How Much is That Replacement Player?by Dave Studeman
December 01, 2006
The concept of a replacement-level player is an important one for those who really want to understand baseball stats. Without a replacement level, you're stuck with stats that are either absolute in nature (40 home runs), rate stats (eight home runs per 100 at bats) or compared to average (10 home runs above average).
The trouble with absolute stats is that they don't include a concept of playing time; 40 home runs in 400 at-bats is more impressive than 40 home runs in 600 at-bats. The trouble with rate stats is that it's easier to have a high rate in a limited amount of playing time; eight home runs per 100 at-bats is more impressive for someone with 600 at-bats than someone with 400 at-bats. The same issue applies somewhat to averages (10 home runs above average, based on the player's number of at bats). Plus, half the players fall below average so there's no solid sense of a lower baseline. No zero, in essence.
A replacement-level stat (such as 30 home runs above replacement) addresses all of these issues. It applies a (hopefully) appropriate baseline that only a small minority of players fall below and it reflects playing time. It's insightful, intuitive and easy to understand.
Lots of baseball analysts smarter than me have written about replacement levels: where they should be set and why. I'm not going to get into that today. Instead, inspired by this discussion, I'd like to ask how much a replacement level player costs on the free agent market.
Let's step back in time, to an article I wrote almost two years ago, which included this graph of how much teams paid free agents and how much those players produced, as measured by Win Shares Above Replacement. The data comprises the 2004 season.
As you can see, players who were paid less than $2 million contributed one Win Share Above Replacement, on average. So in 2004, teams paid something less than $2 million for an average replacement-level player in the free agent market.
We can do better than that, however. For the 2005 season, I monkeyed around with the replacement-level math and renamed the stat Win Shares Above Bench. I think it works a bit better than the 2004 WSAR. I defined replacement level players as 70% of average (said differently, a .350 winning percentage or a 57-win team).
Here's a table of free agents who played in 2005 grouped into $1 million salary ranges up to $5 million, and the average number of Win Shares Above Bench contributed by those players. For this table, I didn't include any pitchers because they're so dang unpredictable—the entire free agent pitcher market is a topic for another article.
Salary Range # of FA's Avg. WSAB $0 to $1M 82 0.5 $1M to $2M 21 1.5 $2M to $3M 16 4.0 $3M to $4M 19 3.4 $4M to $5M 12 5.6This table answers the question more precisely; in 2005, free agents who were paid less than $1 million were only slightly better than bench players, on average, and there was a fairly normal rise in productivity for higher-paid players. Based on this table and the previous graph, I think we can safely say that $1 million is about the going rate for a free agent replacement player. At least it was, heading into 2006.
Here's what the same data looks like for the 2006 season:
Salary Range # of FA's Avg. WSAB $0 to $1M 58 0.6 $1M to $2M 23 0.4 $2M to $3M 12 1.1 $3M to $4M 14 2.0 $4M to $5M 14 3.0Compare the two tables, which represent one year's difference in salary and performance. What a difference it is. The least expensive free agents maintained the same level of performance, but the players paid $1 million to $2 million dropped by more than a Win Share. The $2-$3 million group dropped by nearly three Win Shares. The $3-$4 million group lost almost one-and-a-half Win Shares and the $4-$5 million group lost two-and-a-half Win Shares. On average, the cost of a replacement-level player last year was closer to $2 million than $1 million. What happened?
Young guys happened; that's my theory. As I've mentioned before, the last two years have seen a great influx of young talent into the major leagues. Ryan Howard, Joe Mauer, Justin Morneau, and the entire Florida Marlins roster made peanuts last year compared to free agents. There are a set number of Win Shares to go around, so their great performances decreased the contributions made by free agents.
What surprises me is that the impact of these youngsters was felt primarily among the lower strata of free agents. Here's a side-by-side list of 2005 and 2006 average WSAB by salary class, for free agent position players:
Salary Range 2005 2006 Diff $0 to $1M 0.5 0.6 0.1 $1M to $2M 1.5 0.4 -1.0 $2M to $3M 4.0 1.1 -2.8 $3M to $4M 3.4 2.0 -1.4 $4M to $5M 5.6 3.0 -2.6 $5M to $7M 6.1 6.7 0.6 $7M to $9M 6.6 3.9 -2.7 Over $9M 7.6 7.7 0.1As you can see, the performance of the true superstars—players above the $9 million mark—was the same in 2005 and 2006. It was the lower-priced groups that really dropped in performance. Something strange happened in the $7M to $9M class too. Their names were Darin Erstad, Javy Lopez and, especially, Edgardo Alfonzo.
But the important drop occurred in every salary class from $1 million to $5 million. I'm talking about contracts with players like Abraham Nunez, Royce Clayton, Rondell White, Travis Lee, Vinny Castilla and Carl Everett. These were all bad deals for their teams, and most of them were predictably bad. It seems as though major league teams might have gotten a bit lax with some of their lower-scale contracts in 2006.
So, this offseason, while we're busy ogling the contracts of Alfonso Soriano and (gulp) Juan Pierre, keep an eye on the deals made for less than $5 million a year.
References and Resources
Here's a link to the original article I wrote explaining my overall approach to comparing Win Shares and salary. Although some of the details have changed, the general approach hasn't.
Dave was called a "national treasure" by Rob Neyer. Seriously. Comments about this article can be sent to him through the miracle of e-mail.