Skip to content

Sean Forman: WAR is “GDP for baseball”

Mar 30, 2013, 8:05 PM EDT

Mike Trout AP

There was some big news in the baseball stats community recently, as Baseball Reference and FanGraphs agreed to a common replacement level in their Wins Above Replacement statistic. Though the two sites’ computation of the statistic still differs in other ways, creating some uniformity in replacement level should clear up some of the confusion around the stat among casual observers.

Sean Forman, the creator of Baseball Reference, responded to some general criticisms of WAR yesterday. Most notably, he referred to WAR as “GDP for baseball”. GDP, of course, stands for Gross Domestic Product, an estimate of the market value of a country’s goods and services. WAR often gets tossed aside because it’s complicated, requires a lot of steps to calculate correctly, and is a single number that covers broad subject matter, but Forman shows how those criticisms are not levied against GDP despite being very similar in nature. If you trust GDP, you should trust WAR.

(Obligatory picture of Mike Trout above, just because.)

  1. Bryz - Mar 30, 2013 at 8:13 PM

    It makes me think of PER in basketball – which Minnesota media threw around all the time because Kevin Garnett led the NBA in it for years – and also QB rating (which uses “if… then” statements in its calculation, by the way!). Those two numbers are very hard to compute on your own, but they’ve been accepted by many people, yet WAR and even simpler baseball stats continuously are hounded by the traditional types.

    I’m willing to bet there was similar backlash with QB rating when it first came out, and yet it’s widely accepted now. Sabermetrics’ time will come eventually when the majority of fans display acceptance.

    • hisgirlgotburrelled - Mar 31, 2013 at 1:42 PM

      QB rating is a great example in showing how ridiculous the criticism is of WAR when you consider how QB rating is so widely accepted and rarely rejected.

      A QB rating has a maximum and a minimum, 0 – 158.3. A “perfect” rating is 77.5% completion %, 12.5 yards per attempt, a TD per 11.875% attempts, and 0 interceptions, which were established over 40 years ago and have never been adjusted. Having a maximum and minimum and never adjusting the calculation to how the game has evolved is terrible. The calculation is actually not that complicated in comparison to PER and WAR; you can’t do it in your head, but it doesn’t require a spreadsheet.

      So many accept and defend statistics like QB rating, and pitching wins, as showing a player’s value but have no patience for WAR when it tries to establish actual value. It’s accepted because we know anything over 100 is a good game, a great season, and a hall-of-fame career. Similarly, ESPN has a new total QB rating, much like WAR, which is weighted on down, distance, and situation, and also includes every play, not just pass plays, and this stat is also rejected, even more so than WAR.

      The newer stats will become more accepted when people understand the standard of a great player in comparison to an average player.

  2. kollin7 - Mar 30, 2013 at 9:13 PM

    Makes sense, actually.

  3. historiophiliac - Mar 30, 2013 at 9:45 PM

    Nice pic of ESPN’s #3 player.

  4. mungman69 - Mar 30, 2013 at 9:47 PM

    WAR, What is it good for?

    • senioreditor2 - Mar 30, 2013 at 11:03 PM

      absolutely nothing?

  5. vpisteve99 - Mar 30, 2013 at 10:07 PM

    The WAR/GDP analogy is good. In my opinion, these are great measurements to use as starting points but shouldn’t be used in a vacuum. There is some legitimacy to the arguments against WAR. Any complex statistic that quantifies qualitative measurements is always going to be flawed to some degree but it doesn’t mean it should just be dismissed.

    • paperlions - Mar 31, 2013 at 11:29 AM

      I agree. The problem with the arguments “against” WAR (they really aren’t arguments so much as railing diatribes) is that those arguments are made in a vacuum that doesn’t exist. In most of those arguments, the arguer acts as if someone does use WAR as the be-all-end-all, which no one with any knowledge does not and that the stats they prefer have fewer or smaller flaws than WAR, which they do not. In short, people the mis-use WAR are the same people that mis-use all stats, and the stats some people prefer over WAR (and it’s components, which are really the important stats to look at) are more incomplete, highly contextual, and more deeply flawed when it comes to evaluating performance.

  6. obpedmypants - Mar 30, 2013 at 10:23 PM

    I should hope WAR isn’t as misleading a metric as GDP.

  7. dohpey28 - Mar 31, 2013 at 9:00 AM

    If the same ‘stat’ can be computed differently then it really isn’t a stat is it?

    • 18thstreet - Mar 31, 2013 at 3:10 PM

      Different people would rule the same play a two-base error, a single plus an error or a two-base error. Should we ignore batting average, too? Total hits? Total bases?

      Every statistic has its flaws. I like WAR, but it’s not perfect. I don’t think anyone believes it is.

  8. American of African Descent - Mar 31, 2013 at 9:14 AM

    Here’s the problem with the analogy.

    GDP=Consumption + Investment + Government Spending + Net Exports. (It’s often written as C+I+G+(X-N).)

    WAR=???

    GDP may be complicated to calculate, but at the end of the day the numbers are what they are.

    Wins Above Replacement claims to quantify how may wins a player is worth over a so-called replacement level player (whatever that means) at his own position. In theory, then, you should be able to add the WAR of every player on a team and determine how many game over/under .500 the team is. Ever work?

    Further, WAR doesn’t account for things like how much better certain players make their entire team. For example, a line-up with Barry Bonds in it is going to be much better than the same line-up without Bonds. (Why? Because the players in front of Bonds are going to get more pitches to hit because you don’t want to face Bonds with men on base.) WAR doesn’t account for synergies between a pitching staff intentionally built with ground ball pitchers and middle infielders who are superb with the glove and subpar with the bat.

    Instead, give me OPS, give me UZR, give me ERA, Innings Pitched, and K/9IP, and that really does help paint a good picture of who has talent and who does not.

    You don’t always need to invent new statistics — many times the old tools are perfectly good ones.

    • Kevin S. - Mar 31, 2013 at 9:26 AM

      Oh, if we’re simplifying the formula, WAR is easy – it’s Batting Runs + Fielding Runs + Baserunning Runs + Position Runs + Replacement Runs. The different methodologies get there in slightly different ways, but that’s the basic structure of it.

      • American of African Descent - Mar 31, 2013 at 10:41 AM

        I have no issue with the batting runs and fielding runs. How do you measure base-running runs? Because there are some tremendous base-runners who just aren’t very fast, but are capable of taking the extra base, stealing the odd base, and generally adding value. There are other base-runners (see Coleman, Vince) who are lightening fast but cost their team runs by getting thrown out more than 25% of the time.

        What are “position runs”? Never heard of them. And what is a “replacement player”? How does WAR account for a great defensive player who plays along side of another great defensive player? By way of example, consider a team that has a stud center fielder and a stud left fielder, but a so-so right fielder? The center fielder and left fielder can cheat a bit to their left, allowing them to cover for the right fielder.

        My problem really is that the statistic should not be called Wins above replacement given all the factors that go into winning the game that the statistic just doesn’t measure.

      • paperlions - Mar 31, 2013 at 11:38 AM

        In addition, WAR ignores context (e.g. margin of victory), if you translated run differentials into expected wins rather than realized wins, the numbers will come out much closer. Regardless, this is an irrelevant argument as the importance of WAR is relative, not absolute (i.e. the value of the metric derives from the ability to effectively compare the value of players from different positions, accounting for level of competition and park effects…not from the absolute numbers themselves).

      • manchestermiracle - Mar 31, 2013 at 11:40 AM

        AAD: That’s my feeling toward won/loss record for pitchers, when ERA (among others) is a better indicator of ability. Pitching doesn’t “win” games, pitching prevents the other team from winning. Hitting wins games and pitchers are often not hitters. In the AL they aren’t even allowed to hit. To claim a won/loss record is indicative of a pitcher’s ability is to ignore that a win is a team endeavor, much like WAR. As always, any stat taken out of context becomes worse than no stat at all.

      • American of African Descent - Mar 31, 2013 at 1:31 PM

        I agree with you, manchestermiracle. ERA, Innings Pitched, and WHIP are all good ways to measure a pitcher’s effectiveness. Wins and losses are really something a pitcher cannot control.

    • paperlions - Mar 31, 2013 at 11:43 AM

      FWIW, OPS is a freaking horrible statistic. It sums components with different denominators (OBP includes BB/HBP/SAC, SLG doesn’t), that are on different scales (OBP 0-1, SLG 0-4), and that have different importance (OBP is much more valuable than OPS). It is about as bad a stat as you can get since to guys with an OPS of .800 are not necessarily equivalent at all: a guy that hits .400/.400 is far far FAR more valuable than a guy that hits.300/.500.

      ERA is also horribly flawed. An ERA belongs to both the pitcher and his fielders, it gives pitchers credit for great plays made that should have been hits but weren’t, but puts all the blame for “un-earned runs” on a fielder. It also isn’t adjusted for level of competition, league, or park.

      • American of African Descent - Mar 31, 2013 at 1:38 PM

        Wait, OPS is worse than batting average? RBI? OPS may not be perfect, but to call it “freaking horrible” is just not right either. OPS is easy to calculate in your head (or on the back of an envelope). Much easier than OBP squared plus slugging, which would give you a much better metric, but which is much more difficult to calculate.

        ERA is “horribly flawed”? It may not be as good as WHIP, but “horribly flawed” is a bit much. Your complaint that it “isn’t adjusted for level of competition, league, or park,” doesn’t hold water. At the end of the day, you can only play who you play.

        And show me a (reliable) statistic that’s completely independent of what someone else on the team does? For example, you think having Barry Bonds hitting behind you doesn’t increase your OBP?

      • paperlions - Mar 31, 2013 at 9:05 PM

        None of which changes the fact that each of those metrics are bad at measuring what you want to use them to measure.

      • historiophiliac - Mar 31, 2013 at 6:23 PM

        Still mad?

      • paperlions - Mar 31, 2013 at 9:05 PM

        You being wrong doesn’t madden me.

      • historiophiliac - Mar 31, 2013 at 9:07 PM

        ha ha

        I went easy on you today. Don’t pout now. I don’t want to argue.

Leave Comment

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

Featured video

The Jon Lester-Yoenis Cespedes trade is a win-win
Top 10 MLB Player Searches