2011 Finals - Mia vs Dal

[Mike G]

I have Wade as far and away the most valuable player of the series with 1.76 WARP.
...
I tend to take stats for these short series with a large grain of salt, though.

I've ordered columns and added eWins for easy comparison.

Code: Select all

TEAM   NAME             WARP   eWins
MIA   Dwyane Wade       1.76   1.33
DAL   Tyson Chandler    1.03    .42 
MIA   Mario Chalmers    0.63    .39
DAL   Jason Kidd        0.53    .29
DAL   Dirk Nowitzki     0.50    .95

DAL   Shawn Marion      0.45    .39
DAL   Jason Terry       0.45    .61
DAL   DeShawn Stevenson 0.45    .40
MIA   LeBron James      0.36    .62
MIA   Mike Bibby        0.19    .04

TEAM   NAME             WARP   eWins
MIA   Juwan Howard      0.10    .03
MIA   Eddie House       0.06    .03
DAL   Brendan Haywood   0.04    .01
MIA   Udonis Haslem     0.02   -.01
DAL   Jose Barea       -0.04    .24

DAL   Brian Cardinal   -0.10   -.02
DAL   Ian Mahinmi      -0.10    .02
MIA   Mike Miller      -0.10    .01
MIA   Joel Anthony     -0.20   -.07
MIA   Chris Bosh       -0.24    .37
DAL   Peja Stojakovic  -0.24   -.06

I thought Bosh underperformed, but was he really the worst player on the team?
Was Chalmers really better than Terry?
I don't know how ezpm works; can you explain some of these departures from intuition?

[EvanZ]

Bosh's TS% for each game:

Code: Select all

1 40.8%
2 33.8%
3 44.6%
4 53.3%
5 59.5%
6 78.6%

While Bosh had a very efficient games 5 and 6, his first 3 games were very inefficient and game 4 was below average. It looks to me like the poor shooting was the main factor.

[Mike G]

Bosh was also the team's best rebounder, along with Anthony (for 21 mpg). The Heat won 2 of those first 3 games, and .533 shooting is not below average for either the series or for game 4 (Mia .485, Dal .499).

I'm just having a difficult time grasping how a replacement player could have been expected to help more than Chris Bosh in this series.

[EvanZ]

Bosh was also the team's best rebounder, along with Anthony (for 21 mpg). The Heat won 2 of those first 3 games, and .533 shooting is not below average for either the series or for game 4 (Mia .485, Dal .499).

I'm just having a difficult time grasping how a replacement player could have been expected to help more than Chris Bosh in this series.

Mike, do those team averages go into your calculation?

[xkonk]

Does any other metric support the notion that Wade was a lot more valuable than Nowitzki?

Basketball-reference has LeBron and Wade with more Win Shares than Nowitzki for the playoffs on the whole, although it's close; I don't know if they break it down by series. And Arturo's Wins Produced numbers also prefer Wade http://arturogalletti.wordpress.com/201 ... mavericks/ . So along with eWins and ezPM, I guess that would be four metrics.

[Mike G]

I was definitely inquiring about the Finals. B-R.com steadfastly refuses to keep track of individual series.
And WP is not a metric. I don't know what it is. Dirk at #5 on his team? Does anyone take this seriously?

[Mike G]

Bosh ... .533 shooting is not below average for either the series or for game 4 (Mia .485, Dal .499).
.

Mike, do those team averages go into your calculation?

It's a small factor, but the ratio of Bosh's Eff% to the Heat's Eff% (.485/.531) is raised to the .30 power.
This factor (.973) reduces his per36 scoring output, adjusted by several other factors, to 16.7 .

The biggest element in his scoring contribution, of course, is his 18.5 points per game. This is 20% of what his team scored, and they weren't going to get these points from anywhere else, if Bosh isn't shooting. We know he hit a game-winner, and no one else was going to hit it, because time was running out.

If the 2nd-best team's 2nd scorer and top rebounder is a negative contributor, he must have been doing an amazing amount of bad stuff, too. What would that have been?

[EvanZ]

The biggest element in his scoring contribution, of course, is his 18.5 points per game. This is 20% of what his team scored, and they weren't going to get these points from anywhere else, if Bosh isn't shooting. We know he hit a game-winner, and no one else was going to hit it, because time was running out.

If that were true, Bosh should be +18.5 (on offense alone), but he's nowhere near that value. If he didn't score those 18.5 points, you really think *nobody* else would?

[Crow]

Looking around for possible explanations...

Shot defense being 100% counterpart in EZPM's WARP may have hurt some players, perhaps including Bosh (the defender of record for much of Dirk's scoring) compared to any metric that implicitly assumes equal shot defense credit / blame or excludes it.

The assist weight in EZPM may have helped push up the PGs who passed compared to eWins. The PGs who shot well with a relatively high frequency of unassisted FGs may have benefited in comparison to others with lower frequency of unassisted FGs in EZPM and in comparison to how they were scored in EWins.

[EvanZ]

Crow, one thing I changed recently is that I lowered the weight for 3-pt assists by 1/2. Doesn't make a huge difference to the final outcomes, but it seemed philosophically correct. And 82games had evidence that assists didn't help as much for 3-pt FGA as 2-pt FGA. Also, I'm currently splitting the counterpart defense 50/50 between individual and team. Again, doesn't make a huge difference, but for a short series especially, it seems to reduce the variance to levels that make more sense.

The biggest changes I could see making are to take into account the quality of opponents and teammates (perhaps, using RAPM?), and incorporating some kind of USG*EFF factor. I'm just not sure how to do it yet in a linear weight metric like mine.

[huevonkiller]

We're talking about a player that messed up his elbow trying to play Jordan-like minutes. He needs a bit more time off and rest, he's a massive player.

Mike I give Shaq credit for his high minute usage, but his offense is different than Kobe, Wade, Lebron.

Wade was just as great this time around, but he wasn't equal to Dirk + Jet.
Terry needed just 33 mpg to get the same eWins he got in 40 min. the last time.
Nowitzki got more eW in fewer minutes.

LeBron averaged almost 44 minutes, but that's not unusual for a 26 year old superplayer.
In '08 and '09, Kobe averaged 43 in the Finals.

He averaged 41 in those post-seasons.

In Kobe's most dominant title run he played 36 MPG in the regular season.

Both Wade and Dirk went 44 in '06.

Dirk didn't win a title. Wade averaged 41 the first three rounds and he's two years younger.

Dirk won a title playing 39 MPG.

In '04, Kobe went 46, Shaq 43, Rip 44.

Kobe was terrible, Rip is not in the same category. Shaq averaged 21 PPG.

In '03, both Duncan and Kidd went 44 mpg.

You can't compare the 2011 Mavs, to the 2003 Nets. That was a weak era for the East.

In '02, 5 players averaged 40+ (Kobe 44).

Kobe had a terrible post-season overall for LeBron standards.

2001: Iverson 47, Kobe 47, Shaq 45, Mutombo 42

Kobe was dominant in this playoff run and he struggled in the Finals. Iverson is not on this tier.

2000: Shaq 46, Rose and Miller 43
'99 : Duncan 46, Spree and Houston 44

Rose and Miller, really? That's not comparable. Duncan did it in a shortened season.

Jordan MPG in Finals: 44 (Magic 46), 42 (Porter 44), 46 (Majerle 47, Barkley 46), 42 (Payton 46), 43, 42 (age 35).
Shaq and Hakeem both averaged 45 in '95. Horry 47.
Hakeem and Ewing 43 and 44 in '94.

Hakeem did not have a great 94 Finals, and Shaq played 38 MPG in his Finals run with Orlando. Hakeem has a WS/48 of .143 in his other Finals run.

Mike we're talking about 44 for the entire playoffs, and don't use role players that's a bad comparison.

Most Superstar perimeter players of this era don't normally play that many minutes. Kobe seemed to fail at the end, or throughout the playoffs when he got that much time. Wade doesn't play like that nor does Jordan in his title runs. And neither of those three weighs 250-260 pounds.

[xkonk]

And WP is not a metric. I don't know what it is. Dirk at #5 on his team? Does anyone take this seriously?

Dirk shot 41.6% (not very good) for the series, including 36.8% from three (good, not great, and below his season average). He made up for it at the line, but had two games where he didn't get the line much at all. In the end, he scored less than a point per possession by my calculation. He did rebound pretty well (although virtually all of it on the defensive end), but isn't WP supposed to value that through the roof? He had no assists, steals, or blocks to speak of. In short, all he did was shoot a lot in a not very efficient manner and get a good number of defensive boards. I don't see any reason to say that Dirk had a good series in an absolute sense, although I guess you could argue about how he did compared to his teammates.

[huevonkiller]

Going by game score Dirk didn't have a great series, but then again Wade faced a much easier matchup. Joel Anthony and Chris Bosh are not slouches defensively (not sure what to make of Haslem this season).

Still it was a team win. Reminds me of Kobe's 2010 Game 7 against Boston.

[Mike G]

If he didn't score those 18.5 points, you really think *nobody* else would?

Right. Others may make up 10 or 15 of them, but not all 18.5 .
He's also one of the team's better defenders, so they lose some defense without him, too.

I lowered the weight for 3-pt assists by 1/2. Doesn't make a huge difference to the final outcomes, but it seemed philosophically correct. ...Also, I'm currently splitting the counterpart defense 50/50 between individual and team...

Counterpart? Does this mean that if 2 of the best players in a series happen to play the same position, they can't both be great? One has to be lousy, or both have to be mediocre? That might explain why Kidd was great vs Westbrook 'horrible'.

I advocated that a 3-ptr assist should be 1.5 times a 2-ptr assist. Is that where you are now? Or are they equal?

[Crow]

Alright Evan, I wasn't aware of your recent EZPM changes. They make sense to me. I can put down those suggestions and focus on the data and other changes / possible improvements.

(Are these changes in the season file currently posted at your site or just the playoff postings?)

Taking into account the prior quality of opponents and teammates outside the immediate period of performance measurement perhaps using RAPM could push EZPM into hybrid territory which usually sounds interesting and could be helpful.


Mike G method of valuing player scoring in the context of opponent scoring was one game level / boxscore-based way to address defense and pace different than the approaches of other metrics.

Incorporating some kind of USG*EFF factor could be helpful. At this point you can consider what DSMok1 has done, I think bbstats work as well and maybe MikeGs too. It is possible that incorporating some kind of USG*EFF factor along with taking into account the quality of teammates and opponents would help better assess for exaample how much Bosh's 18 pts per game was really worth or how "necessary" it was. Worth and necessity may or may not be best measured at the same time in the same metric. Perhaps necessity of performance could be a kind of broader clutchness companion metric somewhat similar to what Wayne Winston does with his Adjusted +/- and a separate "Impact Rating".