love is out and Minnesota starts to get killed on the boards, he’s back in. All these on court / off court numbers don’t take into account that fact. Basketball is a game of runs.
Some Dude
03/17/2011
This reasoning makes no sense unless you believe “runs” are forces outside of the players’ control.
If Kevin Love is as good as you say he is, inserting him into the game should stop the run. And him being out of the game should prevent a run from even happening. This theory is absurd. Kevin Love can’t create runs for his team but someone the bench accidentally does it. And he comes in to “stop” runs and always fails and somehow this is out of his control.
Occam again strikes. What is the more likely answer? That Love’s impact is overstated because of his defense or that some wild and crazy “run” thing is doing something super flukey that doesn’t happen to anyone else?
Furthermore, a run is nothing but flipping a coin 10000 times and seeing streaks of 6 tails in a row. Each event has its own probability independent of the other ones.
You’re making up a BS excuse for statistics that don’t jive with you preconceived notions. This season Kevin Love has had no discernible impact on his team’s ability to win or lose games. You can try to make 100 excuses for this, but this is the truth.
I will repeat. if Love is responsible for his 144% of his team’s wins, then it would be nearly impossible for him to have a +/- of 0. Remember, every time he’s on the bench, according to Arturo, his team would produce in the negative win range. Yet, the numbers do not play this out.
entityabyss
03/17/2011
When players on the other team get hot, him coming into the game isn’t going to immediately stop it. Coaches tend to put their best players in the game when things start to get bad. It’s not ridiculous to believe that that’s the case. Kevin Love their best player, but plus / minus suggests that he has not much more of an impact as his replacement pf.
Think about it. If a team gets hot, and k love is on the bench, him coming in doesn’t make that team all of a sudden cold. The inconsistency in plus / minus stats suggests that maybe the results you get from it aren’t always true, unless you want to believe that orlando’s just as good a rebounding team without dwight howard.
Since basketball is a game of runs, players (at least no 1 player) can’t stop a team that’s hot at the moment. Also, what’s not taken into account is who’s on the court when kevin love is off. What line-up is the other team using. Kurt rambis never let’s it get too ugly when kevin love is out.
Some Dude
03/17/2011
Where’s the evidence of teams being “hot.” Hot Hand Theory stuff seemed to blow a lot out of the water.
Yes, I believe it about Dwight Howard. Why? When Howard is in, the other guys don’t crash the boards as hard. When he’s out, they crash harder. What you see, however, is that when Howard is in the game the team scores more and fast break more. That’s because 4 guys go to offense as Howard rebounds, whereas when he’s out say 3 dudes crash the board to get it. So the rebounding is the same as a team but the effect is different. THIS IS WHAT SIMPLE METRICS CANNOT GRASP.
1. You have to establish there is such a thing as “team runs” that cannot be altered by the other team’s personnel. I was data backing up your assertion.
2. Even if it was true, that doesn’t explain Love’s point differential issues. Over time, we’d still expect it to be positive? Why? because over time the runs against Love in the game HAVE TO BE less than the runs with him out of the game. If he’s so good, his team has to go on run with him out there and almost never when he’s on the bench. Secondly, if teams go on runs against Minn when he’s on the bench, him being thrown in the middle of it shouldn’t change much on that part. He should still be seeing a positive affect on the team.
I repeat. For Kevin Love to have no discernible impact and yet it be the result of this run BS would be a miracle. You could argue the runs, if true, skew the stats a little, but in now way could they end up the same. Not in a million seasons.
Arturo once converted WP into PD. According to that, the non-Love players should be a horrific team with a PD that hints at an inability to win a single game. And yet, the numbers play out the complete opposite. It says his teammates should win around 25 games out of 82, not -4 games.
We have actual data of Love’s teammates without him and it proves Arturo’s claim wrong, unequivocally.
entityabyss
03/18/2011
Ok, you seem to have this huge faith in plus / minus stats and believe that what’s being told by plus / minus stats are what’s going on. Since there isn’t too many teams having vastly different rosters year-to-year, do you happen to know the consistency of 5-man units from year to year? Is there anyway that could be done, because I’d like to see that. I personally don’t believe that 5-man units tell you much about a player, but you believe otherwise. What’s the consistency in point differential for 5-man units from year to year? I’d like to know.
Some Dude
03/17/2011
edit: I do not mean a +/- of 0. I mean a +/- splits of 0 (meaning the team fares just as well when he’s on the bench).
And let me say this. The team’s defense is SIGNIFICANTLY better when another PF is in the game, dropping from worst all time levels to just bad.
College Wolf
03/18/2011
Hahahahahaha the “runs” theory is borderline insanity!
dm
03/19/2011
I’m sorry Some Dude but I just don’t think I can invite you to me and Kevin Loves wedding.
You are going really miss a lot of great stuff, like when K Love gives three women concussions rebounding the bouquet.
Gil Meriken
04/05/2011
Did he get much better since college? Why didn’t his team with Russell Westbrook, Nick Collison, Luc Richard Mbah a Moute even make it to the NCAA title game? Not that you have to make it to the NCAA championship to be a superstar (far from it) but wouldn’t that be an indicator of Love’s superior effect on winning, especially considering his teammates? Or were they dragging him down too?
arturogalletti
04/05/2011
Gil,
They were all really,really young.
And I quote wikipedia:
“In the 2008 Pacific-10 Conference Men’s Basketball Tournament, the Bruins defeated the USC Trojans, featuring O. J. Mayo, in the semi-finals. Both Mayo and Love were nominated to the All Pac-10 tournament team. Later, Love guided UCLA to the regular season Pac-10 conference championship, the conference tournament championship, and a #1 seed in the 2008 NCAA Men’s Division I Basketball Tournament. Love helped the Bruins to the Final Four of the tournament, where they lost to the Memphis Tigers. At the end of the 2007–08 regular season, Love was named consensus first-team All-American, Pac-10 Player of the Year, and Pac-10 Freshman of the Year. He led the Bruins with 17.5 ppg, 10.6 rpg, and 23 double-doubles.[24]”
As a freshman. Yeah.
reservoirgod
04/06/2011
Gil:
Two words explain why Kevin Love & UCLA came up short vs. Memphis in the Final 4: Joey. Dorsey.
It’s a perfect example of the problem w/ Kevin Love. Strong, athletic players give him a lot of trouble. Dorsey outrebounded Love on the offensive glass 6-2 and 15-9 overall. It’s a perfect illustration of why I said on last weekend’s podcast that I would take Griffin over Love & why Coach K chose to go w/ Iguodala over Love at PF. His lack of athleticism is a problem. He’s a very productive player but he does have that weakness.
julienrodger
04/21/2011
Anybody who calculates Kevin Love was worth 30 Ws this season and is a top 9 guy since 1978 needs to retire from NBA analysis and go into MLB sabermetrics
Sorry man, but at some point you have to step back and realize “WTF did I just write? Maybe my formula is wrong…” I’d say the same about your numbers saying the Heat win 58 without Bosh or the Raptors win 34 without Bargnani. When your numbers are making statements that clearly wrong, you have to fix your formula. I’m all for advanced basketball stats but not when they come up with something this retarded
I never thought that this: http://www.nesn.com/2011/01/2011-red-sox-will-challenge-1927-yankees-for-title-of-greatest-team-in-major-league-history.html would be the 2nd most embarrasing article I read this week
arturogalletti
04/21/2011
julienrodger,
It’s 25 Wins for Love. Comparable examples show that we can expect him to continue to be exceptional. 22 year old who puts up historic rebounding season and is offensively efficient. I’m willing to be on record with my admiration.
For Bosh and the heat, Bosh produced about 8 wins for them (which is good not great). No Bosh? About 50 wins (if no Haslem as well).
Bargnani really, really sucks. Raptors would be better off without him and he will make any team he goes to worse if he plays for them.
But we will see.
julienrodger
04/21/2011
Can you explain why Bosh’s number is 50 instead of 58? I read the chart as 58 for the rest of the team which would presume 0 for him. Are the Heat projected at 64 from point differential?
Ditto for Love, it sounds like near the bottom of your article you project him at 30 WP.
arturogalletti
04/21/2011
Love got hurt so his projection is a little off from the actual.
Bosh was actually slightly below the level of the rest of the team. Miami was projecting at about 58 wins at the time. Bosh was not impacting that team significantly. He did pick it up at the end of the season. I guess he just needed a good cleansing cry 🙂