By In Stuff

Baseball Reference WAR: A Response

For some reason I don’t understand, my friend Sean Foreman’s response to my Baseball Reference WAR piece went into spam. Sean is president and founding partner of Baseball Reference and someone who has been hugely helpful to me through the years. And he makes his expected excellent points here.

I have pulled his response out here for a full reading.

* * *

I would have been happy to comment on the record for this post. I’m on vacation, so this is a bit rushed.

I’m not sure what you are asking of us here. We state clearly that we don’t find differences of 1-2 wins to be definitive. We also break out each and every component of the stat so users can take that into account. Should we require user training before viewing our stats?

You are correct that the only real difference in WAR between the two pitchers in WAR is on the defensive adjustment.

Then the question becomes the following. While JoeP believes that the Red Sox were a “much, much, much” better defensive team than the Tigers overall, Joe believes that the Tigers were then also way better for Verlander than for their other pitchers.

Maybe it’s true that the Tigers were above average fielders when Verlander was on the mound, maybe not, but keep in mind BIS had the Sox at +59 and the Tigers at -49 for the season. How on earth does a team that’s the 3rd worst defensive team transform itself into an above average defensive team for the 228 innings JV was on the mound given they’d then have to be EVEN worse the other 1200 innings to get to -49?

If we assume for the moment the Tigers were in fact “excellent” behind Verlander, then the question becomes how to handle this. We apply the team’s DRS to each pitcher based on the percentage of the team’s balls in play which in probably 95% of the cases is a good way to do this and may still be in Verlander’s case. If you start to dice things up by the pitcher on the mound you then run into very small samples where something like Mookie Betts pulling back a home run and getting a double play has a dramatic impact on the pitcher’s WAR. I don’t think you’d like the alternative as you run the risk of conflating random variance with real performance differences.

Joe seems to agree that defense aside that Porcello and Verlander were equals. I think it is a huge leap to then say that they were still equal when considering the likely impact of the team defense given what we know. I’m very comfortable saying that Verlander’s performance was more valuable than Porcello’s, though AS ALWAYS the amount of difference is up for debate.

If we find there is an issue with our defensive numbers, we’ll look at making a change. We are always trying to improve.

Print Friendly, PDF & Email

43 Responses to Baseball Reference WAR: A Response

  1. DjangoZ says:

    Appreciate that he responded, but that came off more than a touch defensive (no pun intended). Why not go into the stats and take a look when he gets back from vacation?

    I think Joe found something interesting here.

  2. Pat says:

    Because, as he stated, then you run into issues with sample size and even more randomness than are already in baseball stats.

  3. PhilM says:

    I agree, the tone left a bit to be desired: though maybe feeling he had to defend “work” while on vacation stirred Sean’s invective a bit. This gets to the heart of why I just prefer to work with ERA+. For all its failings, taking errors out does attempt to mitigate the actual performance of the defense when the pitcher is actually pitching. Using ERA+ lets you calculate what an average team would score on behalf of a pitcher in a neutral park. Combine that with earned runs actually allowed using Pythagorean or Paschal or Patriot-Symth or whatever, and you get a win-loss record to hang your hat on. Porcello was marginally better than Verlander, and Kluber was still better than both. . . .

    • invitro says:

      I think Forman’s method does a much better job of mitigating for team defense than ERA does. Certainly we all agree that the statistic of errors measures only a tiny part of the effectiveness of defense, right?

      • PhilM says:

        Totally agree that errors are too subjective and not comprehensive: but I still think there’s too much noise in the defensive data to incorporate it smoothly. Using just ERA+ and calculating neutral win-loss records, I get league leaders like this since 2009: a list that seems fairly defensible (see what I did there?) to me.
        2009: Zack Greinke, Adam Wainwright
        2010: Felix Hernandez, Roy Halladay
        2011: Justin Verlander, Roy Halladay
        2012: Justin Verlander, Johnny Cueto
        2013: Max Scherzer, Clayton Kershaw
        2014: Corey Kluber, Clayton Kershaw
        2015: David Price, Jake Arrieta
        2016: Corey Kluber, Kyle Hendricks/Jon Lester

        • invitro says:

          I agree that that list, and ERA+, are certainly defensible, way more than defensible, really. I think ERA+ is most excellent. But I think RA+ might be a little bit better (I guess that’s a factual question), and something like bWAR/G even better still.

  4. Matt says:

    I really don’t understand the assertion that “If you start to dice things up by the pitcher on the mound you then run into very small samples where something like Mookie Betts pulling back a home run and getting a double play has a dramatic impact on the pitcher’s WAR”. Why is that a bad thing? A fielder making a play like that already does have a dramatic impact on the pitcher’s ERA, FIP, and all of the stats derived from those. If I understand the current system correctly, the way a play like that is handled now is:
    – Give the pitcher on the mound credit for recording 2 outs without allowing any runs.
    – Give the fielder credit for something like 3 runs saved (2 runs taken off the board plus 2 outs recorded, which I’m guessing is around 3 runs)
    – Debit all of the pitchers on the team some fraction of 3 runs based on how many total balls in play they allowed.

    Why would we not just want to debit the pitcher on the mound in that case rather than all of the pitchers on the team? If “random variance” causes one pitcher to have a lower RA thanks to some great fielding plays behind him that other pitchers on the team didn’t get, why wouldn’t we want that to be reflected in his WAR?

    • invitro says:

      “If “random variance” causes one pitcher to have a lower RA thanks to some great fielding plays behind him that other pitchers on the team didn’t get, why wouldn’t we want that to be reflected in his WAR?” — Seriously? WAR is supposed to filter out as much extremely good/bad luck as possible, and such great fielding plays as you describe are extremely good luck.

      • Matt says:

        Right, I think we are arguing for the same thing but I was unclear when I said “reflected in his WAR”. If pitcher A and pitcher B on the same team have the same RA, but Mookie Betts pulled back a home run behind pitcher A, we want pitcher A to have a lower WAR than pitcher B, right?

        If I’m understanding Joe’s explanation of bWAR then both pitchers would end up with the same WAR. And if I’m understanding Sean’s reply, then he is saying it’s a *good* thing that it is that way.

        • invitro says:

          “we want pitcher A to have a lower WAR than pitcher B, right?” — I don’t know. My brain is taxed on this one. I suppose it depends on the reliability of defensive stats in support of a certain pitcher. I will defer to Forman on this one, as he’s of course done the work, I haven’t, and I trust him.

          • Matt says:

            Fair enough. I Foreman’s point is that we can’t trust defensive stats the be precise enough on a play-by-play level to do this, then I can buy that. But if he’s saying that we wouldn’t want to do it in a world with perfect defensive stats for some reason, that’s where I disagree.

  5. heaveecee says:

    As the response stated, I don’t think 1 – 1.5 WAR is that big of a difference when comparing Verlander vs Porcello. Both rate as all-star value, but haven’t reached the high elite levels of peak Pedro or Rocket or Unit or even Kershaw’s recent years. The voting overall showed that it was close, though I’m still in the camp that the voting is flawed and first place votes should be worth more. If defence is the adjustment that separates them by a win or 2 so be it. It was close. There’s no way to know for sure, but I think Verlander got more 1st place votes based on the facts that he lead the league in SO and had an impressive double digit SO/9 while putting otherwise close numbers. I don’t believe K/BB gets that much consideration. The way Curt Schilling’s HoF case has been treated by writers (best k/bb ever) may be evidence of that.

    • PhilM says:

      I’ll second the relative (and largely deserved) lack of K/BB importance: I don’t think anyone believes Phil Hughes had a season for the ages in 2014, although he did have the best K/BB season of all time. And Babe Adams, Deacon Phillippe, Fred Hutchinson, and Jim Whitney aren’t in the Hall of Fame, although they led the league 4x in K/BB, the same as Kid Nichols, Pedro Martinez, and Roger Clemens. Other things, like WHIP and ERA+ and even Shutouts measure much better and matter much more.

      • invitro says:

        Well… the K/BB of the league has had a huge amount of variance over baseball history. So I think you’d need to use K/BB over league average to compare pitchers of different eras. ERA+ does that, so it’s superior as a one-stat comparison (and would be anyway, just for including more information). And I’m feeling really lazy right now, so I’m not going to look at his career, but isn’t Babe Adams one of those guys with an excellent HoF case? 🙂

        • PhilM says:

          Oh, don’t sell yourself short: Babe Adams most definitely has a strong Hall of Fame case! He led his league 5x in WHIP – everyone else with 4x or more is in the Hall. Plus the 3-0, 1.33, 0.889 WHIP performance in the 1909 World Series title, and the fact that he was a real looker (hence the nickname), and he’d be a shoo-in today. Even FIP likes him: 4x leading the league in that newfangled, you-kids-get-off-my-lawn stat.

          • invitro says:

            I had to look up Babe Adams on wikipedia. A note: “Adams later managed in the minor leagues, farmed in Mount Moriah, Missouri, and worked as a reporter and foreign correspondent during World War II and the Korean War.” Well, Adams was 60 years old in 1942, and 68 in 1950… that’s some retirement!

  6. Rick Rodstrom says:

    Joe’s article is one of the best I’ve ever read—thoughtful, well researched, cogently argued, clearly written—and this reply helps pull back the curtain on WAR even more. The gist being that, WAR is only an opinion. It’s an opinion that makes assumptions all the way down the line—informed assumptions to be sure—assigning different weights to different measures to achieve a single emphatic rating. It is not enough to simply state that Porcello pitched in front of a better defense than Verlander, but that defense was worth exactly 13 runs. Or park effects—Wrigley Field can play as big as a canyon or as small as a sandbox depending on which way the wind is blowing. To assign it a single variable that applies to every start is misleading. There are so many variables that can effect performance (remember when Joba Chamberlain was done in by gnats?) that to reduce a season to one digit is the height of hubris.

    Now there’s nothing wrong with Baseball Reference having an opinion—I have my own, one that assigns a greater weight to slugging than OBP, values relievers more than some people do, places a premium on playoff performance, etc. The difference is that I don’t score it to the decimal point to make it an ersatz stat that has the solidity of strikeouts or home runs. Sean Foreman may argue that he doesn’t believe 1-2 wins are definitive, but try telling that to the geek who says that since Player A was worth 4 WAR and Player B was worth 2 WAR that Player A was definitely better, no questions asked. Which is the real sin of WAR: unlike the controversial stats of yore like Wins and RBI’s and Batting Average, which were meant to start arguments, WAR was meant to end them. THIS is the true value of a player, period. It reduces the entirety of a baseball season into one guy’s value judgement. Which is fine if you’re that guy, but not if you’re anyone else.

    • invitro says:

      “one that assigns a greater weight to slugging than OBP” — Well, you’re just factually wrong there, if you’re saying SLG is more helpful to an offense than OBP. I’ve actually done this myself: OBP is 1.8 times as important, or to be more precise, the best linear correlation of runs to OBP and SLG is: runs = C (1.8 OBP + SLG) + D, where C & D are constants. Or to be even more precise, it says an OBP 10 points above league average is worth the same as a SLG 18 points above league average. Don’t just believe me, though, search for it, or better yet, do the regression yourself. FWIW, this is why OPS and OPS+ are fairly broken as stats: they weigh OBP and SLG equally. (I’ll welcome any corrections!:))

      • invitro says:

        If that isn’t enough, here’s a question: would you rather your team lead the league in OBP or SLG? Have you checked to see how often a league leader in OBP also leads the league in runs, vs a league leader in SLG? I think I will do that…

        • invitro says:

          I’m quite surprised, but it looks like it’s better to lead the league in SLG! I checked back to 1967… the leaders in OBP and SLG rank essentially the same in the AL, but the NL SLG leader does much better than the OBP leader. This may be a recent thing (my excuse): in the last 14 years, the NL SLG leader was 1st in runs 12 times, and 2nd 2 times. The NL OBP leader came in 1st 3 times, 2nd 7 times, 4th 2 times, 5th once, and 8th once. So you’d want your NL team to lead in SLG rather than OBP if they played in the last 14 years…

          • KHAZAD says:

            This is partially due to pitchers hitting, which leads to a smaller spread in OBP. The NL has nearly twice as many sacrifice bunts and intentional walks which revolve around the pitcher spot and are evenly distributed throughout the league, leading to less variance.

            For Instance, over the past 5 years (small sample but limited time) the AL leader in OBP was 22 points above average, the leader in slugging was 40.6. This puts the slugging difference at 1.85 times the OBP difference, or very close to your stated value.

            The NL OBP difference was 17.8 points, the slugging difference was 41.6. This puts the the slugging difference at 2.35 times the OBP difference or well above the point where they would even out according to your formula. With such a larger variance, it would follow that the slugging leader in the NL would dovetail more closely with the leader in runs.

          • Rick Rodstrom says:

            You deserve props for reconsidering your opinion in light of some research. My own opinion is that slugging is a much better indicator of offensive prowess than OBP because OBP relies much more on walks while slugging relies on driving the ball. Chone Figgins led the league in walks in 2009, which gave him a .395 OBP (good) and a .393 slugging average (bad). That OBP netted him a 35 million dollar contract, but when pitchers realized he couldn’t hurt them with his bat, they simply challenged him and his career went into a nosedive. There are a lot of factors that will artificially inflate your OBP, like IBB, getting HBP, pitcher wildness, and who is batting behind you in the lineup. Buddy Harrelson once walked 95 times even though his slugging average was .309 because he hit in front of the pitcher. BFD. Meanwhile, making contact can be beneficial even if it doesn’t result in a hit—SAC flies, runners advancing, errors etc—so if anything, slugging is statistically undervalued where OBP is overvalued. It’s not that OBP has no value, it’s that between OBP and slugging, give me the masher every time. Barry Bonds walked 232 times in 2004, and pitchers were happy to do it.

      • Kuz says:

        What’s the coefficient of correlation of the OBG to SLG linear correlation? Is it greater than 0.3?

        r > 0.3 ?

        An r of 0.3 is generally considered to be a weak positive correllation. An r of 0 to 0.3 is generally considered to be a statistically insignificant positive correllation. M

        • invitro says:

          I can’t remember what I got. I think I did that regression about six years ago when I was in a grad stats class, a business school class though, not a math class, and learning ANOVA and those other stat tests. Well I don’t remember how to do it now, but I sucked down all the team stats from b-r, and hacked up a program to do some regression.

          I might have done something wrong, though I’ve tried to check it closely… for 1973-2016, I get the equation R/G = -1.108 OBP + 12.059 SLG – 0.009, with R^2 = 0.640. Now something has gotta be screwy somewhere because of that negative OBP coefficient. But the predicted R/G values really do look close to the actual ones. (The standard deviation of the errors, R/G’-R/G, is 0.270.) I looked several places for a webpage with both complete formulas and an example; I used the ones here:

          I don’t have Excel, I have Libreoffice, and while it has ANOVA and other tests, I either can’t interpret them or they don’t do quite what I want. What I want is a test that says something like 45% of the variation in R/G is due to variation in OBP, and 50% is due to SLG, and 5% is due to something else. I think I remember ANOVA doing something like that.

          I was wrong above and misinterpreted that 1.8 number as meaning OBP was 1.8 times “more important” than SLG. Like KHAZAD pointed out, if that number is/was correct, it just means that the deviation in SLG is 1.8 times that of OBP, or something like that, I think. I retract what I wrote, in any case :).

          Now what I want to do is regress R/G against OBP, SLG, and OBP*SLG. I expect that OBP*SLG is much more “important” than OBP or SLG on their own, or as a sum, only because Runs Created is basically OBP*SLG, and we know there’s a high correlation of runs to RC. At least that’s what I remember… I’m trying to trust my memories a little less…

          • Rudy Gamble says:

            1.8 * OBP + SLG is the quick/dirty calculation for wOBA (it correlates to wOBA but is not scaled like wOBA) which is generally recognized as the best commonly available stat for explaining an offense’s runs.

            I believe OBP gets weighted up for reasons such as: SLG is on a higher scale (max 4.00 vs 1.00), SLG overstates the value of total bases as they relate to runs – e.g., a double is worth less than 2x a single, and SLG does not account for walks. This weighting DOES NOT mean that OBP is 1.8x more valuable than SLG.

          • invitro says:

            Thanks for your reply! I’m glad to hear my memory wasn’t entirely bad recalling that factor of 1.8. Does the 1.8 have anything to do with the ratio between the stddevs of (team, player?) OBP and SLG? (That ratio is 2.16 for teams from 1973-2016.)

            Is it true that 1 point of OBP is as valuable as 1 point of SLG?

            Would you agree that 1.8*OBP + SLG should replace OPS and OPS+? (Both OPS and OPS+ are extremely widely used right now.)

          • invitro says:

            I meant to ask: is 1 point of OBP as valuable as 1.8 points of SLG? 🙂

          • Rudy Gamble says:

            Yes, 1 point of OBP worth as much as 1.8 point of SLG when it comes to run creation. 1.8 * OBP + SLG or wOBA are better than OPS. Worth looking into Tom Tango (aka Tangotiger) and the research behind ‘linear weights’ for MLB events. ‘The Book’ is great.

          • Kuz says:

            Thanks for your in depth reply. Your r^2 value of 0.640 shows a very strong correlation of your line of best fit. So your conclusion is definitely statistically significant. I asked about the r^2 value because I read an article on Fangraphs that had a linear correlation looking at exit velocity versus OBP or BA or something and r^2 value was extremely low, meaning the correlation was not statistically significant. Also, I wouldn’t reject out of hand your -1.108 coefficient of OBP vs. R/G. This of course is anecdotal evidence, but think about the huge number games where the team with the better OBP loses the game (scores fewer runs). Quite often, the team with the greater number of runners left on base loses the game, but has a better OBP. Concerning your question about determining which variables contribute what percentage of variance to an output: use “multiple linear regression”. I’ve used it with good results in identifying key variables in manufacturing processes in real life. But again, the results must be tested for statistical significance.

      • MikeN says:

        At one point, WSJ was always using SLOB.
        Have you tried regressing with aS+bO+cOS+d?

    • Richard says:


      In scientific research, it is customary to add error bars to a number to acknowledge any uncertainty in a measurement. Much of the mathematical field of Statistics deals with how one propagates those uncertainties through a calculation.

      It only takes a single hit to turn a player from a .299 hitter to a .301 hitter. Did a missed strike call extend a player’s at bat and allow him to get that hit? Did the wind keep a ball in the park and within reach of an outfielder? There are so many little things that are outside of the player’s control.

      We really should be including error bars in our statistics.

  7. Mark Daniel says:

    My interpretation is that DRS, used by B-R in dWAR, overemphasizes difficult plays. Players get more points added to their score if they make a difficult play. Considering the vast majority of balls hit to fielders are routine, the differences in dWAR between players comes down to a handful of difficult plays made (or not made) over the course of a season.

    This is a problem because whether a play is difficult or not has no bearing on winning. A missed grounder by a SS results in a man on 1st regardless of whether it was a routine play or an extremely difficult play.

    DRS actually works really well when used to rank players from best to worst. But I don’t believe it is appropriate to include in WAR.

    I’d love to hear a rebuttal of this theory, because it’s possible I’m not thinking of it properly.

    • invitro says:

      I don’t know if I have the right page… I’m looking at and it says: ‘In order to translate each component to Runs Saved, we consulted the “24-States Run Matrix”. We compared the expected number of runs allowed before and after each play and calculated the average change in run expectancy for each event. We then apply these average run values to convert to Runs Saved.’

      So, not more points for a difficult play, but they do get more points for a more important play. Which makes sense to me. But this goes against the context-neutral idea of WAR, and would insert an element of clutch. Maybe the Fielding Bible’s DRS is not the one b-r uses?

      • Mark Daniel says:

        That’s a good point. But I think they use both, actually. From the same link- “In the Plus/Minus system, the computer totals all groundballs hit by right handed batters to Vector 206 (Vector 206 is a line extending from home plate towards the hole between the normal shortstop and third base positions, 19 degrees off the third base foul line) with an average velocity between 65 and 75 miles per hour and determines that these types of batted balls are converted into outs by the shortstop only 23 percent of the time. Therefore, if the shortstop converts a slowly hit ball on Vector 206 into an out, that’s a heck of a play, and it scores at +.77. The credit for the play made, 1.00, minus the expectation that it should be made, which is 0.23. If the play isn’t made, it’s -.23.”
        Then they multiply that number by the run expectancy based on the situation.

        So a difficult play made with the bases loaded and 0 outs would provide a player a lot of points. Like you said, this goes against the neutral idea of WAR, so maybe they don’t use this system.

        • nightfly says:

          You can argue that either way.

          On one hand, an out is an out is an out. On the other hand, guys who turn more difficult plays into outs are more valuable. With the bases loaded and nobody out, the guy who can make the difficult play is more valuable than a guy who can only wave helplessly at the ball as it sails by.

          For the fielder, it certainly makes more sense to credit difficult plays more; for the pitcher, the “credit” is already measured by his not giving up those runs.

          • Mark Daniel says:

            Well, yeah, he’s more valuable. But I’m arguing it doesn’t make sense to award more points for a difficult play. It’s because defense is converted to runs, as in runs saved. The way it’s set up is that in two identical situations, if one guy makes a difficult play at SS with the bases loaded, he somehow saves more runs that a guy who makes an easy play. Why is that? It’s still a ground ball out. It’s not like if both players did NOT make the play that they would allow differing amounts of runs.

  8. James says:

    There’s a rant about WAR I’ve been holding in for a while and I figured I might as well try to put some of it here.

    WAR, whether you’re talking about Fangraph’s or Baseball Reference’s, is an opinion. It is certainly an informed opinion, but it represents an opinion as to what is valuable. It’s a good starting point, perhaps, but it should never be the definitive answer as to whether one player is more valuable than another.

    The biggest problem with WAR for me is that it attempts to define value in terms of a neutral sense of how many wins a player adds to his team, derived from how much particular events are mathematically calculated to improve the probability of a win, completely devoid of context. Even on that level, WAR is flawed because it does not distinguish between an incredibly valuable home run that gives a team a lead in the late innings and one that is in a 14-2 game.

    But even then is the wrong approach. Baseball is entertainment. We watch baseball to be entertained. While it is true that winning is generally entertaining, it’s not a perfect measure. Let me delve into a hypothetical example.

    Suppose you have a player with a noodle bat but a very good eye. Instead of coming to the plate with the objective of hitting the ball fair, he comes to the plate with the objective of fouling off every good pitch and trying to get a walk. Let’s suppose that this player becomes very good at this, to the extent that his pitches per plate appearance is off the charts and his OBP is around .600, almost all of it due to getting walked. His batting average is about .020, and his slugging percentage is identical.

    By the standards of WAR, such a player would be incredibly valuable, and indeed having such a player (if one could possibly exist) would definitely contribute to a team’s wins. A team full of such players would win a lot of games. But in terms of value in providing entertainment, such a player provides very little value. He’d be making a mockery out of the game and would be painful to watch.

    That, for me, is why WAR must always be put in context. What do the other stats say, not about how the player contributes to winning, but how the player contributes to entertainment? Does the player derive value from exciting plays like triples and steals or comparatively boring ones like walks? How has the player performed in clutch situations (not saying there are clutch players, but there are certainly clutch plays, and in a given season, one player might have more than his expected share)? And yes, has the player performed well under the spotlight of a pennant race or does the player toil in the obscurity of a team going nowhere?

    That’s what bothers me most about people equating WAR with value. Context matters. Even if WAR (either version) perfectly measured a player’s contribution to a team’s likelihood of winning—and it doesn’t—it shouldn’t be the final determinant of value. Winning isn’t everything.

    • invitro says:

      I don’t think anyone has said that WAR measures entertainment value. So who are you disagreeing with?

      • James says:

        Mostly the folks at Fangraphs, though there are quite a few commenters on Twitter that think WAR = value and there’s no case to be made that someone is more valuable if their WAR is significantly less.

        Don’t think it’s anyone here. Apologies, I just felt the need to rant about it.

  9. I read Foreman’s tone differently from everyone else. I did not see it as defensive, but as an attempt to curtail his impatience with Joe’s criticism, which had the superficial gloss of rationality but in the end comes down to the completely subjective judgment that the Tigers were a radically different defensive team when Verlander pitched. When people claim Jeter performed better in the clutch Joe rightfully scoffs. When people argue that Jack Morris pitched better only when it really mattered Joe rightfully scoffs. But when Joe gets enamored of a theory that requires that the 2016 Tigers magically perform better on defense every 5 games, he runs with it. If my life’s work were the target of that absurdity I’d struggle to camouflage my disdain, and I think that’s what Foreman did here, because Joe is his friend.

  10. Rudy Gamble says:

    Joe –
    Focusing on each pitcher’s BABIP and team errors to infer defensive prowess incorrectly ignores one variable from the equation: the pitcher’s impact on BABIP.

    SIERA takes the batted ball profile into account and sees a considerable difference b/w the two: 3.42 to 3.78 in favor of Verlander.

    Looking at their batted ball profiles, Verlander gave up softer contact and less line drives. The Tigers have fielded some awful defenses for Verlander and, in his peak years, he always managed a great BABIP that I would largely credit to him.

    Verlander is less defense-dependent than Porcello even after you take his higher K-rate out of the equation. I then view the team defensive adjustment as both neutralizing the defensive contribution and crediting the pitcher’s ability to limit BABIP damage. So the defensive adjustment gap between the two can be described as, “Accounting for both Porcello being the beneficiary of a better defense and Verlander’s ability to make the defense’s job easier.”


    • KHAZAD says:

      Excellent point, Rudy. The popular idea that pitchers have zero control over the type of contact is one that I completely disagree with.

      Joe trying to parse errors to tell me that Detroit happened to be a better fielding team than the Red Sox is a bit silly. The entire reason for runs saved was because of the fact that errors had little to no correlation between whether you had a good fielding team or not. Slow reacting or slow running defenders turn balls that a good defensive team gets an out on into singles and doubles, not usually errors.

      The difference between Detroit and Boston defensively was 108 runs over a season, or 2/3 of a run per game. The idea that Detroit was better during the 15 or 16% of the season that Verlander pitched and somehow more like 9 tenths of a run worse the entire rest of the time completely defies logic and really is not possible over that large of a sample size.

  11. shagster says:


    The spam issue. Why you missed the email. I had similar issues. W Client emails. So looked into it. It turns out, it may be a sender software issue. My emails were being labeled differently if sent via iPhone’s native email app.*. However if I uploaded the app for my email service to iPhone, and sent the email from within the app, then my customers reported their server correctly identified the email came from me.

    So now to make sure emails are correctly identified senders should send them from the email’s app., not send from the iPhone’s email.

    *my email accounts are set up for/plugged into iPhone’s email service.

Leave a Reply

Your email address will not be published. Required fields are marked *