Skip to content

FanGraphs and Baseball Reference agree on a value for replacement level in WAR

Mar 28, 2013, 10:32 AM EDT

Mike Trout Getty Getty Images

When we talk about WAR, we’ve historically been talking about two different metrics, actually. Because the two entities which developed and calculate WAR — FanGraphs and Baseball-Reference.com — have historically used two different values to determine the “replacement level” component of the stat.

While those of you who dig into these metrics with anything beyond a passing glance were aware of and made accommodations for this, it led to at least some confusion among more casual observers and led to a lot of potshots from the fans and the press looking to take issue with any stat more complicated than batting average. “Hey, why should we care about WAR if you guys can’t even agree what it is,” they would say, prior to coming up with some Edwin Starr/”what is it good for” bon mot which they believe to be original and clever.

Now, however: a grand agreement. From Dave Cameron:

You can calculate replacement level a number of different ways, but in the end, it always leads back to a number in this vicinity. Baseball-Reference arrived at a number a little higher than what Tango had used, while we came up with one a little lower. Because they were at opposite ends of the defensible spectrum, the different baselines gave a false sense of difference in the actual calculations. Now, with an agreed upon replacement level, those differences that are solely due to scale will go away.

There will still be some subtle differences in the ultimate calculation (read Dave’s post for more information on that) but now one big difference is gone, which should lead to a bit more harmony and less unnecessary strife when it comes to this stuff.

I’m sure some people will still take pot shots “Hey, if it’s so good, why do you have to tinker with it all the time!” they will claim, ignorant of how math and science work, but this is a good move.

(why the Mike Trout pic? Eh, didn’t want to use the calculator for two posts in a row)

  1. Kevin S. - Mar 28, 2013 at 10:42 AM

    Minor nit – in the first paragraph you mean Fangraphs and Baseball-Reference. B-Pro does WARP, which is a little bit more apart from the two WARs.

    • Kevin S. - Mar 28, 2013 at 10:42 AM

      But otherwise, fuckin’ A yes!

  2. clydeserra - Mar 28, 2013 at 10:54 AM

    Our long national nightmare is over.

    • chacochicken - Mar 28, 2013 at 11:15 AM

      One of them anyway. The other as, evidenced below, is the dire need of an EDIT function on HBT.

      • carbydrash - Mar 28, 2013 at 11:56 AM

        That’s noonsense. Just right properly and proofreed before you post like a adullt.

      • Kevin S. - Mar 28, 2013 at 12:01 PM

        I really, really hope “noonsense,” “right” and “proofreed” were intentional.

      • carbydrash - Mar 28, 2013 at 2:44 PM

        They weren’t! Boy, is the egg on my face red or what!

  3. chacochicken - Mar 28, 2013 at 11:05 AM

    Do that make Mike Trout the Ultimate WARrior?

    • paperlions - Mar 28, 2013 at 12:51 PM

      Yes, yes it do.

  4. philliesblow - Mar 28, 2013 at 11:11 AM

    Miguel Cabrera believes in Peace, not War.

    • proudlycanadian - Mar 28, 2013 at 3:24 PM

      Not piece?

      • tommyshih - Mar 29, 2013 at 4:08 AM

        No. Pies.

      • proudlycanadian - Mar 29, 2013 at 7:41 AM

        Really? That comment takes the cake.

  5. spudchukar - Mar 28, 2013 at 12:05 PM

    The WAR stat still has issues. Glad to see the two camps have come together, but an asterisk or parenthetical caveat should always accompany the stat. WAR can be helpful comparing players, but what needs to be remembered is that the “W” signifying wins doesn’t really translate into actual wins, but is simply indicative of one player’s performance vs. another’s.

  6. El Bravo - Mar 28, 2013 at 12:17 PM

    Have they determined the true calculation for G.A.R. or A.R.G.?

  7. tc4306 - Mar 28, 2013 at 12:40 PM

    The development of statistical metrics is an ongoing process. Partly, it is the continued search of perfection. Partly, it is the modern day version of “You don’t walk off the island.” There are a lot people whose livelihood depends of the development and use of metrics. If you expect anything but constant change, you are hopelessly naive. Several Fangraphs authors have acknowledged “problems” with the defensive metric and have called WAR a “flawed statistic.” Sounds very much like they are preparing us for the “next great thing.” Not to say that metrics are useless: they have some merit. But I’m not buying that they are the be all and end all that their proponents claim they are. Heck, even the moneyball experts in Oakland have tested last year’s MVP race with their metrics. Their conclusion? It was incredibly close and Cabrerra probably deserved the award over Trout. And don’t bother quoting a bunch of flawed WAR stats to tell us how wrong that decision was.

    • stex52 - Mar 28, 2013 at 12:53 PM

      I wasn’t surprised that Oakland came to that conclusion. One comes away from the book with the impression that their metrics always favored OBP + power stats over speed or fielding.

      What would be interesting to me would be how the systems for some of the other teams rated the contest. Everybody uses the tools these days.

      And it is not in the nature of science to ever stop looking for better statistical tools. They should be looking for the “next great thing.”

    • paperlions - Mar 28, 2013 at 12:55 PM

      There is no such thing as an unflawed statistic. Calling any statistic at all flawed is redundant. All statistics are estimates of parameters, whose true value can not be known. Of fucking course, it is flawed…so what. If you really think that is an issue, then you do not understand anything at all about statistics or data or how they reflect “truth”.

      • spudchukar - Mar 28, 2013 at 1:28 PM

        Salient clarification, but not sure I understand where your angst is directed. Many of us who criticize the relatively new statistical analysis, recognize errors in hopes of perfecting the existing measurements. The most ardent followers of Sabermetrics are much more likely to hold them infallible than the critics.

      • paperlions - Mar 28, 2013 at 1:31 PM

        Because the tone of the comment is as if 1) there are some perfect metrics out there, and 2) as if those developing metrics are doing so suboptimally for job security purposes. Which is both offensive and a ridiculous thing to say or think…an attitude that demonstrates a complete lack of understanding of how new knowledge is created.

      • spudchukar - Mar 28, 2013 at 1:36 PM

        Or a lot of “old” knowledge for that matter.

  8. tc4306 - Mar 28, 2013 at 2:22 PM

    OF COURSE job security is a component to the development of metrics. That is neither offensive nor ridiculous. It is just common sense. Be it cars, computers, washing machines or metrics, you have to develop “the next great thing” or the competition will and you will be out of business.

  9. unclemosesgreen - Mar 28, 2013 at 7:03 PM

    We should name this guy, this anonymous Mr. Replacement. I move that we call him Paul Westerberg.

Leave Comment

You must be logged in to leave a comment. Not a member? Register now!

Featured video

This was 'the perfect baseball game'
Top 10 MLB Player Searches
  1. S. Kazmir (5392)
  2. G. Springer (3731)
  3. M. Machado (3063)
  4. K. Uehara (2725)
  5. B. Harper (2618)
  1. C. Kimbrel (2595)
  2. D. Pedroia (2502)
  3. J. Reyes (2433)
  4. J. Chavez (2395)
  5. C. Granderson (2393)