How good is the typical supporting cast of a Finals team?

[colts18]

mysticbb made a post recently stating that the average supporting case of a champion is below average. I thought that was interesting so I went out investigating to find out. Turns out he was right. I went back and looked at plus/minus data of stars of finals winners and losers since 97 and the games missed record of finals winners/losers since 1986. Both of those data sets showed that the average supporting cast of a finals team is average to below average.

For the most part, figuring out the teams star player was easy. The only assumptions I made was Ben Wallace for the Pistons and Allan Houston for the Knicks

Here are the plus/minus of star players of championship teams from 97-12 (16 seasons):
on court: +11.0 per 100 possessions
off court: -1.1
Net: +12.2 per 100
They played 71.9% of their teams minutes (equivalent of 34.5 minutes per 48 minutes)

The average star title winner is picking up his team by over 12 points per 100 possessions. Also the average supporting cast of a title team is below average.

11 out of the 16 had negative off court values. The highest off court value was KG at +4.6, the lowest is Shaq 01 at -7.6. Only 6 of those guys had on court values of less than 10 (MJ, Shaq, Duncan, Wallace, Wade, Kobe).

Here is that same data for the championship losers from 97-12:
On court: +8.0 per 100 possessions
Off court: +0.2
Net: +7.8 per 100
They played 73.9% of their teams minutes (equivalent of 35.4 minutes per 48)

8 of them had positive off court values. Reggie Miller actually had a negative plus minus (-.9). Durant had the highest positive off court value at 5.6.

So based on that, the average finals teams supporting cast is average to slightly below average.

I looked at every finals winner and losers average supporting cast in the games that the teams star missed (86-12):
star player of Finals winner:
games missed: 58-61 record (.487, 40 wins per 82), -0.42 margin of victory
games played: .735 win% (60.3 win per 82)
Difference: .248 win%, 20.3 wins

When a star misses a game, his team plays below average, but when he plays, his team plays like a 60 win team. That is a 20 game lift by a star.

The best supporting cast was 08 KG since his team went 9-2 without him while the worst were 94 Hakeem (3-7), 11 Dirk (2-7), and 92+93 MJ (combined 1-5)

*note: the games played win% is weighted by the amount of games missed. That means 96 MJ's team record is not included since he didn't miss a game while dirk 2011 record is worth 9x more than Duncan 03 since his team played 9x more games without him.


Star player of Finals losers:
Games missed: 58-63 record (.479, 39.3 win per 82), +1.01 margin of victory
Games played: .708 (58.1 wins per 82),
difference: .229 win%, 18.8 wins

Same story here too. The star of the finals loser had his team struggle without him too. The worst record was 93 barkley when his team went 1-5 without him.


Overall the finals teams went 116-124 when their star player didn't play (240 game sample). That is big enough to conclude that these teams were average without their star, and finals worthy with him.

[mysticbb]

No idea, but your off-court numbers are wrong. I guess you picked the pace numbers from NBA.com or even the wrong adjusted values right away.

For championship winners I have -2.2 and for the Losers -1.5. I have 11.2 for the championship winner with the best player on the court, and 8.3 for the runner-up. The championship winner had in average +7.4, the loser +5.8, thus in average the better team won. For the championship winner, the minutes without the star player were 28% of the overall minutes, for the runner-up it was 26%.

The more interesting thing for me is that the correlation between off value and minutes is -0.1 for both, which means, the more minutes the star player missed, the worse got the team without him on the court in that sample. Overall that suggest that the results can't be dismissed with "garbage time".

[Colbinii]

I think you have to take these numbers with context. Championship teams are built around your star player. Obviously your team is going to suffer more with him out as opposed to a team without a star player.

[mysticbb]

I think you have to take these numbers with context. Championship teams are built around your star player. Obviously your team is going to suffer more with him out as opposed to a team without a star player.

That is an idea, but it is wrong. Championship and runner-up teams are above average in terms of performance level without the respective best player. In average a team is about -4 without the respective best player on the court. The other teams are suffering as well.

Chemistry and fit matter, but that goes for all teams, not just championship teams. The difference is that the championship teams had the better star players and the overall more talented supporting casts.

[ronnymac2]

Are you trying to figure out how good these supporting casts are as stand-alone teams, or as teams around a superstar?

[Carmelofan]

It's impossible for a championship team to have a terrible supportin cast. Each championship team has at least a decnt supporting cast. The worst one i can think of is the 2012 heat. Their supporting cast was terrible but they won because they played no title contenders in the east.

[jman2585]

It's usually not how good it is (though that matters too), it's how good it is relative to their opposition at the time. That's why the 94 Rockets (who were not a historically weak support cast per se), are regarded as right there with the 03 Spurs, the 76 Nets and the 77 Blazers- because even though the support casts were worse for the latter 3, Hakeem had (generally) tougher opponents.

[mysticbb]

Are you trying to figure out how good these supporting casts are as stand-alone teams, or as teams around a superstar?

As stand-alone teams. The question was how health should effect a player's ranking for a specific season. Elgee was arguing that health for the regular season is basically meaningless, because championship teams would make the playoffs anyway. He was using examples like the current Bulls or even the 1986 Bulls, not accounting for the fact that both are outliers. No team would make the playoffs with such a bad record as the 1986 Bulls today, and the current Bulls are playing above 0.500 ball, which is well above league average for supporting casts, even for chanpionship supporting casts.

It should also be noted that the best record for a non-playoff teams was in average 43-39, that means a star player can't miss 2/3 of a season and can still expect his team to make the playoffs for sure.

While health is way less important than some make it out to be, it is not as unimportant as Elgee wants to painting it.

[colts18]

No idea, but your off-court numbers are wrong. I guess you picked the pace numbers from NBA.com or even the wrong adjusted values right away.

For championship winners I have -2.2 and for the Losers -1.5. I have 11.2 for the championship winner with the best player on the court, and 8.3 for the runner-up. The championship winner had in average +7.4, the loser +5.8, thus in average the better team won. For the championship winner, the minutes without the star player were 28% of the overall minutes, for the runner-up it was 26%.

The more interesting thing for me is that the correlation between off value and minutes is -0.1 for both, which means, the more minutes the star player missed, the worse got the team without him on the court in that sample. Overall that suggest that the results can't be dismissed with "garbage time".

Not sure where you are getting your numbers from. Here are my theories on why your's are different.

1. You are using different star players than me. I used Ben Wallace, Allan Houston, and Reggie Miller as the stars of their teams.

2. You are weighing each year differently depending on minutes off court. I weighed everyone equally

3. Your possession numbers are wrong because nba.com has a glitch for the 97 season. They overestimated everyone's minutes by like 200 minutes. That will cause you to have a different possession number.

Here is what I had:

winners
on court, off, plus/minus, percent of minutes played
lebron 10.4 -3.6 14 73
dirk 10.6 -5.5 16.1 64
kobe 8.6 -3.4 12.4 71
kobe 11 -0.1 11.1 75
kg 16.4 4.6 11.8 59
duncan 13.7 -0.7 14.5 69
wade 8.3 -6.9 15.2 73
duncan 16.9 -0.8 17.7 56
wallace 7.5 3.2 4.3 77
duncan 9.1 -5.7 14.7 80
shaq 11.3 1.4 9.9 61
shaq 8 -7.6 15.6 73
shaq 11.2 -0.3 11.5 80
duncan 10 3.7 6.3 81
jordan 9.8 -1 10.8 80
jordan 13.6 4.5 9.1 79
avg 11.0 -1.1 12.2 71.9

finals losers:

durant 11.3 5.6 5.7 80
lebron 10.6 1.6 9 77
garnett 7.7 0 7.7 52
howard 10.2 1.9 8.3 72
kobe 9 2 7 81
lebron 6.2 -2.6 8.8 80
dirk 8.4 0.2 8.2 78
wallace 7.2 -0.7 7.9 67
shaq 8.3 -3.3 11.6 62
kidd 8.5 -2.5 11 76
kidd 5.6 -0.1 5.7 77
iverson 5.7 1.6 4.1 75
miller 4.6 5.5 -0.9 76
Houston 1.7 -0.6 2.3 75
malone 10 -2.8 12.8 77
malone 13.2 -2.7 15.9 76
avg 8.0 0.2 7.8 73.8

[mysticbb]

Not sure where you are getting your numbers from. Here are my theories on why your's are different.

I use the raw +/- numbers from NBA.com, and then use an estimate for the possession based on the differences between NBA.com and bbr.com between 2001 and 2013. The NBA.com possession numbers are WRONG! They are counting at least possessions at the end of quarters which are not possessions. And even then, the possession count by NBA.com does not equal the real amount for possessions when counting the possession for games manually.

1. You are using different star players than me. I used Ben Wallace, Allan Houston, and Reggie Miller as the stars of their teams.

With the exception of Houston, who was NOT the best player on the Knicks, I have the same players.

2. You are weighing each year differently depending on minutes off court.

I indeed use minute-weighted averages, which is the way to go.

I weighed everyone equally

Which is wrong!

3. Your possession numbers are wrong because nba.com has a glitch for the 97 season. They overestimated everyone's minutes by like 200 minutes. That will cause you to have a different possession number.

My possession numbers should be more accurate than those of NBA.com even though it is just an estimate.

Here is what I had:

As expected, the numbers aren't accurate. I have slightly different numbers. When I go backwards, I get the correct Net values for the overall teams. With your numbers they are off.

[Texas Chuck]

does this mean people are going to stop making excuses of KG in Minny? Where its now obvious finally to the data boys that even good supporting casts dont have nearly the impact the superstar player has. Something I have been screaming from the mountaintop in KG v Duncan and KG v Dirk threads for a long time. Or are we still going to have to hear that somehow the Mavs supporting cast was worth 25 more wins than the Wolves supporting cast?

[colts18]

I use the raw +/- numbers from NBA.com, and then use an estimate for the possession based on the differences between NBA.com and bbr.com between 2001 and 2013. The NBA.com possession numbers are WRONG! They are counting at least possessions at the end of quarters which are not possessions. And even then, the possession count by NBA.com does not equal the real amount for possessions when counting the possession for games manually.

I used B-R numbers from 01-12 since they use accurate possession numbers. For 97-00, I used nba.com
s pace number to estimate the number of possessions.

With the exception of Houston, who was NOT the best player on the Knicks, I have the same players.

the 99 knicks were a tough team to pick out since Ewing was injured. Camby had the best advanced numbers but no one would say he was the star of the team. Larry Johnson was too old. Sprewell was a good candidate but no one can be the star of the team if they start only 4 games. Houston made the all-star games in 00 and 01 so he was the best remaining choice in my eyes.

I indeed use minute-weighted averages, which is the way to go.

Which is wrong!

I don't see the need in using minute-weighted. I'm looking at the average cast. The 12 heat and 11 Mavs should count equally since they have 1 title. No reason to give the 11 Mavs extra bonus since Dirk missed more time.

[The Infamous1]

It's usually not how good it is (though that matters too), it's how good it is relative to their opposition at the time. That's why the 94 Rockets (who were not a historically weak support cast per se), are regarded as right there with the 03 Spurs, the 76 Nets and the 77 Blazers- because even though the support casts were worse for the latter 3, Hakeem had (generally) tougher opponents.

The rockets are a historically weak cast any way you slice it.

And the 03' Spurs while weak, were facing no significant disadvantage talent wise compared to the rest of the playoff Competetion which was arguably the weakest year in history. Cwebb injury, nets being flat out one of the worst finals teams ever, Kobe injured while the lakers as a whole werent that good outside of their top 2, Dirk injury, and the marbury led suns

[lorak]

Not sure where you are getting your numbers from. Here are my theories on why your's are different.

I use the raw +/- numbers from NBA.com, and then use an estimate for the possession based on the differences between NBA.com and bbr.com between 2001 and 2013. The NBA.com possession numbers are WRONG! They are counting at least possessions at the end of quarters which are not possessions. And even then, the possession count by NBA.com does not equal the real amount for possessions when counting the possession for games manually.

Did you contact with them about that issue?

[colts18]

Did you contact with them about that issue?

nba.com didn't reply back to me, but Jon Schumann did a few weeks ago

great question.

NetRtg is the difference between OffRtg and DefRtg. OffRtg is calculated using only offensive stats to calculate the # of offensive possessions, and DefRtg is calculated using opponent stats to calculate the # of defensive possessions.

I believe +/- per 100 possessions uses an average of offensive and defensive possessions.

Through Monday, OKC had an estimated 5,414 offensive possessions and 5,476 defensive possessions. That discrepancy accounts for the difference between their NetRtg and their +/- per 100 possessions.

Hope that makes sense. Thanks for writing.

[Laimbeer]

It's impossible for a championship team to have a terrible supportin cast. Each championship team has at least a decnt supporting cast. The worst one i can think of is the 2012 heat. Their supporting cast was terrible but they won because they played no title contenders in the east.

Wade and Bosh?

[Laimbeer]

does this mean people are going to stop making excuses of KG in Minny? Where its now obvious finally to the data boys that even good supporting casts dont have nearly the impact the superstar player has. Something I have been screaming from the mountaintop in KG v Duncan and KG v Dirk threads for a long time. Or are we still going to have to hear that somehow the Mavs supporting cast was worth 25 more wins than the Wolves supporting cast?

It would be interesting for someone to go back to, say, the merger and determine which years the better star player won as well as which years the better supporting cast won. I'm too lazy.

[mysticbb]

does this mean people are going to stop making excuses of KG in Minny?

What? Timberwolves had a worse than average supporting cast. What do you expect?

2001: -5.5
2002: -4.3
2003: -17.5
2004: -10.9
2005: 1.0 (greatly related to McHale's approach of balance, who had Garnett play with weaker players and put the better supporting players in lineups without Garnett, the same he is doing right now in Houston, hardly something you can do during a championship run)
2006: -10.2
2007: -15.3

So, seriously, what kind of success do you expect?

Or are we still going to have to hear that somehow the Mavs supporting cast was worth 25 more wins than the Wolves supporting cast?

They were! Or do you honestly believe that the Mavericks in 2007 won 67 games and the Timberwolves won just 32, because Nowitzki made the difference over Garnett?

[mysticbb]

I used B-R numbers from 01-12 since they use accurate possession numbers.

Then you better check some of those numbers, because they are wrong. Take Durant in 2012 (usually the year in which the season ends is used, not the years in which the season started, because the majority of the season actually happens in the year the season ends) for example: He has +7.1 On and +5.2 Off, you have something different.

the 99 knicks were a tough team to pick out since Ewing was injured. Camby had the best advanced numbers but no one would say he was the star of the team. Larry Johnson was too old. Sprewell was a good candidate but no one can be the star of the team if they start only 4 games. Houston made the all-star games in 00 and 01 so he was the best remaining choice in my eyes.

But he wasn't the best nor the most important player ...

I don't see the need in using minute-weighted.

To avoid a bias. All my numbers are minute-weighted.

The 12 heat and 11 Mavs should count equally since they have 1 title.

But we don't have any evidence that the Heat would play the same without James as they did, if he would have missed the same amount of time. You are implying that with your calculation, I take care of that bias with the minute-weighted approach. Because, despite the fact that not all minutes are garbage minutes, the relative amount of garbage minutes is increasing, if the player is missing less games/minutes. So, in order to counter that, we take minute-weighted averages and get a more accurate picture.

[mysticbb]

Did you contact with them about that issue?

No. I contacted NBA.com and ESPN.com multiple times in the past about errors, I never got an answer back. Especially Schumann, he is no better than an average Joe in terms of handling stats. So, let them figure that out for themselves. Why should I do their job?