A + Replacement >= B

Yesterday, Dave discussed the Jon Garland signing, largely criticizing the acquisition based on the fact that they failed to offer Randy Johnson, a superior pitcher even at this stage, a deal as lucrative. While I am not going to continue the discussion about that particular signing, the idea arose that 200 innings from Garland is less productive than 120 or so innings from Johnson should he sustain injuries. This reminded me of something Tango posted last year showing that Albert Pujols‘ numbers were almost equivalent to the production of Mark Teixeira and Jeff Francoeur combined. Whew, spelled both correctly.

All too often, injury-prone pitchers are written off as ineffective. This could not be further from the truth as certain pitchers who fit this mold are wildly productive in the time they spend on the field. Even though they fail to stay healthy enough to log 200 IP in 35 GS, they end up posting some very solid numbers.

Take a look at Ben Sheets, for starters, who produced a 2.43 FIP in 106 IP back in the 2006 season. He made 17 starts, with a 9.83 K/9 and 0.93 BB/9, good for a K/BB of 10.55. Even with a .344 BABIP and 67% LOB, Sheets still managed a 3.82 ERA. All told, his half-season produced +4.0 wins. In the same season, the aforementioned Garland produced +3.9 wins in 211 innings. Yes, Sheets was slightly more productive than Garland even though he pitched almost exactly half of the innings.

This is not the only example either. In 2007, Randy Johnson made just 10 starts, pitching in 56 innings. His 3.20 FIP and 5.54 K/BB helped him produce +1.6 wins that season. In one-fourth of the season, he produced more than 200 innings of Tom Glavine, or 170 innings of Boof Bonser. His win value also surpassed the combined output of Kip Wells, Livan Hernandez, and Scott Olsen in ~520 IP.

Granted, we never know what would have happened if the injured pitchers lasted the entire season. Still, do you really believe Johnson and Sheets would have declined so rapidly that their statistics would drop them into the average category? When pitchers with half of a season or so of statistics are evaluated, the most common reaction is to think they could not possibly be as productive as innings-eaters who stay on the field. This simply is not a universal truth. 106 IP of Ben Sheets in 2006 (+4 wins) + 105 IP of Replacement Level pitching (+0 wins) is equal to, or greater than, 211 IP of Jon Garland (+3.9 wins).

Examples like this will not always surface, but injury-prone pitchers do have value, even if that value is only seen for half of a season.

Print This Post

Eric is an accountant and statistical analyst from Philadelphia. He also covers the Phillies at Phillies Nation and can be found here on Twitter.

34 Responses to “A + Replacement >= B”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Xeifrank says:

    I still think we are working with the fallacy that Randy Johnson is better than Jon Garland. We should be looking at three year weighted values, not just what happened last year. Johnson, who is much older also should get a bump down on his projection due to age, and Garland a boost due to coming over to the NL (West). 3-Year weighted WAR for R.Johnson is 3.02 and for Garland it is 3.00.

    As for a pitcher pitching 120IP + a replacement level pitcher pitching 80 innings being equal to a 3rd pitcher who pitches 200IP. I think there are more drawbacks to the 120+80 scenario than gets pointed out here. First off, what happens if you make the playoffs, and your 120IP pitcher has already thrown his 120IPs and is on the DL. This is not an ideal situation. Secondly, not only do you have to endure 80IP of replacement level starting pitching, but you also put more stress on your bullpen over an 80IP stretch. Just something else to think about. Nice post!
    vr, Xeifrank

    Vote -1 Vote +1

    • Sky says:

      “First off, what happens if you make the playoffs, and your 120IP pitcher has already thrown his 120IPs and is on the DL. This is not an ideal situation.”

      Is starting Jon Garland in the playoffs an ideal situation?

      “Garland a boost due to coming over to the NL (West).”

      No. We know how effective Jon Garland is and already account for the league he’s in. Pitching against NL West talent doesn’t make him more valuable. It just makes his numbers look shinier.

      “3-Year weighted WAR for R.Johnson is 3.02 and for Garland it is 3.00.”

      Average the FIPs, weighted by IP and year, then apply the projected playing time.

      Vote -1 Vote +1

  2. dan says:

    Actually I think it was Sean Smith…. ***looks it up***…. I think you mean this:


    Vote -1 Vote +1

  3. John says:

    Is there any consideration in WAR for going deep into the game (ie getting through 7 IPs or more to ensure your “best” relievers only pitch) instead of only innings? The bullpen comment above prompted this, although any quantifiable advantage from this would seem to be impossible/overly difficult to calc such a small benefits.

    Vote -1 Vote +1

  4. Ed Nelson says:

    I think I agree with everyone else. Sheets has great value to those in roto leagues, but in real life your “replacement” pitcher isn’t an abstract concept he’s a real person. A real person who may get shelled in the second inning so badly he needs to be pulled for 1-2-3-4-5 relief pitchers. Which taxes your bullpen and forces you to push Max Scherzer into the 7th inning the next day when he really should have been pulled in the 6th after 105 pitches.

    In real life guys like Garland and Doug Davis can be incredibly useful by reliably keeping you in games, and pitching deep games predictably day in and day out.

    Ask the Brewer’s what most of a season of Ben Sheets gets you. They will answer Sabathia on incredibly short rest, or Seth McClung. Neither gets you to round two.

    Vote -1 Vote +1

    • Kincaid says:

      Anyone can get chased in the second inning. But most often, it won’t happen. Counting all the possible scenarios you can think of isn’t very helpful if you don’t weight how likely each is and figure out what is most likely to happen. That is no less true of real baseball than fantasy baseball. Not going after an injury prone pitcher just because you think the 6th best pitcher in your organization can’t go 2 innings is either extreme caution to the point of detriment or an indication that your organization is so bad it doesn’t matter who you sign.

      Most of a season of Ben Sheets got the Brewers their first playoff appearance since 1982. Granted, if they’d had someone like Garland instead of Sheets, they would not have had Sabathia or McClung pitching in the playoffs, but I don’t really see how that is a positive for the Brewers.

      Vote -1 Vote +1

      • Rob says:

        They didn’t get Sabathia for Sheets, they got him in addition to Sheets. Now if Gallardo didn’t go down in May they likely wouldn’t have gotten Sabathia.

        Vote -1 Vote +1

  5. Steve says:

    For teams with a strong farm system, I think these “damaged goods” pitchers are a market inefficiency. Most teams can not afford to take on the risk of counting on one of these guys to perform for the entire season. If you have a strong farm system in place it is not the end of the world because that specific teams “replacement level” happens to actually be better than replacement level.

    The Red Sox are a great example of this. They have a very strong farm system with a lot of good young arms that can fill in in a pinch. The Colon signing last year was fantastic. 1.25MM plus bonus money for 0.6WAR in 39 innings. It does not seem like a lot, but it certainly is better than 0. We shall see how the Smoltz and Penny experiments pan out.

    Vote -1 Vote +1

  6. Max says:

    I think the bigger problem in evaluating an injury-prone player going forward from our vantage point is that future injury risk can’t be measured as easily via stats.

    I’m a Mets fan and I know what 120 innings of Sheets means compared to 180 of Oliver Perez but I don’t know what Sheets’ innings will be next year. I can average out his past 3 years, or maybe his past 2 years and come up with a number, I can look at MARCEL or PECOTA projections.

    The Mets, one would think, check actual medical records and could very well have come to the conclusion that the average outcome for Sheets’ ’09 is surgery after 1-15 starts. We just don’t have access to all the facts when it comes to injuries and neither do the projection systems.

    From what I know, I want Sheets over Perez but what do I know really about Sheets’ likely IP total in ’09?

    Vote -1 Vote +1

    • Sky says:

      You would think a MLB team could come up with a pretty decent estimate.

      Or go the other way. Find the innings needed to make a guy’s demanded salary worthwhile. Then decide if it’s a good risk or not. Just because you can’t measure something very accurately doesn’t mean you should avoid one side of the decisions. Not signing guys like Sheets is as much a mistake as signing them.

      Vote -1 Vote +1

  7. OsandRoayals says:

    There should be some consideration for the added value of an innings-eater when the farmFor example if last year the orioles had an innings-eater the team probably would have avoided stunting the growth of a number of prospects.
    I think Arizona probably has the depth to be able to use johnson instead of garland, but we have to remember scherzer has very limited innings and probably a primary factor is that they have lost two of their premier bullpen arms

    Vote -1 Vote +1

  8. Eric Seidman says:

    Just to clarify, I’m not suggesting that in all cases, injury-prone pitchers are definitively more productive than innings-eaters. I’m merely pointing it out that the gap between the two is not as substantial as you may think. If you have 106 IP of Sheets in 2006 and a really awful pitcher worth -1.0 WAR, as in a win below replacement, it’s much different than 0.0 WAR, essentially a pitcher with a 5.50 FIP. Basically, do not discount the productivity of injury-prone pitchers simply because they aren’t on the field for more than 20 starts.

    Vote -1 Vote +1

  9. Joe says:

    But what about the inning Sheets didn’t pitch. Generally that is going to be a below league average pitcher taking those innings, right?

    Vote -1 Vote +1

    • Eric Seidman says:

      Joe, average is set at +2 WAR… Replacement is +0 WAR. So, the idea is that Sheets at 120 IP + Replacement Level at 80 IP is not terribly worse than someone like Garland or Doug Davis at 200 IP.

      Vote -1 Vote +1

      • sabernar says:

        But is that a safe assumption? When the Yankees’ starters went down last year they filled in with Sidney Ponson and Ian Kennedy, et al. I think it’s safe to say that they were probably below replacement value (though I could be wrong about this). Was Wang + Ponson + Kennedy (or whatever the combination is) better than, say, Garland? (Not that I’m advocating switching out Wang for Garland, I’m just using this as an example.)

        Vote -1 Vote +1

  10. AM says:

    This article reinforces what I’ve said about the unrealistic hope that the Twins (with payroll flexibility) should sign Sheets.

    They have 5 starting pitchers already. Their #4, Blackburn, is a bit above average, and their #5, Perkins, is below average. But both have proven they can maintain a rotation spot with at least some form of adequacy.

    Signing an injury risk doesn’t make sense if you are counting on the pitcher to be a piece of the rotation; causes short rest, taxes bullpen etc. But it makes sense if you have passable alternatives to sub in.

    Sheets would be a big improvement for the Twins rotation, and, if is likely, he is injured, the Twins would have Perkins to fill in.

    Oh, and Sheets would add about 2.5-3 wins.

    Vote -1 Vote +1

  11. Eric Seidman says:

    Sabernar, check this out:

    Garland in 2008 = +1.9 Wins

    Wang in 2008 = +2.0 Wins
    Ponson w/Yanks = +0.2 wins
    Kennedy in 2008 = +0.3 wins

    So, injured Wang + 2 virtual replacement level pitchers still exceeded healthy garland by +0.4 wins.

    Vote -1 Vote +1

    • Xeifrank says:

      Garland’s 3 year weighted average is 3.00 WAR. imo, you need to look at three year weighted average and adjust for age/change of league/park effects.
      vr, Xei

      Vote -1 Vote +1

      • Sky says:

        Again, change of league and park are irrelevant, except to the degree that Garland’s skills can be leveraged by a park that “fits him well”. But puutting a low-K pitcher in a parks that rewards balls-in-play doesn’t seem smart, though. And the DBs’ defense was poor last year, although they could obviously make improvements for 2009.

        Vote -1 Vote +1

      • Xeifrank says:

        Sky, if WAR is going to be used to make an argument, then “yes” you do need to take into account what league Garland and RJ pitched in. Garland gets up to a 0.5 win jump from switching leagues.
        vr, Xei

        Vote -1 Vote +1

      • Sky says:

        With all due respect, Xei, that’s wrong.

        The league-adjustment is applied to past performance to put all pitchers (both in the AL and NL) on the same scale. After that adjustment, a 3.0 WAR pitcher in the NL showed the same talent and had the same value as a 3.0 WAR pitcher in the AL.

        Going forward, that 3.0 WAR pitcher might post a 3.75 ERA as a Giant or a 4.50 ERA as a Ranger, but he’s still the same pitcher with the same value. (Numbers for example only.)

        Vote -1 Vote +1

      • Xeifrank says:

        Sky, if the WAR quoted above was league neutralized then yes you are right. I assumed they were not. It was not specified. Secondly, there has been no response to any of my earlier comments regarding cherry-picking 2008 stats to compare Garland and RJ. Don’t you think a three year weighted average with age adjustments should be used when projecting future performance, instead of just what happened in 2008? Three year 5/4/3 weighted WAR for Garland is 3.00 and for RJ it’s 3.02. Going forward with age adjustments, one would think the 40+ year old Johnson would be more likely to get worse. Also, if you calculate WAR off of 2009 Marcels for the two players and adjust for league, Garland has more value.
        vr, Xei

        Vote -1 Vote +1

      • Sky says:

        I definitely agree a real projected is better than 2008 stats.

        Vote -1 Vote +1

  12. Ed Nelson says:

    That evaluation of these pitchers performances does not accurately describe the cascading effect that Wang’s injury had on the Yankees as a team. There are a significant number of other factors that go into making a team successful that don’t show up in Ks, BB, innings ect. For example, Ian Kennedy failed to reach the 5 inning in 6 of his 9 starts. He failed to get out of the third three times. What effect does that have on the Yankee bullpen? What effect did that have on Kennedy’s development? The idea that Kennedy was worth +.3 wins is silly. His effect on the team was far more detrimental than could ever be evaluated by just looking at his numbers alone.

    Vote -1 Vote +1

  13. Matt Harms says:

    I think others have touched on this, bit you can’t just look at Sheets’ production vs a replacement level pitcher. Sheets’ injury not only removes his production, but also inserts the production of his injury replacement. Often, that player will be below replacement value as the team’s long reliever, spot starter, or AAAA veteran.

    I don’t have the tools now, but it would be interesting to see the Sheets+Fill-In Players’s win production vs Garland.

    Vote -1 Vote +1

    • Sky says:

      “Sheets’ injury not only removes his production, but also inserts the production of his injury replacement. Often, that player will be below replacement value as the team’s long reliever, spot starter, or AAAA veteran.”

      This is only true of a poorly run franchise. Replacement-level is definitely for just these purposes — the expected performance of the guy you can find easily for the league-minimum.

      Replacement-level IS the fill-in level.

      Vote -1 Vote +1

  14. Ed Nelson says:

    That’s pretty easy actually. If you look at 2007 (that’s the type of Sheets people are worried about) all of Sheets and Yovanni Gallardo’s starts are made by 3 pitchers. Claudio Vargas, Carlos Villanueva, and Manny Parra

    Parra 2 starts 9 innings 9 hits 5 walks 4 ER and a broken hand in the last start.

    Carlos Villanueva 6 starts 35 innings 29 hits 17 walks 8 earned runs (not bad)

    The problem here was the heavy reliance on Claudio Vargas to cover starts for Gallardo and Sheets

    Vargas 124.2 innings 71 earned runs (75 runs) 52 walks 143 hits 5.13 ERA after his August 24th start when he was demoted to the bullpen. His biggest problem was his inability to work deeper into games failing to reach the 6th inning in 13 of his 23 games started and failed to reach the 4th 5 times. Wonder why the Brewers weren’t in the playoffs. There you go.

    Which leaves me to my last point. When Ben Sheets is on your team you had better have a very good 6th starter, and everybody else better stay healthy because there’s no margin for error.

    Garland’s 2007 208 innings 4.23 ERA. Combined ERA of Sheets/Vargas 4.42 (that’s a little dirty I know) but remember Garland got into the 7th inning or better in 26 of his 32 starts, and only failed to get into the 6th inning 4 times (including giving up 11 earned in 3 and third innings against the Twins July 6th. Ouch!) I think that has great value.

    Vote -1 Vote +1

  15. CJ says:

    At Johnson’s age, the probability that his performance can fall off a cliff at any time is real…it’s a material risk. For that reason, I can understand the Diamondbacks decision to look elsewhere. I might have done it differently and taken the risk, but this isn’t an irrational decision by the Diamondbacks.

    I like Ben Sheets; I agree he has significant value even if he isn’t likely to pitch a lot of innings. And I think he is ideal for a team which needs a No. 2 starter to go with a proven ace. But the Diamondbacks have great pitching in their No. 1 and 2 slots; I can understand the ideal that they want a bottom of the rotation pitcher to soak up innings and save the bullpen. I also think that from a GM’s perspective, there is some disruption costs to using a replacement pitcher for a good part of the year, often in sporadic spurts. Setting aside how good or bad your replacement level pitcher might be, just the sheer act of moving players back and forth between the farm team and the big league team is disruptive. Sometimes other worthwhile players have to clear waivers and be sent down in order to call up the new pitcher. Sometimes players get stuck on the ML roster and can’t be sent down because another team made a waiver claim and the player was pulled back…I have seen that happen with some frequency. How much is it worth to avoid this disruption? I don’t know, but the value isn’t zero. It could make the difference in some cases where the gap between the injury prone candidate and the innings eater is small enough.

    Vote -1 Vote +1

  16. Sky says:

    For those interested…

    When the DBs traded for Randy from the Yankees after the ’06 season, he was coming off a 205 IP season with an ERA of 5.00 but a FIP of 4.27. The trade required an extension for Randy, costing the DBs $26MM for two years of service (between salary and signing bonuses according to Cot’s). 2007 was going to be his age 43 season. Obviously, he was hurt most of the year, but managed to post a good ERA and even better FIP in about 50 innings. In 2008, he pitched 184 innings with both an ERA and FIP in the high 3.00s.

    The DBs tend to be a smart franchise, and while they recognized that Randy’s 2006 ERA was too high for skills, they had to realize his age was a risk. Maybe being two years older and having a defined injury two years ago is enough to scare them away. They don’t necessarily don’t want him, but not at the price he was asking. While the Giants’ deal is a base of $8MM, he has something like $4 to $5MM in playing time and performance bonuses. So if he’s bad, the Giants are on the hook for $8MM. If he’s really good again, they’re on the hook for $13MM. It’s not an awesome deal, but defensible.

    Vote -1 Vote +1

    • OsandRoayals says:

      What MLB.com is reporting is that the Diamondbacks found some money after they realized they wouldn’t have so many high draft picks to pay for. It could be and probably is spin to some extent but there should be some truth to it. Not too many baseball teams go out of their way to say that they made a mistake and could have kept their franchise icon

      Vote -1 Vote +1

    • Eric Seidman says:

      Just to discuss some of what was said here: nobody is debating that a 3-yr weighted projection would be the best way to compare two players. That’s not the point of this post, though. The point of the post is that injury-prone pitchers have value even when they miss half of the season. The entire point of the replacement level is for when pitchers go down with injury or miss time for other specified reasons. The injured pitchers, even in just 17-20 starts, plus the replacement level CAN exceed the production of other healthy pitchers. That is the point of this post, not whether or not Johnson > Garland. The Johnson vs. Garland debate is what gave me this idea, though.

      Vote -1 Vote +1

Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>