The Wins Produced numbers from Andres Alvarez were posted on Saturday night. And we already have our first post using these numbers. Arturo Galletti is an electrical engineer and statistician by education (masters in electrical and computer engineering), by vocation (he was worked across the past ten years for the US government — as well as Baxter and Johnson & Johnson — to convert data into information and improvement actions), and by passion (Minitab is great for stress relief and winning fantasy leagues). He was born and lives in Puerto Rico. But his undergrad years were spent in Boston and he bleeds Celtic green. His first post reflects that passion. Arturo will argue – via Wins Produced – that the Cavaliers’ loss in the second round was not that surprising.
Much will be made over the following days and weeks by the talking heads on the blogs, radio, and TV about the “historic” collapse of the Cleveland Cavaliers and the supposed no-show by LeBron and his teammates. Given that this version of the Cavaliers was finally supposed to have all the pieces around LeBron necessary for a championship run, it’s natural to assume that the Cavaliers losing to the Celtics is an historic collapse on the order of the great playoff collapses (Mavs-Warriors or Sonics-Nuggets) of recent memory. Certainly if we look only won-loss record for the season (61-21 for the Cavs and 50- 32 for the Celts) the pundits would seem to be right. But using Wins Produced we can show that the vilification of the King and his court may be without merit. The Cavs didn’t make history, they were victimized by it.
First off a bit of historical perspective. The Cavaliers are only the second team to win sixty games two years straight and fail to make the NBA finals. The other team was Lew Alcindor’s 1972 & 73 Bucks (63 wins and 60 wins) a team which had the misfortune of running into the 1972 & 73 Lakers (69 wins and 60 wins). In Kareem’s defense that team won the 1971 title and lost in seven in the finals in 1974. The King cannot point to a banner in his kingdom but his current subjects will argue that his knights were weak and his general might have played the wrong guys
Table 1- Cavaliers win projection based on the Minutes allocation in Celts-Cavs series and WP48 after 82 Games
Table 1 clearly shows there is some merit to this theory. If we look at the Cavs as they are currently composed, use the team’s WP48 for the regular season, and the minute allocation during the series with the Celts; we see a team that projects out to 70.7 wins. While this is by all accounts an exceptional number for any team (and it’s indicative of a team that should make the finals at the very least), it is also extremely worrying in that LeBron accounts for 29 of the 70.7 wins or 41% of the total. Based on this allotment of minutes, if you replaced LeBron with a replacement level player this team would only be good for 48.9 wins. And that one injury – in this case LeBron’s elbow — might have been enough to sink the King’s ship.
To put this in perspective, let’s consider the percentage of wins produced by the most productive player on each second round team (this is based on the Andres Alvarez numbers).
- LeBron James: 42.8%
- Dwight Howard: 37.0%
- Rajon Rondo: 35.4%
- Josh Smith: 33.5%
- Steve Nash: 30.1%
- Carlos Boozer: 29.7%
- Pau Gasol: 29.2%
- Tim Duncan: 28.9%
As one can see, LeBron had to do more for his team than any other leading star on the second round playoff teams. So Cleveland is somewhat a one man band.
LeBron also wasn’t helped by head coach Mike Brown and his player selection. Brown was apparently seduced by the ghost of Shaquille O’Neal (WP48 0.86, a below average mark) and all the points scored by Mo Williams (WP48 0.116, just slightly better than average). Meanwhile, Varejao (WP48 0.181), Moon (WP48 0.191), and West (WP48 0.121) were often kept on the bench, much to the detriment to the team.
So are those the final answers? No, we are forgetting the most important fact: The Boston Celtics may be much better than their final record indicates. Let’s just assume that the second half of the season was an injury-driven aberration. If that is true, then maybe the first half numbers of this team are the best measure of the team’s quality.
Table 2- Celtics win projection based on the Minutes allocation in Celts-Cavs series and WP48 after 41 Games
Following this logic, let’s turn to Table 2. This table reports the WP48 of each player on the Celtics after 41 games. It then utilized the minutes allocation in the series against the Cavaliers. With these two numbers in hand we see team capable of winning 68.2 games. The Celtics at 100% are a fantastic team that has multiple options to carry the load (i.e are not 43% driven by the King and his elbow).
And in contrast to Mike Brown, Doc Rivers (other than perhaps overplaying Rasheed) maximized the value he could get from his team. If LeBron was at 100% (and apparently he wasn’t), the Cavaliers would be expected to win based on these projections by the barest of margins. Much like good Cavs teams in the 80’s ran into great Bulls teams, a bit of bad luck – and perhaps some poor decisions – have once again denied the city of Cleveland a title. Let’s hope for the Cavaliers’ sake that these unfortunate circumstances (and the forces of economics as described in Stumbling on Wins) does not lead their relationship with LeBron down the same path as Kareem and the Bucks.
– Arturo Galletti
The WoW Journal Comments Policy
Arturo
May 16, 2010
Dr. Berri,
Thank you for the opportunity, it was great fun.
Joe
May 16, 2010
So your argument boils down to “Lebron didn’t have a great supporting cast, and the Cavs imploded when his elbow was injured.” We don’t need numbers to know that, Arturo.
By the way, why is it that your statistic seems to have no respect among communities like Basketball Reference, 82games, and Basketball Prospectus? I haven’t heard “wins produced” even mentioned there.
huey
May 16, 2010
Joe, you also forgot Arturo’s point about Mike Brown allocating minutes to the wrong players.
Apparently, we don’t need numbers to know that your reading comprehension sucks.
ilikeflowers
May 16, 2010
Regardless of LeBron’s health, this was an epic coaching fail. From the allocation of minutes to the bizarre lack of desperation fouling at the end of game 6, just stunning.
Also, note to everyone – out of respect for Joe – never post anything confirming what the eyes can see.
Arturo
May 16, 2010
Actually, the key point is that the Celts and Cavs at 100% were actually very evenly matched (think back to the year of the superteams post) and very good championship caliber teams. The Cavs and Lebron had much less margin here than is the on-going story in the media. As the magic found out today, Boston and Cleveland were playing in the big boy bracket.
Arturo
May 16, 2010
ilikeflowers
If I remember Wages of Wins, the coaching is not typically that big of a factor what was remarkable here is that we can prove statistically that Brown screwed up.
I agree that the end of that game was the most bizarre I have ever seen in an elimination game. I can’t think of another sporting event where one team has quit like that when they still had a shot.
Michael
May 17, 2010
Really I think this series boiled down to a couple of things.
1- Lebron was injured, something everyone except Hollinger seemed to ignore.
2- The Celtic roster is filled with veterans who were under represented due to the tougher schedule in the regular season. In other words they’re old, and the extra rest in the playoffs has helped them quite a lot.
Also I think if game 1 last night is anything to go by Doc Rivers made the right choice keeping Rasheeds minutes up. His defence on Howard was outstanding.
Alvy
May 17, 2010
Arturo,
Because it seems like LeBron James has been that constant 40% win producer of the Cavs for the past two seasons, I actually do wish LeBron ends his ties with that team. Aside from the composition of those two teams (projected 71 wins for Cavs, and 68 for Celtics), it seems very unfair for a .400 WP/48 mark player to be in competition against three .200, or two .260+ WP/48 players.
palamida
May 17, 2010
Arturo, I would first like to say I enjoyed this contribution immensely and I’m looking forward to see more from you.
With that said I have to say I completely disagree with the analysis.
Just a short time ago we discussed (in the comment section) the outcome of last season’s Cle-Orl series. Was it simply a random result? a random “upset”? perhaps it wasn’t an upset at all?
The latest Cleveland “premature” elimination raises similar questions but Imo – it’s a very different case than that Orlando series.
Your basic argument certainly has merit:
Boston played it’s productive starters heavier minutes, while Brown did not assign PT to some of his more productive “bench” players while giving extended mins.to inferior players such as Shaq,Big Z and Gibson and with LBJ hampered by injury. When taking this particular deployment into account that series was actually “supposed” to be quite close. As it turned out – the Celtic’s managed to take it.
That’s your basic argument as I read it.
HC adv. is very significant in the NBA, and should have, in theory, widened this 70.7-68.2 margin. From a completely subjective point of view, the Series did not appear to be all that “close”. Boston was dominating. It didn’t appear to be a case of a “coin-flip” being played out over 6 games. What numbers can we use to assess this issue?
Boston won the series in an average margin of about 5.33 Pts. I haven’t looked at these type of numbers historically but it sure does seem awfully high. Here’s a quick anecdotal look of all this season’s series that went to 5 or 6 games, just to get a little persepective:
1.)Phx-Por – 10.33 (6 games,Hc – Phx)
2.)Cle-Chi – 9.2 (5 games, Hc – Cle)
3.)Bos-Mia – 7 (5 games, Hc – Bos)
4.)Lal-Okc – 1.66 (6 games, HC – Lal)
5.) SaS-Dal – 0 (6 games, Hc – Dal).
70.7 is roughly equivalent to a 12.4 Efficiency differential. 68.2 equates to roughly 11.5. Considering the 12.4 team had Hc adv., all other things being equal if this is pure chance – Boston sure got lucky winning the series in 6 with an average margin of 5.33.
Lebron’s alleged, phantom “elbow injury” surely wasn’t a factor in this series. As a matter of fact, you would expect LBJ to produce less (compared to his regular season numbers) vs. the Celtics in a playoff series. His WS mark is obtained by playing the field, meaning playing vs. the average competition. Since The Celtics are an elite defensive team and an above average offensive team we would expect him (and everyone else, for that matter) to produce less. In fact his WSp48 mark in this series was superior (ever so slightly) to his season average. That’s hardly the mark of an injured player. If he was in fact, injured, it certainly didn’t affect his production and thus is immaterial to this discussion.
WoW fails to capture the individual match ups. While in the regular season this is a non factor, considering opponents vary, this does have some (even if exaggerated, at times) significance in a playoff series. Yes Cleveland lost this series in part because Brown failed to employ his most productive players – as much as he should have (and could have). But it was also lost in the Rondo-Williams Match up.
In increased mins, even though we would expect Rondo’s production to decline seeing as Cleveland is a vastly above average foe – it spiked. Rondo’s Wsp48 in the regular season was 10.78. Against Cleveland – it was 12.75. That’s similar to the difference between roughly 0.257 WP to 0.323.
This may not seem like such a huge spike, but the spike alone fails to tell the real story – I can’t state this enough: The 10.78 mark is obtained against average competition. We would expect a drop against elite competition. instead we have a significant increase. Rondo further exploited this match up by limiting Mo Williams!
Mo posted a 4.04 wsp48 mark which is significantly lower than his season mark of 7.05. How much of that is attributable to Rondo? That’s a whole different discussion right there, but he is surely due credit for at least some of that reduction, wouldn’t you think?
In summation (and I only touched a few key points here) any analysis of this series that fails to tell the story of Rondo vs. Williams is missing out on something. It is my contention that the result of this series was not (just) the result of randomness. If those rosters were to meet 10 more times, or a 1000 (provided they will be deployed in the same manner) – Boston would win the series more often that not.
You argue that the series was “closer” than some might think, but if match up a 70.7 wins team with Hc adv. vs. a 68.2 Wins team a 1000 times over – Cleveland would prevail more often than not. I don’t think that’s the case here.
Naturally that’s extremely hard to prove :),
But I think that the average margin of victory, coupled with how players actually performed in this series (a small sample, yet not a tiny one) gives us some insight into the matter at hand.
palamida
May 17, 2010
Errr, little correction:
I mistakenly used the SG positional average instead of the PG’s.
Rondo’s WP mark should read: 0.250 to 0.317.
I realize that Andres automated version lists Rondo’s regular season number as 0.290, I used the estimating formula provided in the past by D. Berri. the important figure is the WS mark, I simply used the formula to “put it in terms of Wp48” for those who may not be familiar with Ws.
marparker
May 17, 2010
Joe,
That line of thinking is how dinosaurs become dinosaurs.
Palamida,
How do you account for Boston being a better road team than a home team in the regular season?
re Matchups,
1. Lebron was the most productive player on the most productive team.
2. If his production is always going to be very good when the Cavs lose it will always be because another player was less productive in the post season then they were in the regular season.
3. Are matchups really the only reason that could happen?
palamida
May 17, 2010
MrParker,
While it’s true that Boston has been a great road team, it can certainly be argued that Hc adv. was less significant here than it usually is. However it’s still an advantage for Cleveland albeit, a mitigated one.
As for 1+2 – It’s not a necessity that Lebron will be as productive in the playoffs as he is in the regular season; For instance, I don’t have last season’s numbers on hand but iirc, LBJ did produce somewhat less in the Orlando series than he did in the regular season.
Moving on, while Mo Williams did produce considerably less in this season’s series – he is not a very productive player to begin with. I pointed out his production to demonstrate the net gain the Celtics “received” at the PG position – meaning Rondo outproduced his season numbers while Williams did the opposite.
As for 3 – I don’t think matchups are the “only” reason for anything. I even went as far as saying that : “WoW fails to capture the individual match ups. While in the regular season this is a non factor, considering opponents vary, this does have some (even if exaggerated, at times) significance in a playoff series”.
Meaning I’m not going all “khandor” on y’all :).
The primary argument against the matchup pov (in general) that I myself have pointed out in the past is that if matchups were such a huge deal, it’s hard to explain why “upsets” aren’t more common. In reality, adjustments are made. When slight to huge mismatches are observed – coaches adjust, players adjust and usually even within the span of a series (and in the schemes devised prior to the series – for example – Brown anticipating the Rondo problem and devises ways to mitigate it) teams find ways within their existing personnel to “plug the leaks”. Brown attempted (among other things) to assign Parker to Rondo, but that just created different problem for the Cavs. The data is self-selected: In most cases teams observe obvious “holes” and plug them – at least partially and in most cases the mismatches\”holes” aren’t that big to begin with. Williams’s total lack of defensive skills or even a basic defensive mindset coupled with Rondo otherworldly ability to exploit those very same weaknesses presented a relatively rare occurrence. In this particular series I do believe it was significant.
Don’t want to open up another Usage thread, but there are some observations one could make on the subject of Rondo’s usage in the series, but as I said, I don’t wish to go into that, here.
To wrap things up, I am not claiming matchups were the sole reason, or even the primary reason the C’s won the series.
The C’s are an elite team, especially when their starting 5 (arguably the league’s best) plays heavy minutes which has been the case throughout this playoffs. The Cavs are slightly better. With Hc adv. they were the favorites to advance. meaning the C’s ability (compared to the Cavs), brought them to the well; It was their (hardly surprising) ability to exploit the mismatch at the Pg position that enabled them to drink. from say a 65-35, or 60-40 series in favor of the Cavs, the series turned into a series in which the C’s were the favorites to advance, because 1.) The Cavs did not employ their most productive players (as Arturo noted) and 2.) They failed to address\solve their mismatch at the PG position (and to a lesser extent the KG\Jamison mismatch) in a plausible, satisfactory manner.
In this series, the Cavs should have stayed with West playing as many mins as energy and Pf allowed him to, with Williams backing him up.
WoW would argue that this is the case regardless of matchup – West is the more productive player and that’s true. However in this particular series the damage Williams’s PT caused the Cavs is much greater than the differences in their yearly Wp48 (West and Mo’s, I mean) would suggest.
I hope this clarifies my position, somewhat.
Arturo
May 17, 2010
Palimida,
First off, awesome response. I look forward to reading your first submission. You make some great points and I’m going to respond using data from the deleted scenes from the articles. The initial version of the article did the same analysis but used midseason numbers. In that scenario, the Cavs projected out to 64.9 wins and the Celts to 68.2 wins and you would expect the C’s to win the majority of the time ( with injury & coaching offsetting the homecourt advantage).
I also agree with the Rondo assessment. His 1st half WP48 was 0.276 and his full year was 0.290. This means his 2nd half WP48 was actually .304 and he was getting better going into the playoffs. As for Jamison, I think his WP48 of .184 is deceptive as he didn’t get it playing next to Shaq (who at this point clogs up the middle) he got it next to Varejao and Hickson. So my general impression is that the Celtics are a better team than their numbers and the Cavs (in their final configuration) were a worse team.
As for Lebron, Here’s the number of wins breakdown for the best players of the teams in the second round of the playoffs:
LeBron James Wins 24.778 Rest of Team: 33.072
Dwight Howard Wins 22.27 Rest of Team: 37.865
Rajon Rondo Wins 17.901 Rest of Team: 32.659
Josh Smith Wins 17.824 Rest of Team: 35.433
Steve Nash Wins 16.182 Rest of Team: 37.634
Carlos Boozer Wins 16.316 Rest of Team: 38.638
Pau Gasol Wins 15.519 Rest of Team: 37.616
Tim Duncan Wins 15.695 Rest of Team: 38.603
Only Rondo had a “worse” team around him based on Year end numbers (and we already proved that was an aberration :-) ) . Against a team like the Celts this year or the Magic last year Lebron has to be extraordinary every night or the Cavs have no answer.
todd2
May 17, 2010
Watching the Cavs reminded me of watching Jordan score 63 points in the playoffs against Boston and losing. Is there an optimal distribution of productivity for a team to be successful?
Owen
May 17, 2010
Good stuff Arturo. I agree with your premise. The Celts were a lot better than their efficiency differential indicated. And the Cavs were a lot weaker, especially considering their two best players in the regular season were playing hurt.
I think what I would have loved to seen is the WP for the series itself matched up against the projections. Certainly the Celtics were better than we thought. But how bad did Shaq and Jamison play. What evidence do we have to support the notion that Brown didn’t field his best possible lineups. Were Varejao, West, and Moon as productive in their curtailed minutes as they had been in the regular season?
Ben Guest
May 17, 2010
Great job!
palamida
May 17, 2010
Great comments.
Arturo, seeing as I thought the whole idea was to enable (using the automated version) people who follow a certain team closely – “cover” that team ; and seeing as I do not have any special affinity to any particular team nor do I follow any particular team more closely than others – I did not voulenteer to participate in this welcomed endeavour. I will however continue to follow closely and continue to offer my notoriously lengthy comments :p.
I think you hit the nail on the head with Jamison as well, who wasn’t nearly as productive playing alongside Shaq. Not only was he abused by KG on both ends of the floor, this Jamison\Shaq duo was simply abysmal defensively and and it was so blatantly obvious I think anyone with a set of eyes was asking himself: “Where the f is Varejao”? I know I was.
Seeing your star vs. rest of the team breakdown made me think back to Dberri’s preseason prediction (which I wholeheartedly agreed with): i.e The year of the super teams. Anyone remember that? :)
Owen, I think just about anyone would like to see data along the lines you’re talking about.
For instance, doing the numbers by hand took a couple of mins in my initial comment and that was just for 3 players! that deterred me from doing Jamison for example, whom i’d be interested to see as well. (not to mention, it’s a method prone to clerical errors :p).
I think the day-to-day, month by month or series splits are probably on top of Andre’s to do list anyways, but be that as it may – I second the request…
So Andres, if you can hear us, we need you! :)
Arturo
May 17, 2010
Owen,
I’d love to see that to. The current tools do not allow us to isolate common cause variation (game to game noise) vs. special cause variation (actual change in the process) very well. The Celts in the playoffs are a classic case of special cause.
Owen,
Here’s a table final participants since 2006 and the final four this year (using the corrected Celts numbers)
Team Year Best Player Wins Team Wins % of Team wins (Best Player)
Heat 2006 Dwayne Wade 18.2 51.2 36%
Mavs 2006 Dirk Nowitzki 18 57.4 31%
Spurs 2007 Tim Duncan 20.1 63.5 32%
Cavs 2007 Lebron James 17.4 51.3 34%
Celts 2008 Kevin Garnett 17.9 68.3 26%
Lakers 2008 Kobe Bryant 16.4 60.4 27%
Lakers 2009 Pau Gasol 15.6 61.2 25%
Magic 2009 Dwight Howard 22.2 58.6 38%
Lakers 2010 Pau Gasol 15.5 53.1 29%
Celts 2010 Rajon Rondo 19.7 68.2 29%
Magic 2010 Dwight Howard 22.3 60.1 37%
Suns 2010 Steve Nash 16.2 53.8 30%
Less than a third ( and closer to 25% seems to be the very best)seems to be best with the only exceptions being Wade in 2006 (and we will not go into that one) and the Magic. The Magic would be the most unbalanced team to win in that data set. The data seems to imply that you do need an outstanding player to win but you cannot be dependent solely on him. I’d love to do it for more seasons though.
ilikeflowers
May 17, 2010
todd2,
I would imagine that the optimal distribution of some arbitrary total wp48 would be all starters with the same wp48. This would restrict the opposing team’s defensive strategies and dampen the impact of random effects (both positive and negative) making the team very consistent relative to other combinations.
Someone recently did a study on this issue actually, although I can’t remember where I came across it.
ilikeflowers
May 17, 2010
professor,
has there ever been a team that had five starters where each had approximately wp48 0.200? I would think that it would be an unlikely occurrence.
palamida
May 17, 2010
Ilikeflowers, i think it’s worth noting that while the average NBA player at each position – meaning all the players employed, without weighting for PT, does average around 0.100 – the average is different when accounting for PT is quite different. To illustrate this point, here are the numbers for players who played 30+ mins a game, this last season.
WS48 – (and estimated Wp48):
PG – 7.72 Ws48\ 0.148 Wp48
SG – 6.98 \ 0.129
SF – 9.22 \ 0.164
PF – 12.9 \ 0.187
C – 14.38 \ 0.218
The cutoff is obviously arbitrary (30 mins), but this is just to illustrate a point, roughly, so I don’t think that’s crucial. I used the 15 years positional avg. and not this season’s, but I imagine the results wouldn’t be too far off anyhow.
My point is, that an actual starter who plays heavy mins in the league produces on average, marks that are significantly higher than 0.100.
As a matter of fact a team comprised of “starter’s average” players wouldn’t be so far removed from your 0.200 team – it would average 0.170.
NBA Subs – matter. and the ability to stay on the court for longer stretches (provided you’re a productive starter) also has value. Who needs a backup PG when Rondo plays 44 mins a game in the playoffs? not everyone can do that, and those that can – provide value beyond their usual production merely stretching over a span of a few more mins: It allows their team not to take the hit that is normally associated with using your reserves.
This drop-off is especially steep in the front court.
I’d suggest to anyone whose interested in the subject to read a recent post from Ty over at the Courtside Analyst, regarding this issue.
Not sure if I can link here, by anyone interested shouldn’t find it too hard to find.
Ilike, I just thought it’d be fun to see how close a team of “starters” averages in the league and how close is it to the 0.200 mark.
I think 0.170 can be classified as pretty darn close.
TBall
May 17, 2010
Arturo,
A nice article, and I am a fan of both the Celtics and WoW, but I have some issues.
I don’t think you can cherry pick the portion of the season you think represents the Celtics of the playoffs and use that as a statistical basis for comparing teams. According to Prof. Berri, Rondo’s WP48 is the same if Pierce is healthy or isn’t. And Perkins is banged up now, so which numbers should we use for him? The value of a good statistical model is being able to anticipate future events, not rationalize the past. I worry if the Cavs had won, we’d be comparing the 2nd half Celtics numbers to Cleveland’s full season.
The games have been played, why not use the data? How far did LeBron stray from his predicted performance? If that elbow was Cleveland’s only problem, you’d expect the other players to perform to their season averages.
Compare the regular season and the postseason. LBJ shot 50% from the field and about 75% from the line, averaging 29 points/game, with one less assist and two more boards per game in the playoffs. More blocks, more steals, and more trips to the charity stripe in the playoffs. Was there a WP48 dropoff for LBJ in the playoffs? Blame seems to be assessed here without clear statistical correlation. If anything, I thought Lebron took the team 43% of the way to victory but his team failed to drag themselves the other 57% of the way there.
How about comparing regular season WP48 and post season WP48 and seeing where the C’s stepped up and/or the Cavs fell off? At least then you have some numbers to dissect for the autopsy.
TBall
ilikeflowers
May 17, 2010
palamida,
That’s certainly interesting but what I’m trying to get at is what is the impact of the distribution of wp48 among a team’s players.
There are two ways that I can see to approach this.
[1] You can examine whether teams with more evenly distributed wp48 are systematically over or under performing relative to their efficiency diffs. My guess is that (on average) a team of 5 x 0.200 players is going to have a better record than a team with 1 x 0.400 + 1 x 0.200 + 3 x 0.100 even if they have the same efficiency diff and identical benches and minute allocations. Of course the reality may be the reverse or the ideal distribution may be something else entirely.
[2] You can examine the effect of distribution when two teams of similar overall average wp48 play each other. This probably happens frequently enough so that enough data is available for a definitive answer.
palamida
May 17, 2010
Tball, while I agree with you about the cherry picking argument in general, I think the whole idea here was to cherry pick.
The C’s veterans claimed all through the second half of the season that they’ll show “what they’re made of” come playoff time. Many, myself included were extremely skeptical of this alleged ability to simply “flip a switch”. With that said, a few things should be noted- If ever was a candidate for such a “switch flip” the C’s would have to be it:
Prior to the season, I thought that if Sheed’s poor play last season (his last Pistons season, that is) stemmed from the overall “tanking” and bad overall atmosphere at the then “underachieving” Pistons and not due to his progressing age, the C’s will be quite good, in fact I thought that If the “big 3” would not suffer serious declines in production (due to age), and furthermore if KG would recover fully from his knee injury – the C’s are the favorites to take home rings.
Sheed (if and if) would have been a wonderful backup at both PF and Center positions, Tony Allen is a great backup on the wing and Shelden Williams who has an extensive history of being productive (albeit in limited roles) would also be a great addition – especially if Shelden and Sheed would take mins away from the godawful Glen Davis. My point is, that when the season started and the C’s went 20-2, I wasn’t surprised the least bit. It was expected. I was surprised when they started losing, and when KG got injured again, I figured their season was done for. I do not know how or why they played their second half of the season the way they did. I do suspect though, judging by what this team has done 2 years ago and last year (prior to Kg’s injury) and by their start of this season + my preseason expectations of them that the first half of this season represents their “true” ability better than the second half. I don’t think that’s i’m going on a limb here when I argue this even though admittedly – it’s hardly factual.
With that in mind, I think the “cherry picking” was an interesting take and a logical step in the analysis of their season. I don’t think Arturo was claiming that the second half accounts for nothing, or that you can simply pick arbitrary chunks of any team’s season and make valid arguments on a regular basis. With what we know of the C’s past performance, I think it’s a valid and interesting “experiment” – since this specific “cherry picking” holds some insight and value and has definitely provided food for thought, which is probably the most a post can aspire to achieve.
Just for kicks, I did look at their overall season numbers: If the C’s had assigned PT the same way they did vs. the Cavs throughout the entire season (and using, an entire season’s data) we would expect them to win roughly 61.7 games.
Well off the 68.2 mark, but also far removed from the 50 games they did win, which btw is the same as their Pythagorean suggests.
ilikeflowers
May 17, 2010
Just want to address a typo in my post:
1 x 0.400 + 1 x 0.200 + 3 x 0.100
should be:
1 x 0.400 + 1 x 0.200 + 3 x 0.133
so that this team has the same avg wp48 as the other, more evenly distributed team.
Arturo
May 17, 2010
ilikeflowers,
The closest I could find is the Celts in 08 & 09 and the lakers in 09
http://www.wagesofwins.com/Boston0708.html
http://www.wagesofwins.com/Boston0809.html
http://www.wagesofwins.com/LALakers0809.html
The Celts missed at center and the lakers missed at point (which is still their glaring flaw) but these were some good teams.
palimida,
The whole point was to cherry pick to block out the source of variation for the Celts (injury).
Daniel
May 17, 2010
Long comments…
kevin
May 17, 2010
“By the way, why is it that your statistic seems to have no respect among communities like Basketball Reference, 82games, and Basketball Prospectus? I haven’t heard “wins produced” even mentioned there.”
Because the people who run those sites are fools with an overblown and undeserved reputation to protect?
Just throwing that out there.
kevin
May 17, 2010
Aside from the raw numbers, the Celtics were playing all year for the playoffs, so the team they are bringing to the table now is a lot better than than the one that went 27-27. On top of that, the Garnett playing today and the one that showed up for the regular season are two different people. He has gotten back a lot of his lateral quickness and he is shooting noticably better. He’s back to being a a +.300 WP48 player now.
Using WP48, and balancing the numbers for current PT, I bet the Celtics are a better team on paper, if the paper in question used WP-like evaluation numbers rather than the PER or Wins Shares crap.
You know, in a way, I hope PER and Win Shares win the “court of public opinion” battle. It will make it so much easier to make a killing in Vegas on these playoff games. The cat is out of the bag this year but next year, I’m already planning my trip to the Galapagos with my son, courtesy of moron fans who believe absolutely anything attached to the word “scientific”.
Arturo
May 17, 2010
And we just got linked by truehoop,yay :-)
Dre
May 17, 2010
Kevin,
I will say it is easier to buy WP in a city like Denver, where you are told you have superstars by many of the popular metrics and yet are mediocre most of the time (at least by the standards of how good our superstars are)
Many of these metrics to people who do not understand stats look alike. Win Share, Wins Produced and PER all use similar inputs and even have similar output (Win Share and Wins Produced both estimate wins, PER even has an additional Wins column at ESPN)
Of course some people who dig into the stats may have a better understanding, for instance PER rewards players for taking shots in SEVERAL areas, regardless of if they make them or not.
The way to convince people of the WP48 is to explain and possibly listen to complaints and even stop and think about them if they seem valid. Many of the people on this forum do that, and I hope that is our method of helping other people understand. It took On Base Percentage 100 years to catch on, we can patient with WP48.
Also Basketball-reference.com is a great site and much of my own WP48 would not be possible without them.
palamida
May 17, 2010
Arturo, read that comment again – I was defending the “cherry picking” :)
Experts, experts, experts… If they all agree for instance, that John Wall is such an incredible, one of a kind, franchise changing prospect – then let them put their expertise into practice, or rather put their money where their mouth is:
How many more wins do they expect a team drafting Wall will get? To be fair, they can wait till after the lottery so they’d know whom Wall would be displacing :p
It wasn’t that long ago that another ton of upside\athletic\young scoring PG was selected 1st overall and was heralded all year long by every pundit and expert from here to Timbuktu as the “next big thing”.
As a result, Bulls fans were overjoyed when they defeated the odds and got the 1st overall pick.
In the ensuing season, the Bulls (who were a 33-49 team) went 25-30 until the deadline trade that got them Brad Miller and John Salmons. They then proceeded to go 16-11 and squeak into the Playoffs where they were eliminated by a Celtics team with Big baby at PF. (in an admittebly close and “thrilling” series).
The next season they continued to be the very face of mediocrity and finished the season 41-41 for the 2nd straight season while just missing out on a playoff berth.
All those interviews, all those workouts, all that scouting, and 2 seasons, one Roy award and one AS appearance later a simple box score derived method is still the only one telling the story of Derrick Rose – the Average NBA Pg.
This off course is purely anecdotal but it makes for a great story :p.
While Wall probably has bigger “upside” (age,size) he will not be the best PG in the league until 2020, As dubbed by Chris Reina of RealGm.
He won’t even be in the top 5. In his prime. Well, maybe if gets to play against Mo Williams 82 times in a season.
Demarcus Cousins? that’s a different story and one lucky team – with an emphasis on lucky.
So once again a riddle for all the “experts” out there:
How many wins will Wall add to his new team?
and how many will Cousins add?
John Hollinger just claimed that Rose is the 8th “most ideal sidekick for LeBron James among the other free-agents and players on teams that have enough cap room to sign James”.
Seriously, I’m not making this up. :p
All this makes me think of a classic Hans Christian Andersen tale…
AR
May 17, 2010
Does it matter for winning in the playoffs how WP is distributed on a team or more specifically whether you have 2 vs. 3 strong players by WP?
(or 2 or 3 strong players by scoring or Adjusted +/- or a combination of them)
It might and if it does then it might also be the case that player pairs and lineups and match-ups matter. And player activity not captured in the boxscore.
One way to try to learn more before answering the question is to check things out from those other angles too and then make your philosophical and practical decisions.
At a basic level WP is a regression based linear weights model, as is Statistical Plus / Minus but WP adds a team adjustment.
Win Shares has the team defensive adjustment but uses Offensive Rating based on complicated mathematical manipulation rather than regression.
I am not sure if PER was aided by any regression study initially but I think I heard Hollinger saying it was aided by some, if not initially then in its review. It simply chose to be agnostic about shot defense in the face of poor data.
Adjusted +/- is a stand alone metric for some, but to me it might be best used as a small piece of an overview evaluation. None of the other metrics mentioned above attempts to measure individual player contributions on offense beyond the boxscore or individual defensive contributions beyond the boxscore.
AR
May 17, 2010
Or to be more precise, none of the other metrics mentioned above attempts to measure individual defensive contributions beyond the boxscore or “team level defensive data.
And that is team level defensive data and not for player on the court only or by specific lineups and I think you lose a lot of accuracy about individual shot defense impact when you use team level defensive data.
AR
May 17, 2010
Another clarification: Win Shares’ defensive component is really a blend of individual stops and team shot defense presented together (rather than in separate steps of individual boxscore then team adjustment) thru the use of Defensive Rating.
AR
May 17, 2010
Individual player contributions on offense beyond the boxscore would include passing beyond the assist, spacing, blocking out, saving turnovers, “being a offensive threat” and possibly to the level of requiring a double team, taking and making a decent amount of clutch shots, etc.
Individual player contributions on defense beyond the boxscore includes the very important component of shot defense (1 on 1 and help) but would also include pass disruptions & preventions, block-outs, rebound tips, burning clock with pressure, etc.
John Giagnorio
May 17, 2010
It’s pointless to try to debate guys like Derrick Rose and John Wall. The people who like them made up their minds ages ago, and no evidence is going to change that.
Arturo
May 17, 2010
AR,
All the metrics have some value. I like adjusted plus/minus. But I have big problems with PER. To quote Prof. Berri:
“Hollinger argues that each two point field goal made is worth about 1.65 points. A three point field goal made is worth 2.65 points. A missed field goal, though, costs a team 0.72 points. Given these values, with a bit of math we can show that a player will break even on his two point field goal attempts if he hits on 30.4% of these shots. On three pointers the break-even point is 21.4%. If a player exceeds these thresholds, and virtually every NBA played does so with respect to two-point shots, the more he shoots the higher his value in PERs. So a player can be an inefficient scorer and simply inflate his value by taking a large number of shots.”
PER rewards the squandering of possession which in basketball is failure. It also has no correlation to actual wins (which is what puts fannies in seats).
I do however agree that additional statistics in the Box score would be nice ( say for example more detail on blocks)
AR
May 17, 2010
While acknowledging estimate error and need for caution, it still happens that 1 year regular season Adjusted +/- also suggests that positive impact was highly concentrated in LBJ’s play.
And that Varejao (especially a healthy Varejao) was very important, that O’Neal and Williams were among the weakest Cavs (probably pulled down well below what WP captures because of their indirect impacts) and that Moon and West were at neither pole.
Not always agreement, but there is here.
If team adjustment in WP for a good defensive team lifts everyone, then it may be useful to also look at the distribution of PAWS before the adjustment when judging the breadth of strength by position in a starting core and other data that also gets at individual defensive contribution beyond the standard boxscore as made available at 82games and Synergy.
And given the general compression of rotations in the playoffs, the value of players truly individually strong on both sides of the court is probably elevated in the playoffs.
AR
May 17, 2010
Despite big differences in methodology, WP and PER rated Mo Williams about the same relative to league average in the regular season and the difference in ratings for Shaq was fairly modest and due to his high usage rate.
When near neutral on these 2 boxscore based metrics, the perspective of Adjusted +/- is at least another perspective and might help in making an judgment of whether they are better than average or not in overall impact in the eyes of some.
With respect to these two, traditional Adjusted +/- estimates they were negative on 1 year and 2 year and they were on 4 year RAPM as well. I tend to agree. And losing in 6 is consistent with at least a sense they were not going to be notable forces in beating a high quality team.
Mo
May 18, 2010
If LeBron’s elbow was injured so badly, why was he shooting half-court circus shots before game 6?
Dre
May 19, 2010
Arturo,
Awesome article! By the way I did get playoff data up and an interesting fact about the Cavs the last two years: Lebron has been the ONLY player on the team responsible for more than 2 wins in the playoffs. So for a total of 16 playoff wins in two years, Lebron has been responsible for almost 10 of them!
brgulker
May 19, 2010
First off, Arturo, great post!
Second, I’m surprised that so many people are nitpicking this post. The point (as I read the article) is to show that Boston, when fully healthy and as a very good team on the road, was much more evenly matched with Cleveland, when fully healthy and in spite of their home court advantage.
In order to make that point, one obviously has to “cherry pick” the stats/differentials/records throughout the season — how else could one make that point?
Boston and Cleveland, all things considered, were about evenly matched, and for six games, Boston outperformed Cleveland. Brown played the wrong players, Rivers played the right ones (except Sheed over Williams), and Boston won.
I love this blog, and I love these numbers, but I think we’re over-analyzing this one.
brgulker
May 19, 2010
Typo, “was much more evenly matched … than their final records indicate.”
kevin
May 26, 2010
“Rivers played the right ones (except Sheed over Williams), and Boston won.”
don’t know how you can say that since Wallace played so well and was of critical value on a couple of wins.
Kevin
August 25, 2010
I am a little late posting here. It would be interesting to compare bostons efficiency for the the first three games and h the first half of game 4 to the the second half of game 4 and the last two games. I believe this is when mike brown actually created bigger issues with the williams rondo mismatch trying to plug it than the actual impact of the leak.
keep in mind that even with rondo out performing williams the cavs still won 2 out of 3. plugging the leak and allowing other mismatches were seemingly far more detrimental to the team.
ultimate i do not think the williams/jamison rondo/garnett mismatches is what cost the cavs the series. I would agree with arturo a consistent 8 man rotation of west,varejao and moon would have gone a long way to making the series games more comptetitive.
mike brown has done this before . the previous year he started ben wallce 53 games then had been magically disappear only seeing at most 15 minutes a game for the playoffs. then doping the same to hickson a 73 game start who averaged 20 minutes a game. I would be interested if there was any precedent for a head coach to yank a player from their regular season rotation once entering the playoffs