FanPost

Salaries and wins

Over at The Baseball Analysts there is an article titled "Can Money Buy Love in Baseball?"
By David J. Berri. He is co-author of a book called Berri "The Wages of Wins: Taking Measure of the Many Myths in Modern Sports"

He seems to be saying that salaries don't explain winning very well. I would love to hear from anyone who has read the book. I am skeptical of this. I have not looked at this recently, but below are some comments I posted on the SABR list several years ago and I think sometimes salaries are highly correlated with wins. It may have been different in recent times.

The first post was from 1997. Here it is:

In the AL, the correlation between a team's average winning percentage from 1994-96 and its average salary was .651. In the NL it was .608.

I also did some rank correlations.  The AL had .719 and the NL had .49.  The NL had an extreme outlier, the Expos, who had the 3rd highest winning percentage over the three years but were last in average player salary.  The NL's rank correlation with the Expos removed was .72.  All correlations were significant.

This next post was from 1998

There has been some discussion lately on the relationship between salaries and wins.

Below are the rank correlations between salaries and wins for each league, starting in 1977 and going through 1997.  The two variables being correlated are a team's rank in wins for the year and their rank for salaries paid for the year in their respective league.

AL: .762, .756, .741, .301, .598, .388, .141, .570, .529, .311, .371, .279, .044, .128, .174, .718, .570, .475, .751, .596

NL: .760, .916, .573, .650, .056, .329, .378, .084, -.238, -.364, -.021, .417, .169, .228,
-.025, .301, .486, .777, .286, .364

The year 1981 is not included in this series.  What seems to be happening generally is that the correlations were high in the beginning of the free agency period, then fell, then became higher. So outspending the other teams in your league sometimes will pay off, sometimes not.
Those negative correlations are correct for the NL!

Of course, player performances can vary quite a bit and are hard to predict from year to year, but are less so in the long run.  I also
correlated a team's average rank over  three year periods in wins and salaries for each league.  For example, if a team was 1st, 2nd and 3rd in wins, their average rank was 2.

The first 3 year period was 77-79, then 80-83(leaving out 1981).  The last period is 96-97 only

AL: .797, .566, .755 .578, .152, .717, .838

NL: .837, .479, .090, .210, .224, .719, .407

There is still quite a bit of fluctuation here.  The highest spending teams have not always won. The correlations seem to be higher in the AL.
I would like to hear some good explanations on these two points.

Finally, I correlated the average rank for wins and salaries for the entire period.  In the AL it was .860.  That seems high and seems to
indicate, that in the long run, the higher spending teams do better. I broke down the NL into 77-92 and 93-97 (expansion to 14 teams).  The correlations were.402 and .564.  Again, the numbers were higher in the AL.

Also, in a regression, the R-squared would be .7396 in the AL, meaning that about 74% of the variation in team rank in wins is determined by variation in team rank in salaries.

Trending Discussions