Thursday, October 1, 2009

FanGraphs Salary Values

When trying to quantify a player's contributions to their team, sometimes Matt and I link to FanGraphs because in addition to Runs Above Replacement and Wins Above Replacement, they also translate that player's value into salary dollars. According to their calculations, here are the 15 most notable Yankees and their respective values:
Seems pretty high, doesn't it? That's not even all of them. There are part-time contributors like Chien Ming Wang (yes, he actually had positive value) and Hinske and Hairston and Bruney and Cervelli and Pena that add to that number incrementally as well.

The Yankees' Opening Day payroll was $201.5M, which is roughly what the top 10 most valuable guys on the team add up to (Damon and above on that list). Are the Yanks actually getting that much more than their money's worth?

Well, that depends on your view point.

As David Pinto of Baseball Musings points out, FanGraphs calculates player value based the value of marginal wins, and thereby attempts to valuate all players as if they were free agents. So, when you add up the value of all the batters and pitchers on FG, it comes out to $4.6B, whereas the total payroll of the MLB is roughly $2.7B.

With the obvious disclaimer that the folks behind FanGraphs are much smarter than I am, I would like to respectfully disagree with this methodology.

They use a system that corrects for the artificial forces depressing the salaries of players who are not available to the free market, which makes sense in it's own right. But we are all familiar with these artificial constraints and understand that is the reason why guys like Tim Lincecum are paid a fraction of what they are actually worth.

Instead of creating a system where the value of players is always going to far exceed the payroll, why not base it in reality? When I look at that dollar figure on FanGraphs, I want to know how much a player was actually worth in relation to what other players throughout the MLB are getting paid. I want to be able to tell who is getting their fair share or the pie. Part of that is the fact that guys like Phil Hughes are able to contribute at far beyond their pay grade but someone like CC Sabathia is unlikely to be worth the checks he's cashing, even during a very productive season.

I want to look at salary on a scale that is familiar to me, not one that is based on a contrived scenario in which everyone is a free agent and would make far more than they really do or even would make under those circumstances. It's not like the owners would suddenly shell out an extra $2.1B dollars if everyone hit the market over the next offseason.

Here is that list above, based on the MLB's actual payroll:
  • Derek Jeter -$19.5M
  • CC Sabathia - $16.3M
  • Mark Teixeria - $14.2
  • Robinson Cano - $11.3M
  • A-Rod - $11.2M
  • Jorge Posada - $10.6M
  • Nick Swisher - $10M
  • Andy Pettitte - $9.4
  • A.J. Burnett - $ 8.3
  • Johnny Damon - $7.1M
  • Hideki Matsui - $6.5M
  • Phil Hughes - $6M
  • Mariano Rivera - $5.2M
  • Brett Gardner - $4.9M
  • Joba Chamberlain $4.2M
  • Melky Cabrera $3.9M
  • Alfredo Aceves $3.5M
  • Phil Coke - $59K

  • Total: $152.2M
Seems like a better approximation of their values. At least to me it does.


  1. Almost seems like Win Shares might work well here since they don't have a negative number, and any player with zero Win Shares could be equated with the league minimum.

    I'm not certain how you did your calculation, but it almost seems like someone could take the total MLB payroll and divide it by the total of MLB player Win Shares. This would yield a dollar value per Win Share in a given season, and could then be used to determine player value.

    So, if a Win Share is worth $1MM, a player with a 1 Win Share season would be worth $1.4MM — the value of a Win Share plus the league minimum. I have no idea if that would statistically hold up, but unlike the odd Fan Graphs valuations, this seems more closely tied to reality.

  2. FanGraphs did their calculation based on the assumption that a replacement level team would win about 30% of its games and therefore there are about 1,000 marginal wins available to divide between the 30 teams. Divide the payroll ($2.7B) by those 1,000 wins and they are worth $2.7M each.

    The concept of replacement level is the thing that really makes this difficult. My calculation was simple, just re-scaling the salaries based on $2.7B instead of $4.6M (multiplying by .5869).

    I think the problem with your approach might be adding replacement value on top of every player is going to give you a pretty big surplus at the end (700 or so players + $400,000K) would be an extra $280M. Not positive, but that's what I think at first glance.

  3. My math didn't exist. I simply had an off-the-cuff framework for another approach. Personally, I don't like the FanGraphs approach the more I learn about it.

    Just like your point that it seems odd to base player valuations on a theoretical salary structure, I don't like using the replacement level player idea much since, again, we're dealing with the theoretical.

    Extend that idea to a team — like they did by saying a replacement-level team would only win 30% of its games — and we're talking about something which barely ever happens. Off the top of my head, only the '62 Mets and '03 Tigers were at that threshold of ineptitude since the 162-game schedule was implemented. The more theoretical it becomes, the less applicable to the actual business of MLB it appears to be.

    I've got a loopy schedule for a few days, but if I get a chance to play with the numbers some, I'll see if your criticism of my idea is accurate. Yes or no, it would help further any concepts which remain plausible.