Patrick Minton was born and raised in Minnesota. After graduating with a degree in Psychology, he travelled abroad in Germany. In Germany he did a number of odd jobs to pay the bills, including coaching basketball at all levels. He says he has always been a numbers freak, going so far as to enlist friends to score-keep the games he coached, keep shot charts, track turnovers, rebounds and assists. In his late twenties, he finally turned one of his hobbies into a ‘career’ and became a software engineer at Amazon, where he worked for quite a while before taking a gig at Microsoft. In 2007, he did an MBA at the University of Washington and discovered a love for economics. Scouring the internet for economics blogs, he came across the Wages of Wins Journal. He says he always had a coaching philosophy similar to the tenets of the WoW story (scoring totals don’t matter much, rebounds and turnovers are really important, etc), but this was the first time he had seen someone quantify this philosophy.
In evaluating the Minnesota Timberwolves 2009-2010 season, two thoughts become prominent. First, it appears that the organization has some large misconceptions about who their “star” player is. Additionally, it seems the organization is having as much trouble surrounding today’s “star” as the McHale-run front office had surrounding Kevin Garnett with enough talent to compete for championships. This may be all part of a ‘rebuilding’ plan. Or, it may be that David Kahn is reluctant to rely on advanced statistics.
Of course, the writers and fans of this site are big proponents of using advanced statistics such as Wins Produced to evaluate players. With that in mind, let’s take a look how the Minnesota Timberwolves players performed according to Wins Produced in 2009-2010:
All You Need is Love?
As we can see, although Al Jefferson is often referred to as the “franchise” player in Minnesota, the Timberwolves best player (according to Wins Produced) was far and away Kevin Love. When we look at Kevin Love’s box score numbers, the reasons are obvious; although he is a bit below average in shooting efficiency from the field, blocks, and turnovers; he makes up for this by committing fewer personal fouls, creating more assists, and by being a rebounding machine. In 2009-2010, Kevin Love grabbed 18.4 rebounds per 48 minutes — a mark that not only far surpasses the average power forward’s 11.6, but also makes him the best rebounder in the entire NBA (unless we count Joey Dorsey, who managed to grab 43 rebounds in a small sample of 106 minutes). Of course, the observation that Kevin Love is a good rebounder is a familiar story. But it’s interesting to put in perspective just how good he is; of the top 5 rebounders per-48 minutes in the NBA only Love and Mr. Dorsey play the power forward position; Dwight Howard (18.3), Marcus Camby (18.1), and Samuel Dalambert (17.7) all primarily play the center position (at least, historically).
So Wins Produced suggests that Kevin Love is not only the best player in Minnesota, but possibly one of the best players in the NBA. It has been suggested before in this column that the threshold for stardom in the NBA is a WP48 of .200 (a .200 player produces twice as many wins as an average player per 48 minutes) and that the threshold for “superstardom” is .300. For the record, only 9 players with more than 1,000 minutes had a WP48 above .300 in the 2009-2010 regular season: Kevin Love, Marcus Camby, Lebron James, Dwight Howard, Jason Kidd, Chris Paul, Gerald Wallace, Pau Gasol, and Tim Duncan. That is elite company, and would seem to provide the Wolves with an elite building block, similar to when they had Kevin Garnett.
The Rest of the Wolves
Unfortunately, to win in the NBA, you need more than love, and Kevin Love’s teammates didn’t contribute much. Only Al Jefferson was significantly above average (although Al would probably have posted a higher level of Wins Produced if he had not spent a good deal of time at the center position), and only six Timberwolves managed to post Wins Produced numbers in the positive range. In fact, therein lies the Timberwolves real problem: many of Kevin Love’s teammates didn’t simply fail to contribute wins; they actively hurt the Timberwolves with production in the negative range. Table 1 suggests that if every player other than Love, Jefferson, Gomes, Sessions, and Wilkins had been replaced by players who contributed literally nothing to wins (players with WP48 of 0.000), the Timberwolves could have won an additional 11 games. Yes, switching out those players for a bunch of worse-than-average players would have been an upgrade.
Of course, one might argue that the Timberwolves office is aware of this, and did not care. Much has been made of the fact that this past season was a “rebuilding” year and Minnesota faces the current free-agent class with copious cap space. But then, if this is true — and considering the importance that has been placed on salary cap space in the 2010 free agency market — one might wonder why the Timberwolves signed Ryan Hollins to a two-year deal. Hollins has the dubious distinction of being the worst player in the league (according to Wins Produced). Additionally, why did they chose to pick up the option on Corey Brewer (who is not very far ahead of Hollins with respect to Wins Produced)? Is the 2010-2011 season also a “rebuilding” year?
Going Forward
In the coming offseason, the Timberwolves have some interesting choices to make. Al Jefferson and Kevin Love both play the power forward position, and Kurt Rambis (the team’s head coach) was often reluctant to play both players together. In fact, for much of the past season, Rambis didn’t start Love or give him the most minutes per game (leading to further speculation that management disagrees with us about who the “star player” is). At the moment, it looks very much as though the Timberwolves might end up drafting DeMarcus Cousins, a very productive college player. If Cousins turns into a productive NBA player, the Timberwolves could replace Ryan Hollins’ minutes with his, and improve substantially. What other moves might management make? Only the summer will tell (and when it tells us, there will be another post on the Minnesota Timberwolves).
– Patrick Minton
The WoW Journal Comments Policy
Wages of Wins Journal 
robbieomalley
June 5, 2010
L is for the way you look at me
O is for the only one I see
V is very, very extraordinary
E is even more than anyone that you adore
Sisyphus
June 5, 2010
This is Team X. Player X has a WP48 of 0.XXX and played XXXX minutes while Player Y played XXXX minutes while having a WP48 of 0.YYY. Team X would be better/worse if some of Player X’s minutes went to Player Y because of his higher/lower WP48.
This isn’t analysis, it’s just a regurgitation of the database. I think this blog’s time is over.
Dre
June 5, 2010
Patrick,
Excellent article. I remember being confused when I saw Al and Love contributed 120% of their team’s wins.
So Orlando has had two of the best Centers and made it to the finals with both. Minn may get the honor of having two of the best PFs(if Love keeps this up) and never going anywhere.
Sisyphus,
Gotta disagree. This blog is seeing a huge influx of trolls and old arguments. Although I do like that template! Follow that with a theory why the team did that and I will read that article every day of the week.
Marparker
June 5, 2010
apm vs. wow,
Nerd statistician blog’s version of east coast vs. west coast hip-hop battles in the mid 90s.
I think the new availability of the database combined with multiple writers is a very nice touch.
Just my random thoughts for the day.
Thomas
June 5, 2010
Sisyphus:
This blog has been masturbatory for quite awhile and is starting to edge into Hollinger-land. The only “analysis” that this blog accepts is WP-related, just like the only thing that Hollinger will accept is PER related. Neither have much room for common sense, I’m afraid.
It’s actually kind of funny that these blogs call themselves the “advanced stat community” in the first place. The actual decision makers in the game are years beyond the analysis being offered here. The constant circlejerking on something that’s quickly becoming antiquated is only widening that gap.
dberri
June 5, 2010
Thomas,
Just wondering…how many actual decision-makers from the NBA do you actually know?
Arturo
June 5, 2010
Thomas,
I’ll point out that you can statistically be proven to be wrong. Here’s the correlation between Real wins and Payroll:
Year Correlation (Wins vs Pay)
2002 0.6%
2003 12.7%
2004 5.5%
2005 1.1%
2006 13.5%
2007 18.5%
2008 7.4%
2009 4.1%
2010 17.5%
Total 6.4%
Your statement:”The actual decision makers in the game are years beyond the analysis being offered here.” does not hold up to statistical review. In other words: Cool story, bro.
ilikeflowers
June 5, 2010
Arturo,
Clearly Thomas is referring to the Illuminati who run the NBA. Their goals are not the same as of common men and it is not for us to understand them. The Illuminati’s plans unfold in cycles of prime years and are immune to numerical analysis. During my decades of tracking them all I know is that there is some connection between the leadership of the free world, the distribution of wins vs payroll in the NBA, and the number 42.
// Thomas – masturbatory and circlejerking in the same post. Interesting.
Joe
June 5, 2010
The actual decision-makers of the game are using stats that are far more advanced and relevant than the ones proffered here. Stats from Synergy Sports, Basketball Reference, 82games. Not stats that say Jamario Moon is more productive than Kobe Bryant. Even PER is better than this garbage.
Please alert me to a single front office that takes “wins produced” seriously.
ilikeflowers
June 5, 2010
Cause that explains Arturo’s post perfectly.
Joe Da Troll, epic FAIL. Here’s your sign.
dberri
June 5, 2010
I will ask Joe the same question… how many decision-makers in the NBA do you know? And if the answers is “none” (as I suspect), then how do you know what models people are using?
Schermeister
June 5, 2010
I have been reading this web page for some time and want to say keep up the good work. I also love the new posters which means much more posts to read!!!
As for the timberwolves. I find the way the run the team extremely disturbing and wonder if Love will even get half the chance to play given there incredible prefance to play low productive but media hyped players like Flynn and Brewer. This post seems to reinforce that belief.
It doesnt even take hardcore statistics to see that Hollins, Brewer and Flynn were 3 of the worst players to log serious minutes. And they all on the saem team. I mean hollins is 7 feet and grabs ~7 rebounds per 48 minutes.
At first I thought rebounding was there problem, but not so according to the nba season numbers. FG % and shooting eff is there real problem. Love seems seems to complete fix there rebounding woes and makes them just a shade under average.
Minnesotas team points per shot is only 1.16. NJ is 1.15 for example.
Now it seems they have no eff scorers to move time to or am I missing someone. If they dont have eff scorers maybe they can grab one is Free agency?
As for all the negative trollers here, post some legitamite numbers please. The owners arent using anything above and beyond anyting. I hardly doubt the wolves and clippers have been using something so advance that my small brain cant understand. NJ really new what they were doing this year too.
tywill33
June 5, 2010
A friend of mine who works in the Timberwolves front office told me they are indeed shopping Kevin Love, or will shop him, hard this offseason.
I don’t know what the connection is, but he followed up that tidbit by telling me that the Timberwolves are hot for Rudy Gay of Memphis.
When I suggested to him that he could advance his career by championing the argument against this ludicrous personnel direction, he first commented, “Oh yeah, that’s right… your big on Kevin Love” in the same tone someone would comment, “Oh I forgot you were a socialist” and then he went on to further suggest that my advice amounted to career hari-kari.
That franchise cannot get out of its own way.
dberri
June 5, 2010
Following up on Tywill33… in my contact with people in professional sports across the years, it does not seem to me that you get very far in sports by questioning the wisdom of the people who ultimately make decisions. Your best bet is to do what you can to get along with the people who can fire you whenever they like. In other words, there is no system of tenure in sports.
James
June 5, 2010
Great post, Patrick!
Minnesota is a brilliant example of a team whose front office has no idea how to measure the productivity of its players. They’re sitting on a gold mine in Love, yet refuse to give him the big minutes he deserves. Conversely, they give a great deal of minutes to players who produce very little (Brewer & Gomes). Yet, they don’t seem to realize what the problem is. I feel sorry for Wolves fans, it looks like they probably have many more years of bad teams to look forward to.
Arturo
June 5, 2010
Joe,
First:
If I concede your point that their metrics are more advanced I would logically expect to find correlation between payroll and wins as I would expect advanced relates to locating winners. We don’t, therefore the existing decision making model (the Yay!Points! model) is not effective.
Patrick Minton
June 5, 2010
Sisyphus,
As others have pointed out, “regurgitating the database” isn’t without value, since not everyone actually studies ‘the database’.
Thomas/Joe,
Many decision makers are actually publicly on record that they do not, in fact, use any such statistical analysis. In fact, it appears to be a point of pride with many front offices.
Love’s value can be dumbed down to the following argument: You get to shoot more often when he’s on the floor, and your opponents get to shoot less often when he is on the floor. Yes, I am aware that the same argument can be made regarding Dwight Hoard and Marcus Camby. However, those players are highly-regarded for a different reason: everyone looks at them and recognizes their freakish athleticism, and “KNOWS” that they are good players without having to resort to such skullduggery as box-office statistics.
Scondren
June 5, 2010
Arturo- While you’d probably expect the correlation to be higher, I doubt it will ever be super high. Injuries and age can really change productivity, whereas contracts are longer term and guaranteed. I think Elton Brand is the poster boy of this; I believe his WP was pretty good before falling off a cliff with Philly due to injuries (age).
Rookie contracts probably mess it up also.
There is probably a way to work around that, but I’m not smart enough to think of it.
Schermeister
June 5, 2010
Cavs should be interested in Kevin Love.
That would be a major land for memphis. Though would they play Love and Gasol at the same time. I am totally convinced that if Love gets more minutes people will finally come around to thinking he is a great player(all star)
Marparker
June 6, 2010
I’m convinced that there is a correlation between arm muscles and playing time.
Thomas/Joe,
In this instance if I were a T-wolves fan I would at least want the decision makers to try everything. They have not even given Love a chance which doesn’t make much sense considering how high of a draft choice he was.
You can hate WoW or its “followers” all you want but the real issue is that Minnesota won’t even give Love a chance. That’s all I am saying.
SadWolvesFan
June 6, 2010
Poor David Kahn. The link illustrating his aversion to stats shows that he recognizes the importance of acquiring a top-notch difference maker, even as he lacks the ability to properly identify the difference maker right under his nose. I pray he reads this article.
A word in defense of Corey Brewer: Of course WoW will hate him for his inefficiency. Since they’ll never put up great rebounding numbers, WoW will hate any wing who is inefficient. It will also continue to underrate defense, since the obvious defensive stats (defensive rebounds, steals, blocks) are all it can use. Corey Brewer is first and foremost a defender, and his shooting efficiency has improved by leaps and bounds both years (TS%: .429, .473, .502). WoW can’t quantify this, but his playing style is such that he’d contribute more to a winning team than a piss poor one like the Wolves. He doesn’t deserve to be lumped in with Hollins and Flynn, who are rightly terrible.
To use a different metric, 82games.com shows him to have been a net neutral for the Wolves while still showing Hollins and Flynn to be godawful.
http://www.82games.com/0910/09MIN5.HTM#onoff
Leon
June 6, 2010
You don’t need to have watched much basketball to see that Hollins is bad. BAD.
palamida
June 6, 2010
SadFan, I find it odd you would link to that 82games page and I’ll tell you why:
Personally I put very little credence in the the PM branch be it RPM or Adjusted. With that said and correct me if I’m wrong but you linked to that page so I’m assuming the stats are “supposed” to support your argument, right?
According to these numbers, Brewer was indeed a “net neutral” for the T-Wolves: The team was 1.8 pts per 100 poss. better with him offensively and 2 pts WORSE defensively.
I thought WoW slights him because he’s a “defensive specialist” and WoW underrates defense.
Wasn’t that your argument?
Where exactly do you see evidence of this defensive prowess in the numbers you linked to?
in the negative def rating? in the godawful counterpart stats?
Enlighten me please.
Tom Mandel
June 6, 2010
Please, whatever you do, *do not* share this article with David Kahn.
I am hoping that the Wizards can somehow trade for Love (I bet Kahn would *love* Andray Blatche!!).
brgulker
June 6, 2010
I am curious, what happened with Sessions this season? I have heard he struggled in the triangle, the primary wise for his decline. Can anyone address that?
robbieomalley
June 6, 2010
Tom,
As a huge Wizards fan, I do too. If I was Grunfeld I’d offer the number one for Love and the #4 or Love and Ricky Rubio (with other filler like Blatche and Young). I bet Kahn would do one of those. But sadly, the one time Grunfeld should trade the draft pick he’s not going to.
And I think the forum here has missed the point of this post. The point is you have to put Love in the game. This is a basketball metaphor for the need to incorporate love in this discussion.
Marparker
June 6, 2010
Fellow wizard fan,
Its only wishful thinking that the Wizards would value Love. I bartend in a place that Flip Saunders frequents(he doesn’t drink) and though I try to leave the poor guy alone I spend the whole 30-45min beaming anti John Wall thoughts into his head.
Maybe Grunfeld thinks the Wiz have a glut of too skinny/too short guards already and will draft Evan Turner.
With all this Minnesota is awful talk, maybe they will be the lucky bastards who get the rights to Evan Turner. I could see Philly taking Cousins or Favors and the Nets taking the guy who is left over. If I’m a Minnesota fan that is what I’m hoping for.
Patrick Minton
June 6, 2010
SadWolvesFan,
Ah, the beloved “all we care about is rebounding” argument. Sure, it’s part of it. An average SF grabs 7.5 rb/48. Corey Brewer only grabbed 5.5. What isn’t apparent to me is why you think it’s not relevant to mention that a player is a terrible rebounder (relative to other players at his position).
But “we” also think turnovers are really important (as in, not committing many of them). And, by the way, I doubt you’ll find many coaches who disagree with that. Well, Corey turns over the ball a lot. About 50% more often than the average wing.
“We” also care a lot about personal fouls (again, coaches probably think fouls matter too). And, it turns out, Corey fouls a lot. About 20% more often than the average wing.
“We” also care about free throw shooting (do coaches think free throw shooting is important?). Corey’s about average at getting to the line, but pretty bad at hitting free throws (64% vs. the average 78%).
brgulker,
In looking at Session’s numbers, his assist rate was low this year (for a point guard), his turnover rate was high (for a point guard) and his personal foul rate was high (for a point guard). He also shot poorly from the free throw line and from 3point range (although he didn’t take many 3point shots). He continued to be an excellent rebounder.
To what extent these things are the result of playing in the triangle instead of a more traditional NBA offense is something for the coaches to evaluate (see? That’s me admitting that numbers don’t tell the whole story).
Edmond
June 6, 2010
Chatter on the internet about the Warriors trading Anthony Randolph and the #6 pick for Love and the #4. As a Warriors fan, it sounds just too good to be true. We could potentially land Love AND Cousins. If it happens, I will personally give Don Nelson a foot massage after every game.
Michael
June 6, 2010
Surely to measure if these stat guys are having an impact is to look at the correlation of wins to dollars in any new contracts they award?
Also, even if stat guys had no effect on the payroll to win ratios of their team it wouldn’t count against the value of stats, only the stats that they themselves were using.
I disagree that Hollinger is only using PER in his analyses as well. He insider stuff is actually quite diverse.
Italian Stallion
June 6, 2010
>I am curious, what happened with Sessions this season? I have heard he struggled in the triangle, the primary wise for his decline. Can anyone address that?<
From what I gather, the Triangle is not particularly friendly to PGs (from a stats point of view) and is very difficult to learn – potentially impacting both Sessions and Flynn.
I read some quotes from Flynn suggesting he thought he would be way more effective in a pick and roll offense.
Scondren
June 6, 2010
I know we’re not supposed to believe in “being on fire” but damn if Ray Allen isn’t making me forget that.
Alvy
June 6, 2010
I know we’re not supposed to believe in “Glen Davis is a good player” but damn if Glen Davis isn’t making me forget that.
kevin
June 6, 2010
What aren’t we supposed to believe that being on fire isn’t something real? Getting “in the zone” is an athletic reality and right now,
Allen and Rondo are in it.
kevin
June 6, 2010
I’m a Celtics fan an so what I’m about to say might be laced with a tinge of bias but Rajon Rondo has a freakish amount of athletic ability.
robbieomalley
June 6, 2010
“The likes of LeBron James, Dwyane Wade, Chris Bosh and Amare Stoudemire will soak up most of the big money. On the second tier, Gay will be highly coveted.
One team that will make a strong run at Gay is the Minnesota Timberwolves. Team owner Glen Taylor and president of basketball operations David Kahn have already targeted Gay as an attainable player to uplift the struggling franchise.”
http://www.commercialappeal.com/news/2010/jun/06/gays-ready-to-answer-the-bell/
kevin
June 6, 2010
Rondo just abused Kobe again on that strip.
ilikeflowers
June 6, 2010
Once again the Laker’s frontcourt was dominant, but this time the C’s backcourt advantage overcame it. It’s going to be interesting to see who can negate the other’s advantage most.
kevin
June 6, 2010
I agree, ilk.
Now though, the next 3 games will be played in Boston, where the Celtics players will have the benefit of familiar rims and a supportive fanbase.
I expect Pierce and Garnett to refind their stroke. They will withe wind the next 3 games or win 2 out of 3 and close the Lakers out in game 6.
The only reason is was this close tonight was the horrendous play of Sheldon Williams. My god, the minutes he turned in in the latter minutes of the second quarter define the concept of “playing scared”.
kevin
June 6, 2010
I’m looking at the boxscore tonight and if you wanted any further hardware to nail the coffin shut on +/-, Derek Fisher led the Lakers with a +4. This is the guy who got himself, his wife, his children, his relatives, his furniture carbonized by Ray Allen and his lethal jumpshot tonight.
ilikeflowers
June 6, 2010
Kevin,
Yeah, I was just thinking that LA really can’t expect much more than they’re getting out of Gasol and Bynum – and with the strategy of giving Odom reduced minutes, I’m not sure where they’re gonna get the likely additional needed production from. LA can’t count on KG to continue having an awful series. We’ll see, maybe Phil has some more strategic tricks up his sleeve to try and negate some of the C’s backcourt and SF advantages.
kevin
June 6, 2010
Yeah, but what can Phil do? Rondo is an impossible matchup for him. He tried putting Kobe on Rondo but all that accomplished was exposing Kobe’s defensive limitations.
As I said before, the Lakers match up poorly against the Celtics. No matter who they put on Rondo, Allen and Pierce, they have a mismatch problem. If Garnett were actually shooting well, this series would already be over. Overlooked tonight was Pierce. He had lousy offensive night but he completely dominated Artest, made him irrelevant. His defense allowed Rondo and Allen to run wild.
I was also impressed with Garnett’s interior passing. If the Lakers hope to win, they have to neutralize Garnett both offensively and defensively. They have generally been successful at that but are still peering out from behind the 8-ball at this point.
ilikeflowers
June 6, 2010
Kevin,
Well, they won the first game because Rondo was just ok and Allen was awful, of course KG was subpar on to of that as well. They’ll need to find some way to replicate that, and hope that Rondo and Allen didn’t just have off nights. I don’t expect LA to be able to, I already picked the C’s in 5 or 6, but you never know.
Arturo
June 6, 2010
Kevin,
You nailed it. Here’s a quick and dirty boxscore analysis for the game:
Player +/- Est WP48 MP Team
Nate Robinson -3 0.944 6 Celts
Sasha Vujacic -7 0.625 7 Lakers
Rajon Rondo 12 0.406 42 Celts
Pau Gasol -12 0.388 42 Lakers
Ray Allen 12 0.267 44 Celts
Rasheed Wallace 15 0.220 18 Celts
Tony Allen 3 0.188 12 Celts
Andrew Bynum 1 0.051 39 Lakers
Kobe Bryant -9 0.033 34 Lakers
Derek Fisher 4 0.029 35 Lakers
Shannon Brown -3 0.010 15 Lakers
Kevin Garnett 1 -0.056 24 Celts
Kendrick Perkins 3 -0.086 32 Celts
Paul Pierce 3 -0.187 40 Celts
Lamar Odom -10 -0.197 15 Lakers
Jordan Farmar -9 -0.257 13 Lakers
Glen Davis 3 -0.315 18 Celts
Ron Artest 0 -0.438 41 Lakers
Shelden Williams -4 -1.165 4 Celts
The celtic guards killed the Lakers. Gasol showed up but all the other Lakers got shot down. I do love that the ESPN guys still claim Kobe had a good game. The really scary thing is that the Celts may actually be better with Rasheed on the floor.
Arturo
June 6, 2010
Yep, I have Rasheed with a .197 WP48 in 36 minutes in the series. Unbelievable.
ilikeflowers
June 6, 2010
Arturo, I believe that your calculations for Kobe are off. Did you incorporate the clutchiness of the two 3 pointers that Kobe actually made? Checking the conversion chart…that should be another 0.200 added to his wp48.
kevin
June 6, 2010
Nice tally there, Arturo. Kudos and thanks.
kevin
June 6, 2010
“I don’t expect LA to be able to, I already picked the C’s in 5 or 6, but you never know.
I picked the C’s in 6 so we’re totally on the same page.
You do never know. But, barring catastrophic injury, I can’t see how the Lakers win this one. They just don’t have enough versatility and defensive prowess to close this one out.
Arturo
June 6, 2010
Ilk,
That the boxscore’s fault. Not mine :-)
Kevin,
My biggest worry is that the best 5 for the celts right now is Rondo,Allen,Pierce,Davis and Sheed and unless Garnett finds his legs (or a handy dandy cortisone shot) they need to go to it more often. I really can’t believe I just typed that though.
Anon
June 6, 2010
The lakers will win the series. I have an advanced statistical model where I decide who I want to win a playoff series, and then the team I’m not going for wins 100% of the time. I want the celtics to win, so therefore the lakers will win the series. You can’t question the stats.
Marparker
June 6, 2010
Its amazing Kobe goes 5to’s,5pf,21 pts on 20 shots and he “had a good game”
Meanwhile that bum Gasol better get his stuff together and stop playing so soft.
todd2
June 7, 2010
Artest; 1/10. Where have you gone, Trevor Ariza? Better #’s, same $, younger/better defender.
kevin
June 7, 2010
Yeah, I like the way how Bynum and Gasol carried the Lakers all game and then, down the stretch, Kobe just forgot about them both and threw up some hellacious bricks.
This could get really interesting. In order to have a chance, Phil might have to tell Kobe “Look, if we are going to have a chance, then Pau, and not you, is going to have the be the go-to guy. You take too many ill-advised shots and don’t pass out of double teams very well.”
How much you wanna bet, if that happens, Kobe’s gonna pout?
For those of you who weren’t around then, Kobe is almost exactly the same player, from a size, skills and demeanor standpoint, as Rick Barry. Rick was a little bigger and played a different position but they are almost exactly the same player.
ilikeflowers
June 7, 2010
I agree Kevin. Center is really the only place that LA can press their productivity advantage. I watched the entire fourth quarter but I didn’t even notice that Gasol didn’t take a shot (there’s a knock against my eyeball model – it was mainly focused on the ball, i.e. Rondo and Kobe). LA fans can take heart in that maybe they would have won if they’d just kept pressing the issue at center. Playing a two man game with Gasol and Kobe (skewed heavily towards Gasol) with Bynum/Odom crashing the boards sounds like one of their few effective strategic possibilities given the matchups.
— Celtics Advantages —
PG
Backup Backcourt
SF
— Laker’s Advantages —
C
Backup Frontcourt
Homecourt
With Boston, I’m not sure that homecourt even matters. I think that they’re one of the few teams that could actually win games 6 and 7 on the road after having lost 2 of 3 at home. Ultimately, I think that LA needs one of their scrubs (i.e. everyone other than Gasol, Odom, Bynum, and Kobe) to seriously overperform and they have to continue to play Gasol almost the entire game.
ilikeflowers
June 7, 2010
It was also an interesting anecdote watching the difference in the fourth between a 0.290 player dominating the ball versus a 0.164 player.
kevin
June 7, 2010
“With Boston, I’m not sure that homecourt even matters. I think that they’re one of the few teams that could actually win games 6 and 7 on the road after having lost 2 of 3 at home. ”
Well, homecourt advantage for the Lakers is gone now. The Celtics have the homecourt advantage now.
ilikeflowers
June 7, 2010
True, but those advantages are basically a repost of a summary that I did looking at the series prior to any games being played. LA could crumble as they’ve done in the past against Detroit and Boston and lose all three games, but I think that it’s reasonably likely that they’ll be able to win at least once. For what it’s worth, I’d give LA a 40% chance of winning at least one game in Boston.
kevin
June 7, 2010
Agreed, but I give about a 10% chance of winning 2 and I think they have to win 2 if they hope to repeat. the Celtics are not going to lose back-to-back title games.
ilikeflowers
June 7, 2010
I keep reading about how Pierce hasn’t played well due to Artest, but Pierce was awesomely efficient in game 1. He scored 24 points on 13 shots! 9 boards as well. His other stats weren’t great but he certainly played well on offense.
kevin
June 7, 2010
Agreed again. And his defense on Artest in game 2 was phenomenal. Artest was irrelevant. And it would have been even worse if they hadn’t called that BS foul on the come-from-behind block.
I expect Pierce to get unlocked in Boston. he always shoots better at home.
Arturo
June 7, 2010
Ilk,
I’ve got Pierce at .199 WP48 for Game one and -0.187 WP48 (ouch!) for game 2. So Artest did do a good job on him in game 2.
kevin
June 7, 2010
“I’ve got Pierce at .199 WP48 for Game one and -0.187 WP48 (ouch!) for game 2. So Artest did do a good job on him in game 2.”
This is where WP48 needs some work though. Artest himself went 1-10 from the field and fouled out. Pierce had a great deal to do with that. Pierce should have been given credit for shutting Artest off like a faucet.
Useitall
June 7, 2010
It would be foolish to use raw +/- and nothing else. But use it and / or Adjusted +/- alongside boxscore stats and boxscore based metrics, it can help understand the big picture.
You use counterpart stats to get at Artest vs Pierce or the story with Fisher.
Arturo
June 7, 2010
Kevin,
Actually, Artest was .258 WP48 in the first game and -0.438 in Game 2 (only Shelden williams was worse). Pierce did do a great job on him.
ilikeflowers
June 7, 2010
Useitall,
Or we could just use counterpart wp48 and winscores – there’s no reason to add in even more noise from trying to remove backup effects.
dberri
June 7, 2010
Useitall,
Plus-minus data — whether it is adjusted or not — are not consistent from season-to-season. We suspect this is because you cannot really filter out the impact of teammates. Given this problem, how are you supposed to use this information? Or to put it this way…. if a player does well by this measure in a game or a season, how do you know that isn’t just the impact of teammates you are picking up? And if you can’t tell (and I don’t think you can), how does this information help someone make a decision?
Just looking at numbers doesn’t really help if you can’t tell what the numbers mean. And that is a real problem with plus-minus data.
useitall
June 7, 2010
Dave,
if you use raw counterpart match-up boxscore data and Adjusted +/- as tools in your overall data analysis along with other things and see where they agree or disagree and think about all it further aIong with everything you see and know about the game and apply your judgment I think you can do better more often than compared to not using either of these tools.
I am still thinking about whether you can systematically blend these metrics in an objective way. Right now my synthesis of them requires judgment and caution.
If you have decided that it can’t possibly be useful to think about this data at all, in any case, or you personally can’t come to any better overall conclusions when consulting these tools vs not or can’t get comfortable with the uncertainty vs what they might suggest, that is your decision.
I find it useful to see and consider many perspectives.
ilikeflowers,
I’ll look at and consider counterpart wp48 when it is made available.
useitall
June 7, 2010
If Adjusted +/- is not consistent over 2 seasons then I’ll look at Adjusted products that are available that use 3, 4, 5 and 6 seasons and I look at the range and direction of the values from each season and I’ll look at everything else.
If the various values and metrics are mostly in agreement then I will cautiously conclude as much as I feel comfortable concluding or guessing. If there is sharp disagreement I am going to note that and think further and realize it is likely not a simple call. I am not going to say one metric is better than the others and the only one I am going to use.
ilikeflowers
June 7, 2010
Useitall,
then how does it get at Artest vs Pierce or the story with Fisher?
ilikeflowers
June 7, 2010
Useitall,
I find it useful to see and consider many perspectives.
What are the insights that APM has given you so far in these playoffs/finals that wp48 or winscore has missed?
useitall
June 7, 2010
When the Adjusted +/- method and estimates first went public that author almost immediately went on to produce an “Overall +/-” that essentially was about 4 parts “Statistical +/-” and 1 part Adjusted +/-. None of the later Adjusted +/- authors followed that direction and I think that is a missed opportunity.
I don’t know how much better this blend would be at retrodiction or prediction today because I haven’t seen a solid recent test.
I also don’t know how much better other conceivable blends might do. Jon Nichols’ Composite Offensive and Defensive Scores notably included counterpart data (to help get at 1-1 defense) but I would probably use different weights or search for the weights that help best explain the complexity of the data.
I tend to think blends might do better but would appreciate a rigorous test by someone well qualified to conduct and assess the test.
I don’t know if or how much the Cavs use Overall +/- or a descendant of it.
My layman interest in meta=metrics is based on examples of cases where it has been used productively in other arenas.
useitall
June 7, 2010
Raw counterpart boxscore data gets at Artest vs Pierce or the story with Fisher more directly and to a greater degree than anything else.
Adjusted +/- is not the primary tool to get at that dimension, though if you have Adjusted splits it can extend beyond what counterpart data shows to get at impact on teammate offense and impact on overall team defense.
I personally would not use Adjusted +/- from a single series to say much or maybe anything. I would use Adjusted +/- from 15+ games with the caution the error term suggests and then some.
But looking at season long Adjusted +/- of players and lineups I think there are things you can learn and say.
I’ll pause to review that data and will decide how much I want to go into now or later after that.
dberri
June 7, 2010
useitall,
Even with an entire season worth of data we see..
1. most results are not statistically significant
2. the season-to-season explanatory power is around 10% (that would be very low)
3. there is no statistical relationship between a player’s season-long APM when he switches teams.
Given all this, what do you think you are learning by looking at this?
Again, the results seem to depend a great deal on a player’s teammates. Since you can’t separate the player from his teammates with this measure, what are you learning?
ilikeflowers
June 7, 2010
Of course there isn’t much insight to be gained from playoff or finals numbers since the sample size is so small. So really the numbers from the season are more important. As the Celtics post from the Cav’s-Celtics post-mortem details, looking at the wp48’s of the Celtic’s starters, tells you that they are a formidable group when given the lions’ share of the minutes and were a championship caliber collection. It also tells one that the Cav’s left a lot of productivity on the bench in that series. I’ve posted what I think wp48 says about this series above and elsewhere. I typically ignore APM, since every time I see the standard errors they’re so consistently large (frequently bigger than the APM itself) as to be useless. But perhaps APM tells the same story, I’ll leave that for other interested parties to address.
useitall
June 7, 2010
The 2 finalists have the starting lineups with the very best Adjusted +/- of any used 5+ minutes per game for the season.
The Suns were 5th best and real close to #3. The Cavs would have been #3 for the stretch they had Jamison.
Atlanta had the 4th best that qualified. Dallas, Portland, Utah, Milwaukee and San Antonio also had good biggest minute lineups but did not use the best as much in the regular season as some others and you could argue that they did not use them enough in the playoffs in some cases.
Orlando had a lineup that worked well regular season and playoffs. They used it a little more in the playoffs. Perhaps they should have used it even more, perhaps a lot more. Adjusted lineup +/- was the best approach to find this.
Using lineup Adjusted +/- it was easy to spot a trouble spot for OKC, Denver, Chicago, Miami and Charlotte. They did not have a strong top lineup and I think it was reasonable to assume that would be a problem in the playoffs matching up against other top lineups.
Point differential or team level Wins Produced would have predicted playoff series right but looking at top lineups gave a stronger signal to favor the Jazz over the Nuggets.
Looking at a lineup Adjusted +/- for the regular season split based only on play against playoff level teams or just for within your conference might be even more useful than the complete season rollup which includes lottery teams. performance against the weak may not be not important for predicting performance in the playoffs against better teams.
useitall
June 7, 2010
If as you say player level Adjusted +/- “results seem to depend a great deal on a player’s teammates”, and I agree- you will never be able to completely separate individual player impact perfectly- then you use that player level information with caution and in conjunction with everything else to look agreement or disagreement and you look at player performance in specific lineups- something which Wins Produced proponents seem to argue or assume does not really matter much or at all or in a measurable way.
But each will use what they want.
ilikeflowers
June 7, 2010
Adjusted lineup +/- was the best approach to find this.
Point differential or team level Wins Produced would have predicted playoff series right but looking at top lineups gave a stronger signal to favor the Jazz over the Nuggets.
These are the only two concrete statements in your post. Presumably, you have the numbers to back both of them up?
useitall
June 7, 2010
I meant to say Point differential or team level Wins Produced would have predicted “most of the” playoff series right.
I guess the only miss by point differential was the Spurs over Dallas.
The Spurs had top regular season lineups that were strong like the Mavs’. It doesn’t look like the best lineups faced each other much or any in the regular season given the Dallas trade and Parker’s absence.
The Spur’ failure to get beyond the first round may be related to their failure to use their top lineups as much as other teams typically do.
Parker- Ginobili- Jefferson- McDyess – Duncan was a smashing success on small sample (10 game) playoff lineup Adjusted +/- but they only used it less than 7 minutes a game. It was strong in small sample for the regular season too and despite the small sample size I would have thought it would have been a better idea to play it more than they did.
If you just think at player level and add up boxscore based stats you wouldn’t even grapple with the issue.
useitall
June 7, 2010
To find lineups that work well… you look at lineups. You could look at raw +/- or Adjusted +/- or both. I usually look at both. Since in this case the message from both was consistent for both time periods I went with mentioning Adjusted +/- because of its effort to account for differences in quality of opponents faced.
useitall
June 7, 2010
My statement about the top lineup matchup between Utah-Denver might have been right… if Kirilenko was available from the being and healthy. He wasn’t. Sorry for that oversight.
The Nuggets lost the series by a net 7 points. Their most used lineup lost by 11 in just over 20% of the total time, a pretty bad rate of loss.
I should have just said the weakness of the Denver top lineup was a concern. If you looked at raw team +/- they were +6 per 48 and you might have thought they were fine. But if you look at Adjusted +/- they were barely over neutral.
And against a good team that might not be enough.
It wasn’t. Utah found acceptable lineup replacements to beat Denver’s top lineup and win the series.
useitall
June 7, 2010
At player level, Adjusted +/- that includes the playoff data suggests players like Gortat, Haywood, Collison, Dudley, Varejao and Korver were probably positive help for their team.
In how many of these cases does playoff WP agree with that? You can check and tell me if you want- I can’t get into that database due to software incompatibility.
Whatever the answer, I would use it all to develop summary judgment and specific action plans.
useitall
June 7, 2010
in above, should be
…if Kirilenko was available from the “beginning”
marparker
June 7, 2010
Where have I seen this writing style before
dberri
June 7, 2010
marparker,
Obviously everyone who likes plus-minus has the same writing style. Multiple posts in a row. Arguments that don’t quite address contrary evidence. Something about plus-minus makes people behave this way (or maybe, just maybe, it is the same person with multiple names).
marparker
June 7, 2010
I saw the post count and thought I missed out on some really discussion. Oh well
useitall
June 7, 2010
If multiples posts seems rude to you, I’d say that is just an opinion. I consider it a form of paragraphing.
ilikeflowers
June 7, 2010
useitall,
Those posts are reminiscent of some earlier rambling nonsensical APM gobbledigook from another poster I wasted time with once. Any reason for the subterfuge?
useitall
June 7, 2010
I’ll use a name with a topical statement if it suits me.
Across a few threads, I’ve addressed a few topics, responded to contrary views, dealt with context and complexity and gave some novel suggestions.
Dave, you repeat your favorite questions and habit of not liking any answers. I provided the answers I am going to. If they don’t sway you, I accept that and am not surprised. Your perspective and mine don’t have to be the same and apparently aren’t.
Your questions and answers don’t sway me from believing that using multiple metrics is better than just one.
I probably won’t post much more. I am not hear to argue endlessly. I decided to recognize the new era on the site with multiple authors and express some viewpoints and debate a few rounds but I’ll probably leave it at this or not much more.
useitall
June 7, 2010
If you are turned off by multiple posts, I think you are either oversensitive or looking to try to pick on something other than the substance.
useitall
June 7, 2010
“So Wins Produced suggests that Kevin Love is not only the best player in Minnesota, but possibly one of the best players in the NBA.”
Traditional one and two year Adjusted +/- have Love as a modest negative on overall impact.
Regularized 2 year had him as slightly positive. Regularized 1 year as of the All-Star break had him as improved to the 50th best player in the league.
I’d consider all these marks and the boxscore based marks in forming my overall pretty positive impression of him so far.
RTG
June 7, 2010
usitall,
On the topic of multiple posts: I don’t think the users in this forum are being oversensitive or trying to pick on something other than substance. Instead, I think its just a matter of their (and my) displeasure at your total disregard for the protocol of communication generally observed in this forum.
Imagine being at what is probably a normal workplace meeting for most of us: most people take a few seconds to collect their thoughts mentally before speaking. Then, if after speaking, some new thoughts occur to that speaker, they normally either wait a few minutes to fully develop them or even if the thoughts are fully developed, they usually wait for a significant number of new thoughts to accumulate before speaking again. Now, imagine if in such a setting, one person violates this protocol and instead speaks every thought he has when he has it regardless of whether he has fully molded the argument in his own mind or not. The issue here is not that speaking “half-baked” thoughts in and of itself is inappropriate (it’s definitely appropriate, for example, during a “barstool argument”); the issue instead, is that that person has violated the appropriate protocol for that forum.
For an online comments board, it appears to me that the “appropriate protocol” would be the rules either set by the site owner (dberri) or, absent enforcement/guidelines, the protocol commonly followed by other speakers (posters). When you violate the appropriate protocol, you annoy everyone because it is clear that either a) disregard/do not understand their method of communication and b) think it is too cumbersome and is not worth your time. As a result, the biggest injustice to the “substance” of your posts is caused by YOU, not US.
If you’re still confused by my mini-rant/speech, imagine this: Dberri goes onto ABPR Metrics (sp?) forum, takes every piece of work there and begins applying the same rigorous review process that he (likely) applies when working with his grad students. ABPR Metrics (sp?) members would likely be upset/annoyed because Dberri would then be violating the protocol of their forum. See what I mean?
useitall
June 7, 2010
A digital forum is a bit different than a verbal live conversation, in my opinion.
In a verbal live conversation I follow the norms for that.
In a digital forum everyone has the ability and right to say whatever they feel and thinks adds value. There is no competition for airspace. Read and react as much and however you choose. Even “long” threads are not that long.
If you feel it is too much or not sufficiently refined that is your opinion. It may be the opinion of several or many but I don’t think it is a big breach.
“If you’re still confused by my mini-rant/speech, imagine this: Dberri goes onto ABPR Metrics (sp?) forum, takes every piece of work there and begins applying the same rigorous review process that he (likely) applies when working with his grad students. ABPR Metrics (sp?) members would likely be upset/annoyed because Dberri would then be violating the protocol of their forum. See what I mean?”
No I don’t see what you mean at all. I think most there would read whatever he were to write there and react however they felt and however much they cared to or felt compelled to.
useitall
June 7, 2010
I made a 2 sentence statement relatively late in the life of the thread. No one can say they didn’t have a shot to say whatever they wanted. Then or now.
I got a question, I responded, once.
I got more questions, so I responded further as I wanted.
I got more feedback, so I gave additional commentary.
I returned the focus to a point made in the original article.
And then shifting from substance back to style views, I responded to you about posting style. I’d rather talk about substance.
makeitup
June 7, 2010
This whole Wins Produced is dull, and it’s reliance on concepts like hard data, correlation with team results, and accuracy of prediction are of little interest to me.
makeitup
June 7, 2010
I much prefer measures that leave room for interpretation. That’s the element I find most interesting.
RTG
June 7, 2010
useitall,
So I guess we’re doing this all the way. It’s on like donkey kong. I’ll try to be civil.
“A digital forum is a bit different than a verbal live conversation, in my opinion.”
>Yes, it is a bit different. I’m fairly certain most would agree that the expectation of following the prevailing protocol is not one of these differences.
“In a verbal live conversation I follow the norms for that.”
>OK
“In a digital forum everyone has the ability and right to say whatever they feel and thinks adds value.
>In any forum, everyone has the ability to say whatever they feel/thinks add value. Even in a meeting with a VP several levels about little old me, I have the ability to say anything. I don’t because I don’t want to break protocol. It’s not about a lack of ability, its about decorum.
‘There is no competition for airspace. Read and react as much and however you choose. Even “long” threads are not that long.”‘
>I’m not complaining about length. I do, personally, take objection to some of the longer comments in this forum, and exercise my objection by not reading them. Note that I don’t criticize such commentators. They usually follow protocol and collect their thoughts before posting. Therefore, they don’t ruin the shorter comments or the overall forum, which I do like to read.
“If you feel it is too much or not sufficiently refined that is your opinion.”
>Yes, it is an opinion insofar as the degree of refinement is a subjective qualification. However, because I’m not trying to prove anything to you w/ facts. I’m only trying to give you my thoughts on why others object to your posting format so that you can perhaps move beyond this barrier and make the substance of your posts known. I’m trying to help you.
“It may be the opinion of several or many but I don’t think it is a big breach.”
>You don’t think it is a big breach. Clearly, others do, so they lash out at you/don’t address your comments. If you’d like to move past this barrier and into discussion, I suggest you reevaluate whether your actions are a “big breach” or not.
“No I don’t see what you mean at all.”
> I can’t make myself any clearer. And so I won’t try. Maybe someone else can take a stab?
“I think most there would read whatever he were to write there and react however they felt and however much they cared to or felt compelled to.”
>I would hope they would react however they felt and however much they cared to or compelled to. If they reacted in a way different from how they felt or if they reacted more/less than the amount they cared to or felt compelled to, I would have to conclude they were insane. I doubt they are insane. However, their reactions would convey their annoyance at the violation of protocol.
makeitup
June 7, 2010
Wins Produced does not lend itself to the kind of uncertainty that promotes lively discussion. If one can simply look at hard numbers and draw clear conclusions, there would be no reason to post dozens of one- or two-sentence comments about lineups that work for reasons that are not only thoroughly obscure, but arguably are completely inexplicable.
makeitup
June 7, 2010
I would much rather speculate on why certain lineups work, then make recommendations for coaches and GM’s. This may have the potential to lead to a lucrative consultancy business.
makeitup
June 7, 2010
Wins Produced does not offer the same consulting potential, and thus holds much less appeal. A metric which results in the advice of “get better players” is unlikely to garner much repeat business.
makeitup
June 7, 2010
Even if it does quite well at identifying which players are more productive than others.
makeitup
June 7, 2010
Simply identifying the better players is not very useful; teams have the players they have, and changing a roster takes some time. Trial and error with lineups whose success cannot be predicted, explained, or replicated allows for a great deal of fees to be paid before the client can see it’s all guesswork.
makeitup
June 7, 2010
Wow, I’m even annoying myself. Of course, that’s only my opinion, and I have every right to ignore it.
RTG
June 7, 2010
makeitup,
thanks for the shoutout
useitall
June 7, 2010
RTG,
I hear your perspective.
I’d rather my posting style at times didn’t bug people, but I’d rather say what I wanted to say when I when the time to say it and for me that takes precedence as long as I am addressing legitimate topics.
If I can do the latter without pushing buttons, I will. If questions and critiques, causes me to post more and frequently than some would prefer that is a side effect and point of criticism that I will bear in those circumstances.
Your statement about “reactions would convey their annoyance at the violation of protocol” doesn’t register well with me because there is no written protocal at APBR forum- or unwritten assumption in my view- that you can’t offer as much analysis or critique as you want. It may or may not go over well for substantive or stylistic reasons but make your points and listen to the hopefully substantive response, if any.
You are assuming that Dr. Berri posting a lot of critique there would annoy and provoke hostile reaction by many. I disagree; I don’t think posting a lot by itself would provoke much criticism on that point specifically.
The hostile reaction if / when it occurred from some would mainly be based on substance or tone or past history.
Speaking strictly for myself I would not mind at all any commentary he’d wish to make there, even a lot of critique of methods he deems flawed or not rigorous enough. To be clear I am not I am not pushing for or critiquing the choice not to, just responding to your statement.
But some may feel differently, there as here.
At APBRmetrics or any serious online discussion spot like this one, the more thoughtful commentary and response the better to me. If you reach your limit, exercise your right to stop or skip over as you wish. I will read generally almost everything anyone else had to say as long as it is about basketball and thoughtful.
useitall
June 7, 2010
Makeitup, if you want to chant your talking points go right ahead. Valid perspectives to you and in some cases to me and overly simplistic characterizations in some cases to me.
RTG
June 7, 2010
useitall–I’m glad you hear my perspective.
On the whole Dberri/APBR metrics example, I just want to clarify that that was only an example. I didn’t want to say anything about the person Dberri or the forum APBR. I was instead just trying to give an example of how breaking protocol in a different forum would not (in my opinion) be received well.
For the sake of complete clarity, let me remove Dberri’s name (who I must apologize to for somehow having dragged into the whole APBR discussion again) just say a different example: Let’s assume I get invited as a panel speaker on ESPN for the finals (I wish). Then, in the middle of a standard “ESPN discussion” about killer instinct, clutchness, and making your teammates better, I simply start reading off WP48’s. I would be breaking ESPN protocol. If I was kicked off of ESPN, I would not complain about their oversensitiveness or the fact that they are trying to pick on something other than substance. I would have to instead understand that their disregard for hearing me out would be based on the fact that I destroyed a show meant for the casual viewer by breaking protocol.
Alright, I’m done. I can’t believe my first foray into the forum was about this.
dberri
June 7, 2010
It is interesting that more than one APBRmetric person has posted under multiple names. You can tell that it is more than one person because the writing style (which as anyone who has graded papers can see) is unique to the person. So in the past you would have one writing style under multiple names. Now we have another writing style under multiple names. Is there something about plus-minus that leads to this behavior?
And I loved the “makeitup” comments from jbrett.
dberri
June 7, 2010
RTG,
Along the same argument… people who study statistics in sports often note that certain numbers don’t have much meaning because they don’t have any predictive power. Imagine if you were invited to provide color commentary on a game and you kept saying “yes, that number doesn’t mean anything. Really we haven’t learned anything about the quality of that player from that event.” I am sure this line of reasoning would get old very quickly.
useitall
June 7, 2010
Wins Produced is based on a regression which has error in its estimates of the value of discrete boxscore stats relative to others and for the action of one player in one moment vs all the rest just as other Statistical +/- has error and Adjusted +/- has error in their estimates of value assignable to a player.
The team adjustment serves a purpose and need, but the method and the accuracy of what it tries to do can be debated.
Under the WP team adjustment all players have their scores adjusted based on the total team defense across all of the team’s time and team shot defense is presumably a big part of what it is trying to capture. Better to try to capture than not to me, so I credit the inclusion.
But the shot defense, individual and team or help shot defense, and all other non-boxscore defense of a specific player for the unique set the time when they are on the court is not really equal in value for all players and uniformly at the average rate found for the team’s total time. The adjustment is an practical approximation with plenty of real error in judging the real defensive performance of individual players.
useitall
June 7, 2010
“in the middle of a standard “ESPN discussion” about killer instinct, clutchness, and making your teammates better, I simply start reading off WP48′s. I would be breaking ESPN protocol.”
If you think ESPN’s protocol is to not allow others to say what they genuinely believe and want to say and you agree to not say what you want to say because of this belief, then yes we have different perspectives.
useitall
June 7, 2010
If what you want to say can be said in a few minutes or read in a few minutes it doesn’t cross my general threshold for being too much. But if your threshold is lower or more conditional to style of delivery or the perspective it comes from and whether you start out in agreement or not, do what feels right to you.
useitall
June 7, 2010
Direct boxscore stats are more stable.
Instability in player Adjusted +/- which is not tied directly and exclusively to the boxscore stats of a specific player and at the league level is not sufficient proof that there isn’t some useful or potentially useful signal in at least some cases at an individual player level if the signal rises to a certain level and lasts long enough. It is far some certain or exact but no metric provides a flawless answer.
Instead of noting instability and immediately and rather completely rejecting it as mainly or entirely the result of a flawed metric, you could try to research further and look at split data and try to understand any sources of real variation from the actual use of the player.
I wouldn’t say you can ever precisely know how much variation is error and how much is real but you can either try to understand and make judgments based on that understanding or not try and bet that your analysis without this effort will be better for being less diversely or singularly based. Different people will select different alternatives.
I’d use and consider the results of a lot of tools, separately and blended.
Weighting what is directly tangible heavier than what is indirectly estimated is a reasonable preference and I would still use boxscore data pretty heavily.
useitall
June 7, 2010
Use of screen names and different screen names can have various reasons including thinking that it is the content of the post that matters more than the screen name. I have stayed with the same screen name or an abbreviation in each specific thread. I changed them in these several recent threads because I found that a convenient way to highlight the somewhat different main points being presented. It was probably a mistake to give folks something extra to get annoyed at or react to. I am somewhat sorry for that- it was a casual choice but done in a way where I assumed people would see the connection (all saying what else matters or use it all) and I wanted them to see it rather than not see it . If I was going to stay longtime I’d probably select a permanent handle, but that is probably unlikely and I didn’t really have a permanent one in mind.
dberri
June 7, 2010
One of my major criticisms of APM is that so many results are not statistically significant. Defenders of this method echo the arguments seen above. Essentially the argument is that people can make whatever judgements they want. This suggests the rules of statistical inference are whatever you like them to be.
Not sure how it work in other fields, but my experience in economics is that you don’t get to do this. Insignificance is insignificance. We don’t look at results that are insignificant and say that despite the really large standard errors we have really learned something (other than the observation that the results are not statististically significant).
Unfortunately, insignificant results are hard to sell. So you are probably better off just ignoring this issue (at least, you are better off if you wish to sell something).
useitall
June 8, 2010
If Minnesota is to improve they can do it almost anywhere, captured by one metric or another, at player level or lineup level. Outside shooting and better defense are probably the greatest needs but they also need to cut own turnovers and force more and get to the line. And they probably need better lineup fit to achieve more on these things more consistently.
Flynn – Brewer- Gomes – Jefferson- Milicic has positive in short minutes on lineup Adjusted +/-. Three lineups with Sessions and Love were the only other lineups that were positive on Adjusted +/- among those use over 50 minutes.
After the draft and off-season, I’d pick at least 5 lineups to try to test to a somewhat higher level of estimate accuracy (hopefully to or on pace to 300=400+ minutes for the season each) and then assess the player level data from various metrics and the lineup level data and make some further choices.
useitall
June 8, 2010
Insignificance year to year for predicting a metric’s value at league level to a high level of confidence is clearly insignificance for predicting that metric the next year at that level of roll-up to a high level of confidence .
There is indeed a choice whether you stop there entirely or proceed cautiously. We disagree on what we will or want to do. I respect your training and reasons.
I still regard it often to be pretty much “ought to do the best you can’ managerial analysis and then judgment. I don’t believe the standards for this type exercise are universally agreed upon and fixed in exactly the same spot.
I think there are individual cases with pretty consistent Adjusted +/- values which are probably a decent basis for estimating a player is more likely somewhere on the positive or negative range or more likely near neutral, or in fewer cases the information is likely enough to say a player is probably nearer one extreme or other or at least is likely not near the other. Unless it is strongly contested by the majority of other information.
But in each individual case, if you were responsible for presenting a professional judgment, of course use the rules of statistical inference for that particular case or question and be clear about your level of Statistical Significance and any leap of judgment beyond it that you might select to make (instead of stopping and not making an estimate of that facet), if you feel you have a stronger basis to make the judgment (all things considered) than to think the opposite or to be agnostic.
I won’t consciously ignore the flaws, error or limitations of any method. While I might critique other methods more here, I have critiqued Adjusted +/- plenty elsewhere and some here because it needs it and I think it can be improved.
I don’t promote just one method over all others. I make it pretty obvious in my writing and private conversations that I see complexity and the need for consulting many tools and expect that there will be some roughness and some unknowns and some mistaken impressions despite best efforts. If I did stick to one method and believed it answered everything or nearly everything it probably would be easier. I would only promote research of my own or others to the best of my current or future ability to produce or use. I would have both cautious assessments of everything that I can find and think might be reasonably solid or useful within some bounds of confidence clearly stated and I would have judgments and even opinions that I would want to share but would largely share them based on the context of the conversation and if it was requested or might be allowed or considered. I might need to separate these two parts out more distinctly sometimes but there is a place for both and certainly in an online public forum a fan can do both.
kevin
June 8, 2010
“Trial and error with lineups whose success cannot be predicted, explained, or replicated allows for a great deal of fees to be paid before the client can see it’s all guesswork.”
It’s obviously not guesswork. Red Auerbach build three successive championship teams (4 really, when you consider that the 1969 teams had only 1 player left over from the 1957 roster). That was hardly guesswork. he knew exactly what kind of players he was looking for.
Marparker
June 8, 2010
In short,
If you are a wp48er you believe in the accuracy of a model that has been proven to be accurate.
If you are a APMer you believe that no model is accurate and prefer to derive data from data that has already been derived.
Dre
June 8, 2010
Kevin,
There is a team dynamic to basketball that would be great to understand. WP48 does include team adjustments that are added based on per minute for each player. It would be great to be able to get a better feel for how lineups and individual defense affected this. That said, I am not convinced of using small sample sizes of highly dynamic data.
I do appreciate you actually citing an example rather than just calling people wrong. However, your example is 40 years old, pre advanced stats (I am talking offensive rebound and steals not PER and WP48) based on a League of 8 teams using the team with one of the League’s best players as an example.
jbrett
June 8, 2010
Kevin,
To build on Dre’s reply:
Red was certainly a smart guy, and an excellent coach; however, his own accounts of how he got players tell a story of guesswork. His legendary backcourt, Cousy and Sharman, fell into his lap when other teams folded; he’d already passed on Cousy in the draft, to the consternation of the Boston press, and his response was that he got paid to win games, not play local talent. He drafted Russell and Sam Jones, among others, without ever seeing them play; it is also safe to assume he lacked access to useful boxscore information. He relied on what was largely a social network, which also had no advanced metrics at hand.
I’ll give him one other thing; when it came to making trades, he had the soul of a used-car salesman. (Mel Counts for Bailey Howell? Rick Robey for Dennis Johnson??) He won a couple in the 70’s with Cowens, made a shrewd draft in ’78 and a steal of a trade in ’80 to win 3 more.
But his dynasty was Russell and the right supporting cast; they won nothing before he arrived. So he was shrewd, but he was more than a little bit lucky–not ‘sour grapes’ lucky, but second-chance-with-guys-he-didn’t-want lucky. To his credit, once he got the right players, he seemed to do well figuring out the optimum lineups. Then again, I doubt he was the only one who could have made those calls, with that roster. And there may well have been someone willing to advise him on maximizing those lineups, for a modest fee.
kevin
June 11, 2010
“He drafted Russell and Sam Jones, among others, without ever seeing them play”
That’s not true. He saw Russell play.
And your forgetting about a million brilliant moves large and small he executed to keep pumping the talent pipeline.
It’s also not true he won nothing before Russell arrived. He went 49-11 with a Washington Caps team when he was 30 years old.
“To his credit, once he got the right players, he seemed to do well figuring out the optimum lineups. Then again, I doubt he was the only one who could have made those calls, with that roster. ”
That’s the whole point. He had control of the roster and ALWAYS got the right players and figured out the optimum lineup. havlicek was the last player chosen in the 1962 draft. He didn’t just draft Russell. He shrewdly manouevered to get the chance to draft him. He got KC Jones was a second round draft pick. He got Havlicek as the last player in the first round. He signed Don Nelson as a free agent after the Lakers let him go. He signed Larry Siegfried as a free agent after Cincinnati gave up on him.
He got Paul Silas in shrewd screwing of Jerry Colangelo when Colangelo thought he could sign Celtics draftee Charlie Scott without consequences.
I could go on and on and on. Every player, every brick in the wall, was carefully chosen by steadfast manouevering by Auerbach, year after year after year. That’s why he ranks first in the annals of the NBA in terms of alpha-dog status. He’s ahead of Jordan. Clearly ahead.