No correlation between salary and wins?

[xkonk]

The poker analogy is a poor one. In the 'unregulated' game you describe (is that even the word for it?), the guy with the most money wins every pot because the other players are forced to match his bet to keep playing, which they can't. This is a game that would never start because the richest player would simply go all-in pre-flop and win the blinds every hand. Even if another player had the absolute nuts he could not win. In basketball, the teams still play the games. If Michael Jordan went crazy and decided to give every player on his team a raise so that he was spending more than the Mavs or Lakers, they would still have to play the game and the Mavs/Lakers would be the winner most of the time.

Your description of a league with no cap sounds like baseball to me. The connection between salary and winning is typically low in MLB as it is in the NBA. This year, either the 10th, 11th, 13th, or 17th highest-spending team will win the championship. Last year's Giants were 9th. Of the last 20 teams to play in the World Series, only 5 (the Yankees and Red Sox) have been in the top 5 in salary (after this World Series, it will obviously move to 5 out of 22). So it seems empirically false that teams willing to spend more will acquire all the best players and just run away with things. It's true that the Yankees typically do well in the regular season, but they would appear to be an outlier rather than the rule.

[Mike G]

Well, the last time I played poker -- and it was around 30 years ago -- nobody knew how much money anyone else had. That was part of the bluffing. And if you bet money you don't have, you better be sure you are going to win.

Basketball playoffs are more likely than baseball to advance the better team to the next round. Rarely does an upstart get to the conference finals, much less the Finals.

A team with very good players who are underpaid soon becomes a team whose players are looking elsewhere for more money. Success is self-limiting that way. Unless you have a superstar (or 2) willing to play for less than market value, you don't get to have a dynasty.

I don't follow baseball, but I'm going to theorize that much of the Yankees' success has to do with their buying up good players. And probably paying some more than they're worth. They still may not be getting maximum wins per million dollars, but they tend to win.

If you remove the 10 worst-spending teams and the 10 best from the last 10 NBA seasons -- avg 1 of each per year -- I'm guessing the correlations are better. Due to the mega-bad money managing GM's that have done so badly. It's literally impossible for any GM to offset an Isiah Thomas Knicks season.

[xkonk]

If you ignore the Knicks, the correlation from 2002 to 2010 jumps from .256 to .377. Still not a stunning number.

A better question for this thread might be, is there any evidence (besides the anecdotal/single-team "look at the Yankees!/look at the Lakers!") that teams can pay their way to consistent victory? Obviously they can to some extent; the correlation is positive. But it's also small. Can anyone convince the Zimbalists of the world that it's such a strong, common occurrence that it's in everyone's best interest to have a salary cap? The NHL lockout is the best example I can think of, but before that they had literally no restrictions on spending and player salaries were a very large proportion of revenue, so the analogy to other sports is somewhat limited. I also have no idea what the correlation between salary and winning was in the NHL pre-lockout or how much it's changed since then.

[Bobbofitos]

The poker analogy is a poor one. In the 'unregulated' game you describe (is that even the word for it?), the guy with the most money wins every pot because the other players are forced to match his bet to keep playing, which they can't. This is a game that would never start because the richest player would simply go all-in pre-flop and win the blinds every hand. Even if another player had the absolute nuts he could not win. In basketball, the teams still play the games. If Michael Jordan went crazy and decided to give every player on his team a raise so that he was spending more than the Mavs or Lakers, they would still have to play the game and the Mavs/Lakers would be the winner most of the time.

Your description of a league with no cap sounds like baseball to me. The connection between salary and winning is typically low in MLB as it is in the NBA. This year, either the 10th, 11th, 13th, or 17th highest-spending team will win the championship. Last year's Giants were 9th. Of the last 20 teams to play in the World Series, only 5 (the Yankees and Red Sox) have been in the top 5 in salary (after this World Series, it will obviously move to 5 out of 22). So it seems empirically false that teams willing to spend more will acquire all the best players and just run away with things. It's true that the Yankees typically do well in the regular season, but they would appear to be an outlier rather than the rule.

I think this is more a product of baseball's playoff system, which is a total variance/joke fest.

[xkonk]

Baseball's playoff results might be mostly due to luck, but the low correlation in the regular season is presumably not. A number of people are more concerned with rich teams buying their way to championships, not regular season victories, so I thought the World Series point would be useful. From 2002 to 2010, looking at relative payroll (relative to that season's salary cap), the top 20 NBA spenders consist of the Knicks (8 times, no championships), the Trail Blazers (4 times, 02-05, no championships), the Mavs (4 times, 04-05 and 08-09, no championships), the Spurs (06, no championship), the Suns (07, no championship), the Raptors (09, no championship), and the Timberwolves (04, no championship). If you notice a theme, it's that spending money doesn't guarantee you a title. I'm sure that people will point out that the Knicks were horribly mismanaged and shouldn't count, but they're still part of the pattern: spending money doesn't guarantee anything. Take them out and the top 12 spenders still don't have a title. Teams have to spend wisely, and if fans or owners are angry that their team doesn't have a title, maybe they should look at the guys who picked the talent on their team.

[Mike G]

From 2002 to 2010, looking at relative payroll (relative to that season's salary cap), the top 20 NBA spenders consist of the Knicks (8 times, no championships), the Trail Blazers (4 times, 02-05, no championships), the Mavs (4 times, 04-05 and 08-09, no championships), the Spurs (06, no championship), the Suns (07, no championship), the Raptors (09, no championship), and the Timberwolves (04, no championship). If you notice a theme, it's that spending money doesn't guarantee you a title. ...

Holy cow. Is that what they call a 'straw man'? Nothing will "guarantee you a title".
Whatever the reason that 2002 is picked as the starting point -- salaries before then are unknown? the 2000 and 01 Lakers were one of the highest spending teams? -- we know the Spurs won twice and the Mavs once in this 9-year span.
The Blazers were competitive in that time, and the Suns later; the Wolves that one year. (Their fans will tell you they "should have won"). Three of 20 named high-enders won titles, compared to just 1/30 of all teams winning titles in any interval. The 20 highest spenders did well, even with the Knicks going 0-8.

I was already wondering why the Yankees' spending-success was limited to:

Of the last 20 teams to play in the World Series, only 5 (the Yankees and Red Sox) have been in the top 5 in salary

-- which cuts off the Yanks' titles in 1998 and 2000; and who knows what else?

Can't we get an unbiased report in here?

[xkonk]

My apologies; I thought this thread existed because people (by which I mean whoever thinks that a salary cap is necessary) are concerned about 'rich' teams buying wins and titles that the 'poor' teams can't/won't pay out to get. The correlation between regular season wins and salary has already been discussed (and I thought dismissed long ago, although apparently I'm wrong since it keeps coming up), so I thought I would move on to titles. The basketball salary data I have covers 2002-2010, so that's why those are the numbers I present. I'm not much of a MLB guy, so I found a website and covered about as much ground as I cared to.

The teams you picked out from my list have done well, but they never won in the year they spent a lot. Isn't that the concern? Sure, the Mavs and Lakers spend a lot every year, but shouldn't we also worry about an owner saying "I'm taking my shot" and buying as much as he can? The Cavs did something like that once they got LeBron, but no one cared because the Cavs managed to lose in the playoffs despite having a top-tier team. The Celtics did that when they made their run at Garnett and Allen, but they did manage to win (albeit still without spending as much as some other teams).

Apologies again Mike, for not having perhaps 20 or 30 years of salary data on-hand. Perhaps I've heard too many of the owners' arguments and thought the effect should be so strong and obvious that it would appear in the last 10? Maybe if you'd like to contribute we'd get an unbiased report? Or maybe you could just reframe the salary cap argument into something that isn't a straw man so we can test theories you would find more reasonable?

[Crow]

There is a difference between being the absolute highest spender and "spending a lot".

As I stated earlier in this thread 8 of the 9 teams who spent over $400 million in the last 6 seasons got at least one trip to the conference finals and they got 15 of the 24 such appearances.

As I stated at Real GM, all the champs in the last 6 seasons spent in the top 30% on spending that season.

Upon further research if you take it back another 17 years only 9 of those champs were top 30% on spending. But that was under different CBAs with different rules and a different cap system or no cap and a different labor market and probably fair to say a different perspective on player spending given a different revenue environment and a different level of average owner wealth compared to today.

The last 6 seasons were all under the most recent CBA (making it a good period designation) and the pattern for conference finalists and champs and spending in the top 30% is pretty strong and perfect respectively.

Swinging from 9 in 17 champs being in the top 30% on spending to 6 of 6 is a pretty strong swing. It could random variation or it could mark a distinct change in the competitive playing field. Or some of both.

New CBA, at least somewhat different rules and climate. We'll see what the level of success of the top 30% of spenders is compared to the rest of teams and compared to the top 30% in those previous 2 very different time periods. It probably won't be 6 of the next 6 champs in the top 30% but I'd be surprised if it less than 4 of 6, when the expected share of titles for top 30% spenders over that timeframe (all else equal) would be 2.4.

[Crow]

Another cut of analysis. in the top part of the article:

http://nba-point-forward.si.com/2011/10 ... nce-issue/

[Bobbofitos]

Another cut of analysis. in the top part of the article:

http://nba-point-forward.si.com/2011/10 ... nce-issue/

Really good article

[Mike G]

Another cut of analysis. in the top part of the article:

http://nba-point-forward.si.com/2011/10 ... nce-issue/

“The statistical correlation between payroll and win percentage is practically nonexistent.”

Whoa. Nonexistent? Zimbalist tells me he was referring to the correlation between spending and winning within a single NBA season. That’s an important qualifier, ...But we have to take the longer view, too, and see if money helps teams remain consistent winners over long periods of time. That’s where things get complicated.

What is stopping anyone from simply summing team payrolls for 5-yr, 10-yr, 15-yr intervals, and finding the correlation with team wins/playoff series wins/titles in those intervals?

[Crow]

xkonk: "If you ignore the Knicks, the correlation from 2002 to 2010 jumps from .256 to .377. Still not a stunning number. "

Hartford Whalers at RealGM found a .258 from 2000 to 2010.

For the last 6 seasons under the current CBA I found the correlation to be .35.

The correlation is suppressed by the distorting effects of the rookie contract scale and other elements of the CBA (exceptions and maxs and minimums).

Remove just the Spurs even in the tail end of their good fortune on value from stars and the correlation goes to .4.

Remove the top 6 spenders (overspenders?) and it also goes to .4. Remove the bottom 6 spenders as well and it also goes to .44.

At Real GM it estimated that if a team spend 25% more than an average team and would be expected to go from a 41 win team to about 46 win team. Because of this I'd think the impact of salaries on playoff appearances and seeding and playoff performance will be greater than on regular wins.

The yearly correlation between wins & spending the last 3 seasons has been among the 4 highest of the last 13 seasons. The correlation has bounced around a lot. Several owners successfully spent a lot to enhance their chance to win a title. As mentioned several times, 8 of the 9 teams who spent over $400 million in the last 6 seasons got at least one trip to the conference finals and they got 15 of the 24 such appearances.

Correlation of total number of playoff games played by teams in the last 6 years and salary surprisingly is still only .35 in the last seasons (I guess it doesn't take or mean a lot to be a first round loser). Correlation with number of playoff series wins is only .38. But weight 2nd round series wins as worth 3 times as much as first round wins and then just double it the rest of the way (a weight of 6 for 3 wins and being runner-up, 12 for 4 series wins and being champ) and the correlation between weighted playoff series win "value" rises to .45.
That seems to be about where the correlation tops out. Only value Finals appearances and the correlation is .42. Give winning the title 3 times the weight of being runner-up and it only goes to .45.

Relatively speaking a pretty low correlation but it still probably should not be entirely minimized or dismissed either.

The marginal win efficiency (and playoff series and deeper success) of an extra dollar probably goes down as the total spending rises, starting out higher than the average efficiency then fading. Those values could be computed with more time and interest.

On simple average, the 9 most successful teams in terms of playoff series won spent an average of 8% more during the last 6 seasons than the average team. It seems like a fairly good investment (if you are in this game for glory and / or the heightened revenue that comes with success) but other factors are bigger than salary- talent, coaching and match-ups. Remove Mark Cuban's Mavs and the other 8 top performing playoff teams only spent 5% more. These 9 teams averaged 1.3 playoff series wins per year. No other team made the conference finals more than once and none made the Finals without being top 30% of spending over the recent 6 year period. That seems like a pretty sharp separation. (But as noted before it wasn't that way in the deeper past. And might or might not stay that way under a new CBA.)

[Mike G]

Alright, I got team payrolls from 1998 to 2011 (14 seasons) from this page: http://www.eskimo.com/~pbender/
Then hand-entered all team wins, found each year's average team payroll, and converted each payroll to multiples of that year's average, then to equivalent 2011 (avg = 67.124) million dollars.

This entails 413 team-seasons. It turns out, 413 = 59*7, so when 413 teams are ranked by their standardized payroll, broken into equal 7ths, their wins break out like this:

Code: Select all

2011$   %avg   W    W/mil
93.9   1.40   48.1   .53
74.0   1.10   45.1   .61
68.3   1.02   43.8   .64
65.0    .97   38.0   .58
61.9    .92   41.2   .66
57.3    .85   37.6   .66
49.4    .74   33.3   .67

Note that the highest-spending 1/7 of teams paid out twice as much, on average, as did the lowest-spending 1/7. And on average, they got not quite 50% more wins.
So the last column, Wins per million dollars, shows a diminishing return of sorts.
Some owners would rather pay twice as much for a 48-34 team, and others may be content to put out half as much for a 34-48 team.

It's certainly not a proportional relationship, but it may be a linear relationship. The closest such fit I can find between payroll (P, in millions) and wins is:
W = .372*P + 17.112

This is also equivalent to:
W = (46+P)*.372

The average discrepancy between wins and expected wins from this formula is 9.90

The 2006 Knicks paid out 134.8 mill (in 2011 equivalent dollars), and thus 'should have' won 67 games. They won but 23, thus missing the mark by 44 wins.
Knicks teams from 2002 thru '08 have 5 of the 6 most underachieving teams relative to their payrolls. (Miami '08 has the other.)

The 2005 Suns, meanwhile, have the best record relative to payroll, paying just 75% of that year's average salaries and winning 62 games. Such an outlay would expect to return just 36 wins.

Just 5 of the 25 biggest variances between expected and achieved wins are positive. Proving that it's much easier to mess up big-time than it is to achieve some coup d'etat on the rest of the league. This year's Bulls look 2nd-best, followed by the '06 Pistons, the '98 and '99 Jazz.

[Crow]

That is a very good way to look at it.

As I mentioned before, the league-wide correlation of salary vs performance will be heavily affected by the uneven / not strong relationship between the two amongst average or low spending teams. The pay vs performance trend amongst top spending teams looks pretty strong. They can of course use the previous year's winning% as a gauge of whether the extra spending is "worth it".

Might be interesting to look at average change in year to year spending trends amongst teams after first breaking the 40, 45, 50 and 55 win barriers moving upwards and 50, 45, 40, 35 moving downwards.

This is different from the suggestion just made (not year to year for the same team) but moving from the 7th tier to the 6th, the additional average wins cost $1.8 million each. It actually cost less to move up to the 5th, 4th and 3rd tiers. Only the second and first tiers are pricey to move up to on average, costing $4.4 million per win to move from the average of 43.8 associated with 3rd tier spending to the second tier and the average of 45.1. Sometimes that might make th e difference of being in the playoffs or not (a distinction with fairly significant real monetary value I'd think, making it generally a good investment). To move from the 2nd tier of spending / performance to the first, a extra win costs $6.6 million on average. Owners will definitely vary in how much they are willing to spend and how much the prospect of more winning would be worth to them to make that extra financial effort.


If the new CBA has a pretty hard cap structure, it will make extra spending more difficult / expensive and probably reduce the occurrences on spending 20+% above the average. Teams will need to work the remaining tools for edge harder- coaching , analysis, vets working below fair market, the draft on its own and tanking for it (probably more often for multiple seasons to maximize the impact).

[Mike G]

As reported above, the average discrepancy between wins and wins expected from team payroll is 9.90 in a team-season.
Taken as 3-year intervals, 5, 7, 9, 11, and 13-years, the average annual discrepancy progressively lessens.

Code: Select all

1yr   3yr   5yr   7yr   9yr   11yr  13yr
9.90  8.22  6.95  6.29  5.83  5.50  5.32

The normal difference between wins and pythagorean expected wins is probably 2 or 3 wins per season, and these intervals approach just about twice that. Probably a couple of extreme outliers, a couple of extra wisely- and foolishly-spending teams, make up most of the difference.

The Spurs own the 8 best 7-year intervals in the last 14 years. In fact, all teams (except the Bobcats) have 8 7-year intervals in that span, and no one can match the Spurs in wins vs. payroll. They paid like a 40-win team and averaged 59 wins, 2001-07 and 2002-08.

The Pistons have the next 2, and 4 of the next 7; their max was 2002-08, when their payroll would suggest a 40 win team, but they averaged 55.

The Knicks, of course, dominate the lower end, snapping up the bottom 8 spots; their worst-of-the-worst 7 years being 2002-08.
The next worst 7-year underachievers (overspenders) would be the 2005-11 T-Wolves; averaging just 27 wins while spending the league average on salaries.