Shutdowns and Meltdowns Should Kill the Save

On Friday, Jonah Keri penned an impassioned plea to kill the save statistic.

If you’re an Indians fan still rattled by Thursday’s disaster, or just a baseball fan sick of suboptimal decisions resulting in painful losses, the man to blame is a late Chicago sportswriter named Jerome Holtzman. Fifty-three years ago, in an effort to shift more recognition unto underappreciated relief pitchers, Holtzman invented the save statistic. Today, that invention is responsible for far more unintended consequences, and far more heartache, than Holtzman could have ever intended. Bloody battles are fought over the ill-begotten riches that saves bestow on those who can get them. Managers lose games for their teams by getting seduced by saves. Pitchers who fail in save situations get labeled as gutless pariahs.

It needs to end now. It’s time to kill the save, send it to hell, and strand it there for eternity.

In case you missed the subtlety, Jonah’s not a big fan of the save, or more accurately, not a big fan of the repercussions that valuing saves has had on modern day bullpen usage. As the save has grown in popularity, so has the rigidity of bullpen roles, and reliever usage has been modified to specifically fit the definition of the save that was created by Holtzman. It is one of the few instances in sports where obtaining a statistic, rather than maximizing a team’s chances of winning, actually drives how teams utilize their players.

However, as Jonah notes, the current system is so entrenched in the economics of today’s game that it isn’t as simple as just convincing teams to stop managing in a certain style. Reliever compensation is driven very heavily by the amount of saves a pitcher can compile, and any attempt to move away from the save as a measure of value could be viewed by skeptical players as simply an effort to drive down costs. If Major League Baseball is ever going to move away from the save as a driving force of reliever valuation, it has be replaced by a system that would be acceptable and understandable by the relievers themselves, and teams would have to make a concerted effort to explain why a redistribution of funds without a reduction in total expenditures on relievers would actually be an improvement for relievers as a whole.

In order to get that kind of wide scale acceptance, the metric replacing saves would have to be approachable and easily explained, and also line up with what relievers understand their job to be – come into close games and preserve a lead or keep the deficit to a minimum. Any metric that is context neutral will never be accepted by those within the game, as reliever performance is inextricably tied to the situation into which a pitcher is placed. Any stat that seeks to displace the save has to take the inning, score, and base/out situation into consideration.

That leads us to WPA, but getting teams and players on board with using a stat that gives them credit for +0.24 wins in a given performance isn’t going to be an easy transition. So, what’s needed is a metric that accounts for leverage-specific performance and preferably looks like saves and blown saves, so it can be offered up as an easy-to-understand improvement on an outdated system.

Enter Shutdowns and Meltdowns. Quoting Jonah again:

This might sound a bit complicated, but it really isn’t. By using 6 percent as the cutoff, you get a stat that runs on a similar scale to saves and holds. Elite closers and setup men will rack up 35-40 (or more) shutdowns and very few meltdowns, just as a dominant closer can earn that many saves, while blowing very few. If you’ve ever watched poker on TV, you’ll see a player’s odds of winning a hand rise or fall by a certain percentage based on the cards the dealer flips over. Same easy-to-follow concept here: If you retire the side 1-2-3 in a big spot (say, two runners on, none out, and you enter with the game tied in the seventh), you get a shutdown, just as hitting your nut flush on the river will usually win you a hand. The only difference is the pitcher has more control over the outcome in this case, rather than it being left to random chance.

The key is that SD/MD puts closers and other members of your bullpen on even ground. That way you don’t end up overpaying for a pitcher who happens to record the final out of a ballgame. Greatness is greatness, and it gets rewarded whenever it might occur during the course of a game. We know that at least one relief pitcher has adopted SD/MD to track his own performance: Daniel Bard kept close tabs on the stat before getting converted back to a starter’s role this offseason.

The big barrier to acceptance is going to be getting players on board with being evaluated by something other than saves, and we know that at least one Major League pitcher has already bought into the value of SD/MD. Naturally, it’s a reliever who was dramatically undervalued by the current system, and found that SD/MD better represented how valuable he actually was to his team. Getting good setup men and middle relievers on board will obviously be easier than getting closers on board, since a shift in reliever valuation will move money from ninth inning guys to seventh and eighth inning guys.

Still, that redistribution is something that the majority players should be in favor of. Very few players come up as closers, and most elite relievers spent a few years toiling as a setup man before getting promoted to the big chair. In order to become a closer, they had to perform very well in that earlier inning role, and they generally received little to no compensation for those performances. By moving to a system that paid relievers based on Shutdowns and Metldowns, relievers would become more highly valued earlier in their careers, and the shift to receiving real paychecks earlier would transfer money to guys who have previously been underpaid and then gone down with severe injuries before they could ever cash in on their success.

In reality, many of the best relievers in the sport don’t last six years, and if they don’t become a closer early in their career, their arbitration payouts are limited due to their lack of saves. It’s completely possible for a high quality reliever to have a good five year run and never really get rewarded financially for that success. For instance, Mike Adams has recorded 103 Shutdowns and only 24 Meltdowns over the last six years, and his 81% SD/MD rate is eighth best in baseball since 2006, ranking between Francisco Rodriguez and Rafael Soriano on the leaderboard. Despite that, Adams will make $4.4 million this year after his third trip through arbitration, and he’s garnered a grand total of just over $9 million in his career. Soriano made $18 million before he became a free agent, while Rodriguez made just over $21 million in those same six years.

By replacing saves and holds with Shutdowns and Meltdowns, Major League teams could more accurately reward those players who actually perform well out of the bullpen, and could move towards a system where managers weren’t beholden to an outdated statistic. By getting rid of the save as a valuation metric, teams could more efficiently utilize their best relievers while also still ensuring that they were financially rewarded for pitching well in critical situations. The binary nature of SD/MD will appeal to players who see their job as “success or fail” based on how their performance directly affects the scoreboard, and the fact that it scales very similarly to saves can help it receive acceptance that other measures never will.

Jonah’s absolutely right – Major League Baseball should begin taking steps to kill the save statistic, but they can only do that if they have a worthy successor in place that Major League players will adopt as a replacement. Shutdowns and Meltdowns can be that successor. We don’t have to be slaves to the save anymore.




Print This Post



Dave is a co-founder of USSMariner.com and contributes to the Wall Street Journal.

141 Responses to “Shutdowns and Meltdowns Should Kill the Save”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. jcj5y says:

    I’ve also seen Tom Boswell quote Tyler Clippard as someone who was well aware of his high WPA, presumably because he needs an alternative way to measure his considerable value. (Clippard was second in the league in WPA for pitchers last year, and third in the league with 40 shut downs. Interestingly, Nats closer Drew Storen also had 40 shut downs.)

    Vote -1 Vote +1

  2. nittany87 says:

    Agree 100% with the basic premise.

    The fight (as always) is educating the general public. Will be curious to see how hard it will be to gain traction.

    +6 Vote -1 Vote +1

    • Ty says:

      I think the bigger problem is getting commentators to adopt the system. If fans tune in to games and hear about SM/MD rather than saves, I imagine most people would catch on pretty quickly. In the past decade-ish I recall concepts like WHIP and OPS being accepted pretty quickly as soon as broadcasts started talking about them.

      Vote -1 Vote +1

      • vivalajeter says:

        I wouldn’t be surprised if it’s mentioned this week during a Mets broadcast. Their announcers seem to keep up to date with stuff like this.

        Vote -1 Vote +1

      • Isolationist says:

        I’m sure the word “meltdown” will be mentioned during the Mets broadcast

        +45 Vote -1 Vote +1

      • Llewdor says:

        I remember a few years ago a Mariners broadcater (Ron Fairly, I thnk) pulled out the Game Winning RBI stat, one that even MLB retired years earlier.

        The commentators are the key.

        Vote -1 Vote +1

    • JimNYC says:

      This will never, ever happen — at least not any time soon.

      In order for stats to be mainstream, they have to be easily explainable to people. This will never be easily explainable to people.

      As it is, saves are too complicated for most fans — I’d be willing to bet that more than half of all baseball fans don’t know the other two ways to get a save (aside from pitching the full ninth inning with a three run or less lead). You mention the words “win probability” to most fans and their eyes will immediately glaze over and they’ll stop listening to you.

      Saves are a stupid statistic, but people want some way to evaluate relief pitchers, and you’re not going to be able to sell the populous on one that requires more math than saves do.

      Vote -1 Vote +1

  3. The Kid says:

    Yea, cause getting the final 3 outs of a close baseball game is pretty irrelevant anyway. No pressure whatsoever.

    “Hey let’s use Mariano Rivera with runners on 1st and 2nd in the 7th inning with two outs. I’m pretty sure Cory Wade or Luis Ayala could get the final 3 outs of a 1 run game with no problems.”

    -62 Vote -1 Vote +1

    • Eric Dykstra says:

      I… can’t tell if this is sarcasm or not.

      +33 Vote -1 Vote +1

    • ThundaPC says:

      The sad part is that “get the final 3 outs” isn’t even the only way to record a save.

      Vote -1 Vote +1

    • termain says:

      Perhaps you haven’t heard of Leverage Index. The leverage index for the first situation you described is about 3.2 for a one run game. The leverage index for the second situation is 2.9.

      It is (on average) a good use of Rivera to have him pitch in the first situation. And it’s not like you can’t have him pitch for a couple outs in the 8th.

      +10 Vote -1 Vote +1

    • Bip says:

      The thing you have in quotes seems reasonable, especially when you phrase it as which situation to use the lesser reliever in. Would you rather make Cory Wade try to diffuse a situation with two men on base or have him start a fresh inning? What about the ninth inning makes it that much harder to pitch in than previous one? Plus, in that situation, I’d probably use Mo in the seventh and if it goes smoothly, I’ll leave him in for the eighth so he’s not being wasted on a single batter.

      Vote -1 Vote +1

      • uclabennett says:

        “What about the ninth inning makes it that much harder to pitch in than [the] previous one?”

        Clearly you haven’t ever pitched in the 9th inning of a baseball game before.

        -10 Vote -1 Vote +1

    • Mark says:

      On the other hand, you don’t get to use Mariano Rivera in the 9th inning if Cory Wade, Luis Ayala or whoever the hell else Girardi uses in the 7th inning blows the lead.

      Cause then Rivera has no lead to save. I’d rather have Cory Wade pitching the 9th with the lead then have Rivera pitching the ninth while the team is trailing.

      Vote -1 Vote +1

      • Bob says:

        The question no one has asked is who is at bat? If it’s Ellsbury at the top of the Red Sox lineup, Rivera could make a lot of sense. If it’s Yuneski Bettancourt, Rivera is probably overkill.

        Vote -1 Vote +1

      • jim says:

        if it’s yuni betancourt coming up, ryan madson 1 month off TJ surgery is probably overkill

        +12 Vote -1 Vote +1

  4. jj says:

    I wondering if MLB likes the fact that it can get by with using the Save stat and making it easier to keep 5 or 6 RP getting paid a relatively lower figure. Having this one somewhat arbitrary stat may actually be helping them keep total bullpen expenses lower. So overpaying for one guy (the closer) may seem silly, but if the owners have a system for keeping the rest of the bullpen prices low, then it seems like maybe this isn’t as silly as first glance makes us think. Also there is a benefit of signing the big name closer in that it makes you look like you are trying to field the best team possible and excites the fans (Heath Bell to MIA).

    +7 Vote -1 Vote +1

    • Rangerjoe says:

      This. So much this. Smart teams know the disparity between “name” closers and the often unknown “firemen.” With the rangers last year, Feliz got the saves, Adams was the true shutdown man. In fact, this year they have several good shutdown guys and one name close making more than all of them (and who is now less skilled). I’m sure Jon Daniels knows this for example; but it does keep the total cost down. A market inefficiency if you will.

      Vote -1 Vote +1

    • Bip says:

      Any reliever in the game today could get 20 saves over a full season. I heard there’s a strategy where you turn a totally mediocre reliever into a “closer” by putting him in low pressure save situations like where there’s a 2 or 3 run lead, he racks up 20-30, and now you have a trade chip that will be seen as way more valuable than it actually is.

      Vote -1 Vote +1

      • DD says:

        Matt Capps, who returned Wilson Ramos. If Capps wasn’t collecting saves, he’d have returned a PTBNL.

        +12 Vote -1 Vote +1

      • vivalajeter says:

        Not sure where you heard that Bip, but I don’t know of any team that has one closer for easy saves (to increase trade value) and another closer for tough saves. There might be teams that use lesser relievers to increase their trade value (like Capps a couple years ago, or Broxton this year), but they don’t just hand them easy saves.

        Vote -1 Vote +1

      • baycommuter says:

        Sure, Billy Beane played that game with Bailey. He took the No. 38 prospect in the A’s system, converts him to a reliever (closer), and after three cheap years sends him to Boston.

        Vote -1 Vote +1

      • uh says:

        Hello Mark Melancon.

        Vote -1 Vote +1

      • CircleChange11 says:

        Billy Koch

        Vote -1 Vote +1

      • Greg says:

        Whether you think the Red Sox overpaid for Melancon and Bailey, they were the best relievers on their respective teams last year by WAR.

        Again by SD/MD ratio, no one on the Astros was really comparable to Melancon last year. By that measure, Balfour was a little bit better than Bailey last year but if you look at the preceding years, Bailey outperformed Balfour. The job was Bailey’s to lose and he didn’t perform such that he should have been replaced. The difference was small enough again last year that had he stayed with the As, he still would have been expected to be the best reliever by any measure but total IP, especially given his superior traditional stats.

        So based on the above, the notion that Bailey and Melancon were used as closers to inflate their trade value is ludicrous.

        Vote -1 Vote +1

      • cable fixer says:

        ^isn’t that a bit of a false dichotomy? it can’t be both that: 1. they were their best reliever on the team and 2. using them in the CL role inflates their perceived value?

        Vote -1 Vote +1

    • Max says:

      Well, the first teams to do this would gain a significant advantage, as they could obtain the best relievers that don’t get any saves for a similar amount that they are getting now, while other team’s scramble for people with a ninth inning mentality.

      Vote -1 Vote +1

  5. dbake005 says:

    So you’d rather Cory Wade go in and give up 3 runs in the seventh inning, making Mo Rivera irrelevant for that night?

    Vote -1 Vote +1

    • Jason says:

      So, you’d rather make Mo Rivera irrelevant for tomorrow’s game by having him pitch in a losing cause?

      Vote -1 Vote +1

      • wealthy_waffles says:

        But you don’t know if you’re going to be in a position to win or use Mariano Rivera tomorrow night. You DO know that you need to use him in this one to have a shot at winning today. You have to try and win when given the chance

        Vote -1 Vote +1

      • Jason says:

        right, you do not know it, so the expectation is equal to the frequency with which you need Mariano over the course of the season (a pretty good likelihood you will need Mariano in the next game).

        Also, you do not NEED to use Mariano to win that game. He might help you get through that inning marginally better than your next best reliever. However, it is not definitely the case that that marginal difference exceeds his value later that game plus the value of saving him for tomorrow if you are destined to lose. We don’t know the answers to these questions.

        -7 Vote -1 Vote +1

  6. glenstein says:

    Awesome. I would just think twice about making +/- 6% as a cutoff point- SD/MDs should exist on their own terms in counted in whatever way best encapsulates the value of the performance being described.

    Vote -1 Vote +1

  7. Is FanGraphs going to start keeping track of shutdowns and meltdowns on the players’ stat pages?

    Vote -1 Vote +1

  8. glenstein says:

    Another issue: people more or less understand what a save is as they’re watching the game. Do people clearly understand which pitching situations involve 6% shifts in WPA?

    +33 Vote -1 Vote +1

    • gonfalon says:

      +1

      Vote -1 Vote +1

    • Will says:

      This here is the biggest problem.

      Since it’s a counting stat it needs to be easily countable. There are obvious situations where a reliever is pitching for a SD/MD (ie bases loaded, bottom of the 9th, no outs, up by one run), but there are many more scenarios where it’s not apparent. How about bottom of the 9th, bases loaded, two outs, up by 3 runs. Would that qualify? How about bases loaded, no outs, bottom of the 3rd, up by one run?

      You wouldn’t know if it was a SD/MD until you got on FanGraphs to check the WPA of that particular situation. And that’s the sole reason why it will not catch on. But I applaud your efforts to do away with the save.

      +12 Vote -1 Vote +1

      • Zac says:

        It’s really not that complicated once you look at it. For starters, you really need to not allow any runs to have a shot at a shutdown. 96.3% of shutdowns in 2011 were from relievers who did not allow a run. Another 2% of shutdowns were pitchers who allowed a run, but finished the game without relinquishing the lead (i.e. coming into the game with a 2 run lead in the 9th, allowed 1 run). The majority of the remaining 66 pitchers had their run score after they had been taken out of the game, meaning that they deserve only some of the blame for the run scored.

        So if the pitcher allows 0 runs, the question of whether it’s a shutdown or not just depends on how much they pitch, and the score of the game while they pitch (and runs left on base when they come into / go out of the game). So we can just look at some likely scenarios depending on the type of pitcher and their role on the relief staff. Now I’m not taking part factors or the strength of the offense the pitcher is facing, which affects the run environment and slightly changes the WPA. But it’s not by much.

        For a traditional closer or setup man, who pitches one (and only one) of the last two innings, the only change you need to make with them is to get them to think about 0, 1, or 2 run lead instead of 1, 2, or 3 run lead. A pitcher who pitches the entire 9th inning (or in extra innings) with a 0, 1, or 2 run lead will get a shutdown as long as they don’t give up the lead. A pitcher who does the same in the 8th inning will get a shutdown as long as they don’t allow a run. 60% of shutdowns in 2011 were from relievers who recorded exactly 3 outs during their relief visit.

        22.5% of shutdowns are relievers whose team records between 4 and 6 outs while they are pitching. Pitching that length of time while your team has anything between a 2 run deficit and a 3 run lead is probably enough for a shutdown.

        4% of shutdowns are relievers who pitch more than 2 innings. With few exceptions, you’re still looking at the same relative closeness as the reliever who pitches two innings. Charlie Furbush came into the game in the 4th inning on May 27, 2011, with his team down 4 runs. He pitched 5 shutout innings (pitching through the 8th), and left the game with his team still down 4 runs and got .074 WPA. So with 5 IP with his team down 4 runs, he got enough for a shutdown, but not by much.

        The last 13.5% of shutdowns in 2011 wee pitchers who pitched less than a full inning. Obviously this is a pretty high leverage situation. In the 9th inning, if you come to the mound with the tieing run at the plate, a single out is enough for a shutdown. In the 8th inning, if you come to the mound with the tieing run on base a single out is also worth a shutdown.

        Vote -1 Vote +1

      • Will says:

        Zac, thanks for the explanation. It’s very interesting.

        But I think you’ve made mine and many others’ point. It’s not at all easy to determine in real time whether it is a SD situation.

        And that’s only for shutdowns. It’s a whole different story calculating meltdowns.

        Vote -1 Vote +1

      • Greg says:

        Not that I think it would happen, but they could always start showing the win expectancy during broadcasts. For single inning stints at least, that would make it pretty easy to track shutdowns and meltdowns.

        Vote -1 Vote +1

    • chuckb says:

      Probably not but I think that most viewers can recognize tough situations. Any viewer will recognize that 2nd and 3rd w/ no one out in the 7th is a tough situation, so calling it a “shutdown” when the reliever works his way out of it won’t be difficult to comprehend. Or, conversely, calling it a “meltdown” when someone comes in up 1 w/ 2 men on and gives up the 3 run homer is pretty easy to understand as well.

      You don’t need to be well-versed in WPA or leverage index to understand that some situations have a much greater impact on the game than others do.

      Vote -1 Vote +1

    • atigersgrin says:

      Also when a closer comes in for a save there is very little chance they will leave before the game is over or they have blown the save. Right now managers and relievers don’t worry about shutdowns, so pitchers get pulled when they were in line for a shutdown. Will managers be pressured into keeping guys on the mound to record a shutdown?

      Vote -1 Vote +1

    • BIP says:

      This argument is really more applicable to fantasy players. I strongly perfer wOBA and UZR and FIP to evalutate players, but I certainly can’t tell you what a player’s value in these statistics will be based on a game’s performance. It’s a reason I haven’t yet done a fantasy league that uses these stats.

      Vote -1 Vote +1

    • Herbstr8t says:

      They will start to understand if it is commonly reported and tracked. As long as SD/MD intuitively match what people see it will make sense to them. Shutdown and meltdown are terms you already here at the ballpark, or at the bar, if the stats reflect what people are talking about that can only be an improvement.

      I’ve been chasing holds in my fantasy league for 5 yrs now, and I still can’t tell remember all the nuances of what does and doesn’t count as a hold.

      Vote -1 Vote +1

  9. JT says:

    Great article. I would take it a step further and do away with giving wins and losses to the pitchers too. The stats for pitchers should be quality start (QS) vs non-quality start (NQS) and quality relief appearances (QR) vs non-quality relief (NQR).

    Currently a reliever who blows a save can get credit for win, though the original starter (who may have pitched 8 innings of shut-out baseball gets credit for nothing). In the new scenario the starting pitcher would get credit for the QS and the reliever who blew the save would get a NQR.

    Vote -1 Vote +1

  10. dbake005 says:

    What are your QS and NQS qualifications?

    Vote -1 Vote +1

  11. gonfalon says:

    Personally, I’d rather keep the save, although I agree in spirit with the idea that your best relievers should be used in high-leverage situations, whether there is a save situation or not. over at Buc’s Dugout, there is a good article comparing the usage of Joel Hanrahan and Jonathan Papelbon last weekend:

    http://www.bucsdugout.com/2012/4/9/2935429/clint-hurdle-pirates-phillies-dodgers-charlie-manuel

    of course, this isn’t to say Manuel is an idiot, or Hurdle is a genius — over the course of a year, all managers make their share of smart and dumb decisions at different times. but maybe we just need smarter managers overall.

    regarding the idea that good middle relievers are undercompensated, I’m not sure that’s actually true for free agents (such as Rafael Soriano and Frankie Rodriguez), especially now that draft pick compensation has been overhauled, or for pre-arbitration eligible players (who all play for near league minimum). and for arbitration eligible players, the poor middle relievers get non tendered (like no-hit utility infielders or backup outfielders), while the good ones get a contract. so I don’t think there is a problem, compensation-wise… of course, YMMV.

    Vote -1 Vote +1

    • gonfalon says:

      one other note is that Rafael Soriano and Frankie Rodriguez are poor comps for Mike Adams, as he has really never been used as a closer. Soriano and Rodriguez are currently signed to expensive free agent contracts as middle relievers, but during their first 6 seasons, Soriano was a closer for two years, while Rodriguez was an elite closer for 4 of his pre-free agency seasons.

      in addition, Rodriguez was healthy for all of his pre-free agency seasons, whereas Adams missed a year, apparently due to injury.

      Vote -1 Vote +1

    • Fatalotti says:

      Good article. I’m not sure if you read all of it, but the point of the article (from a Pirates perspective) was that Manuel definitely was an idiot with his bullpen usage, and the writer was pointing out that Hurdle did the exact same crap last year. I suppose he’s hoping that Hurdle seeing it up close and personal will allow him to not make the same mistake again this year.

      Once again, using a weaker pitcher to pitch in a situation (tie game on the road, 9th inning or later), where there is absolutely no room for error, while saving a much better pitcher for a later point in the game when your team has a lead and, by definition, an actual margin for error, is just terrible tactical managing, and a direct result of the influence of the save statistic.

      Vote -1 Vote +1

      • PiratesHurdles says:

        This.

        Hurdle had an easy choice thanks to the save nonsense not applying to the home team, but he did the exact same thing as Manuel did (let the closer sit in a tie) on the road last year as the season went down in flames around him.

        Die Save Die.

        Vote -1 Vote +1

      • gonfalon says:

        FWIW, I did catch the writer’s opinion of Manual’s bullpen usage last weekend, and Hurdle’s bullpen usage last year, and agree completely. I was just trying to be ‘diplomatic’, for lack of a better word.

        Vote -1 Vote +1

  12. Corey says:

    Couldn’t a rogue arbitrator start instituting this on his own? Also, what is WPA? I’ve never heard of it.

    Vote -1 Vote +1

    • Corey says:

      Win Probability Added, got it, my brain hates acronyms.

      Vote -1 Vote +1

    • DD says:

      Arbitrators are only allowed to look at certain information (certain pitching/hitting stats, comparable players’ pay, the player’s historical pay, and awards)

      Vote -1 Vote +1

  13. Table says:

    YES YES YES. I’ve been agonizing about this very thing lately. Please let this happen in my lifetime, preferably in the next 10-20 years

    Vote -1 Vote +1

  14. sprot says:

    “In order to get that kind of wide scale acceptance, the metric replacing saves would have to be approachable and easily explained…”

    Which is why it will never ever ever be SD/MD. Ever.

    +16 Vote -1 Vote +1

    • Kevin says:

      That was my initial reaction, but after Zac’s explanation above, I think it might work. A shutdown, for late relievers at least, is basically just a “save” except with a 0, +1, +2 run requirement rather than +1, +2, +3 requirement, and a lack of a need for it to finish the game. The accompanying explanation: the 7th and 8th innings matter just as much. A meltdown is a failure in those same situations. This shouldn’t be too hard to accept, especially with some broadcaster explanations. Indeed, it might need some simplification for broader acceptance, but for one, think it’s a good starting point.

      Vote -1 Vote +1

    • bstar says:

      But how would that work? Boog Sciambi will be the only commentator who understands it. How will it seep down into the general public’s awareness if the commentators don’t get it?

      Vote -1 Vote +1

  15. poland4evr says:

    If the metric changes to more accurately financially reward relief pitchers what ends up happening is that market efficiency becomes a default standard and the value of exploiting market inefficiencies by smart clubs and FO’s is thereby further circumscribed. Sounds like a poor outcome to me.

    Vote -1 Vote +1

    • Eminor3rd says:

      So you favor unnecessary and artificial imbalance in baseball?

      Vote -1 Vote +1

      • poland4evr says:

        I just don’t see any problem with its presence. In the same way, it doesn’t offend me when inefficient contracts are given out or inefficient strategies are employed on the field. There is no need for baseball to be rational, objective, or balanced.

        Vote -1 Vote +1

  16. CaptWBligh says:

    One problem I have with SD and MD counting: frequently, the reliever is only really eligible for one or the other, and the manager controls that. For instance, the mop up guy who only ever comes in with the team up by 3 runs won’t ever get a Shutdown. And the mop up guy who comes in with the team down by 3 won’t ever get a Meltdown.

    Now, your standard save situation or close game can usually go one way or the other. But the hypothetical mop-up guy above could get 50 appearances, blow 5 of them and be lights-out in the other 45, and end up with 5 MDs and 2 SDs on the year…

    Vote -1 Vote +1

  17. chuckb says:

    I think those of you who are pessimistic toward these statistics’ adoption aren’t giving people enough credit. A few years ago, stats such as OBP and OPS weren’t commonly used or recognized and now they are. Most people understand ERA even though they’d fail if you asked them on the spot how to calculate it. Some of the rules about saves are a bit obtuse (tying run in the on-deck circle, for example) but people understand that stat. Most people understand what a hold is, despite its ridiculousness. It’ll take some time, to be sure, but I think that people will eventually get it as it becomes more widely used. WAR is much more complicated but it’s even becoming fairly widely used.

    Vote -1 Vote +1

    • Indians fan in SF says:

      Exactly. Many of the stats that those of us who read this site dislike will continue to be used – perhaps forever. But that doesn’t mean it’s keeping us from looking at things in a more meaningful way and using better stats and measures of a player’s contribution to a win or loss. Recent Cy Young voting is showing that pitchers’ W/L records are no longer the final word in a pitcher’s quality. As for Keri’s story, I actually don’t at all think the Indians are guilty at all of managing to the Save stat. It’s exactly the reason that they haven’t had a big name (or even decent!) closer for at least a few years. The front office knows that it makes no sense to pay for saves. That’s why, I’d much rather continue to see Pestano going in in high leverage situations – and think that that’s exactly what we will see. Case in point – compare Joe Borowski’s 1.42 WPA to Rafael Betancourt’s 5.40 WPA in the not-quite-pennant year of 2007. Borowski had 45 saves, but also had a 4.12 FIP (and xFIP). The Tribe knew exactly what it was doing I think. If anything, it’s the media and fanbase that I think ends up pushing a team into assigning the role and misusing their best relievers.

      Vote -1 Vote +1

    • bstar says:

      Chuck, you really think a majority of the people at an MLB game know what a hold is?

      Vote -1 Vote +1

  18. atigersgrin says:

    As mentioned closers would not like this as it’s taking money from them to give to others. Since many middle relievers likely believe they are good enough to eventually be a closer they may be opposed to this as well. There are many people who are not rich who don’t want to tax the rich more because they believe they will be rich in the future.

    Vote -1 Vote +1

    • gonfalon says:

      “There are many people who are not rich who don’t want to tax the rich more because they believe they will be rich in the future.”

      I think everyone will agree that someone making $11M a year, like Rafael Soriano, is rich. But are you really saying that someone like Octavio Dotel, who is making $3M this year, would be considered “poor”?

      Vote -1 Vote +1

  19. Dan in Philly says:

    When you hear the same argument for 15 years and yet nothing changes, including how “smart teams” employ and pay their closers, you begin to suspect the argument may, just may, be wrong. It’s entirely possible that paying one great guy to finish close games in the 9th is actually a pretty good use of your money. I would hazard to guess that if it were as grossly and obviously a waste of resources, the practice would have changed by now.

    Vote -1 Vote +1

    • PiratesHurdles says:

      Disagree, TB and Pittsburgh (among others) have figured out the cheap pen just fine the last few years. Things have changed, but its the minority still. Just because it hasn’t become obvious doesn’t mean its not there. One reason teams still pay top closers is because they are really good pitchers to begin with. That is no way means that you have to pay a guy alot to have a good closer.

      Vote -1 Vote +1

    • tz says:

      The only benefit that jumps out is that by having “predictable” roles (closer, 8th inn guy, 7th inn guy, etc.) relievers can start more innings and warm up specifically for “their” inning. This would relieve some wear and tear on guys from warming up repeatedly without being used.

      Vote -1 Vote +1

      • Dan in Philly says:

        Your argument ignores the fact that the leverage in the 9th inning is far greater than in the 8th, 7th, etc. Let’s say you have Pitcher Ace and Pitcher Beta. Pitcher Ace will shut down the opposition 95% of the time. Pitcher Beta will only be successful 80% of the time. The saber crowd argues that there should be no difference in using pitcher Ace in the 8th or the 9th inning of a one run game.

        However if Pitcher Ace gets through the 8th there’s a 20% chance the other team will win or go into extras, wasting his efforts and meaning he won’t be able to help you later. If pitcher Beta blows the 8th, you can either not use Ace at all or use in only if you get a lead in the 9th or extra innings.

        Without going through all the permutations of outcomes I looked at the typical usage of ace closers and concluded that almost all the time it’s better to use them in the 9th with a 1-3 run lead than any other situation. There are some exceptions, but if you only used your best relief pitcher in that role mindlessly you would come quite close to optimizing his production to your team. Looked at that way, the save is not a bad statistic at all for evaluating the ability of a closer, and not a bad measure for compensation.

        Vote -1 Vote +1

      • Eminor3rd says:

        You aren’t accounting for the fact that blown 1-3 run differences in early innings make 1-3 run innings in the 9th impossible. Take any situation and plant it in the 9th inning and the leverage will get more extreme, but if you look at games retroactively, you’ll see that the highest leverage situation is not always in the ninth inning, because the outcome of the given situation affects the formation of the 9th inning situation.

        Vote -1 Vote +1

      • Phrozen says:

        The problem with that line of thinking, Dan, is that Pitcher Alpha should be used against the opponents best hitters. If it’s the 8th inning in a one-run game, and I’ve got Albert Pujols, Miggy Cabrera and Joey Votto due up, you should throw Pithcer Alpha out instead of saving him to face Michael Martinez, Melvin Mora and Adam Dunn. Let Pitcher Beta deal with them.

        Vote -1 Vote +1

      • NadavT says:

        Dan in Philly’s argument is just a strawman. The SABR crowd definitely does not argue that “there should be no difference in using pitcher Ace in the 8th or the 9th inning of a one-run game.” Rather, the argument is that Ace should be used during the highest-leverage moments. All else equal, the 9th inning is higher leverage than the 8th inning, but all else often is not equal. As Phrozen notes, the batters faced matter, as do other factors, like the presence of batters on base when the reliever comes into the game.

        Vote -1 Vote +1

      • Jason says:

        “Dan in Philly’s argument is just a strawman. The SABR crowd definitely does not argue that “there should be no difference in using pitcher Ace in the 8th or the 9th inning of a one-run game.” Rather, the argument is that Ace should be used during the highest-leverage moments.”

        But how does the manager know when the highest leverage situation of the game is during the game? …it gets easier to predict the later in the game it is….

        Vote -1 Vote +1

      • CircleChange11 says:

        But how does the manager know when the highest leverage situation of the game is during the game?

        Nobody can answer that question, yet numerous commenters keep talking as if they had hindsight knowledge in present time.

        They don’t.

        If you blow a lead in the 7th or 8th inning, you still have time to make it up … if you you do the same in the 9th, well you have less time … and you (as a manager) may get fired if it happens more than once (bring ‘ace’ in the 7th, and 3rd best reliever in the 9th).

        Almost ALL of the comments presume that the manager knows what point in the game will be the highest LI. For an intelligent crowd, it’s a pretty dumb comment to make.

        Vote -1 Vote +1

      • CJ says:

        Sure, that’s a reasonable concern.

        My counterpoint: analyse the relative frequency of each LI value, and use this in combination with an “outings target” for each reliever (ie. closer pitches at most 70 innings, etc).

        You can quite easily cross-reference this expected frequency of LIs against a list of LIs for common situations by base-out state (6th inning or later, at most 3 run margin; you don’t particularly care when you’re down by 6 in the 4th because you just throw those innings to mop-up guys) so your manager always knows the leverage index.

        Then, give the manager some leeway (Pujols on deck, midst of pennant race, etc) on interpreting this LI if you believe it necessary, much like Tango’s “go with gut” option on intentionally walking Barry Bonds.

        If you can nail down the way LI is distributed, you could even account for rest patterns, etc; at least on the balance of probabilities. In this way, you could give you manager ONE number: if leverage hits 2.7, say, send in Ace; even accounting for the fact he can’t pitch tomorrow because of it.

        Vote -1 Vote +1

  20. tz says:

    What about showing the WPA in the corner of the screen during baseball games, like they show the odds during poker? Eventually fans will buy in and understand the shutdown/meltdown scenarios.

    +5 Vote -1 Vote +1

    • vivalajeter says:

      Are there different ways to calculate WPA? I assume there are a lot of factors, such as the batter on deck, who’s on the bench, how you calculate the odds of any single event happening, etc. If there are different ways to calculate it, which calculation do we put on the screen?

      With Poker it’s much easier – you know which cards are left and what odds of each card coming up.

      Vote -1 Vote +1

  21. Jason says:

    “It is one of the few instances in sports where obtaining a statistic, rather than maximizing a team’s chances of winning, actually drives how teams utilize their players.”

    This is a truism that I would actually like to see tested. It is assumed by SABR people that all managers use their bullpens wrong to accrue saves for their closer. It is assumed that the reason for this is the rules for recording a save (and that baseball managers are stupid).

    To me, the directionality of this relationship is not so apparent. I see no evidence that a team would actually do better if they used their bullpens differently. In my lifetime every single time a “closer by committee” has been tried it has failed. I suspect this is the case because having defined bullpen roles is actually more optimal than the strategy advocated by the SABR community. I suspect something like Darwinian evolution occurs in baseball and teams tend to adopt successful strategies and ignore unsuccessful ones.

    The modern method of bullpen management largely began in the late ’80s with Tony LaRussa. This is long after the save was defined. I suspect teams started doing it because they saw how successful LaRussa was at finishing ballgames with his strategy. I suspect teams are managing to win and not managing to get their closer saves. It might be the case that the two are related.

    Vote -1 Vote +1

    • Dan in Philly says:

      Well said, much better than how I tried to express it myself above. I have been following the sabermetric crowd long enough to recognize established sacred cows, and denegrading the save is one of them.

      Don’t get me started on how 300 wins actually is a pretty good indication of greatness in a pitcher :)

      Vote -1 Vote +1

    • Dave Cameron says:

      You’re arguing against a point I didn’t even bring up. You can absolutely keep the modern day rigidity of bullpen roles and still be in favor of eliminating saves and using SD/MD as the metric of choice instead.

      Instead of having your closer be your “saves guy”, have him be your ninth inning guy. Same role, same inning, different barometer for when he should pitch – instead of 9th inning up by three, it’s ninth inning down by one, tied, up by one and sometimes two. You’re not deploying him any differently, you’re just avoiding the inefficiency of using your best reliever to protect a three run lead with three outs.

      If you really don’t think managers are being controlled by the save statistic, please explain why teams almost exclusively use their closer in situations clearly defined by the save rule.

      Vote -1 Vote +1

      • Jason says:

        Dave,

        I am arguing against a point you did bring up. I quoted your point that I was arguing against. …I’ll do it again:

        “instead of 9th inning up by three, it’s ninth inning down by one, tied, up by one and sometimes two. You’re not deploying him any differently, you’re just avoiding the inefficiency of using your best reliever to protect a three run lead with three outs.”

        Yes you are employing him differently. Very differently. It is not just the inning that a closer pitches in, but also the situation. Using your best relief pitcher in a losing game might very well be a terrible strategy over the course of the season! Especially if it means he can’t pitch tomorrow in a game you need him in.

        “If you really don’t think managers are being controlled by the save statistic, please explain why teams almost exclusively use their closer in situations clearly defined by the save rule.”

        I already explained this. Perhaps it just so happens to be the case that using your closer in accordance with the save rule actually maximizes your chances to win the most games over the course of the season. As far as I know, no one has shown this to be the case. You are making the claim, so it is incumbent upon you to provide evidence for it. The null hypothesis ought to be that the managers know what they are doing. If you want me to believe that you, Dave Cameron, would be a better manager of bullpens than every single manager in the game for the past 20 years you had better provide evidence of it rather than just making an argument from plausibility.

        Vote -1 Vote +1

      • Jason says:

        Dave,

        It is not that I think you (and the multitude others that have made the claim) are wrong about the save rule dictating managerial strategy. I would just like to see some evidence of it. In my opinion, the SABR community differs quite starkly from real science in that it is extremely reluctant to challenge their own assumptions. You guys commendably challenge the assumptions of what you believe the status quo to be. But you are unwilling to challenge your own assumptions.

        As far as I can tell it is an assumption that managers cost their teams wins over the course of the season by using their bullpens the way they do. As far as I can tell, the rational for this belief is based upon a simplistic model of winning a single game in a vacuum and does not consider the trajectory of a season. Also it is based upon a model where player performance is unrelated to the situation they find themselves in. Managers don’t just manage to win one game. They manage to win as many games as they can across the season (as evidence, managers often alter their strategies for playoff games). Today’s game is affected by yesterday’s game and also affects tomorrows game. We know this is true, yet the model you use to conclude that managers are wrong does not consider this. Managers do. Also, players are human and the situation they are in likely affects them. Managers consider this. Your model does not. You have to be willing to test the conclusions that your model derives, especially when you know that the model’s assumptions are not entirely correct. This is the only way to get better models and better predictions.

        Vote -1 Vote +1

      • NadavT says:

        The evidence is in the win probability and leverage statistics that are presented at this site and discussed in this article. From a pure statistics perspective, the optimal approach is to use your best resources (i.e., your most talented relief pitchers) during those moments in the game when the outcome is most in doubt. Being up 3 runs in the 9th inning is a game situation with much lower leverage than the situations in which a “fireman” reliever was traditionally used before the save statistic was invented.

        You make a lot of gross assumptions about the SABR community’s lack of data about reliever usage. Here’s your data — it’s called leverage and win probability. On the other hand, you propose a sort of panglossian alternative in which the way things are done just happens to be the best way to do things. Where’s your evidence for that? You throw around a lot of ideas like how today’s game affects tomorrow’s game or how managers have to account for the fact that managers are human. Fair enough — what’s your hypothesis for how these factors affect optimal reliever usage? How would you test that hypothesis? In short, if you want to attack a data-driven argument, be prepared to back it up with data of your own.

        Vote -1 Vote +1

      • Jason says:

        NadTV,

        “The evidence is in the win probability and leverage statistics that are presented at this site and discussed in this article.”

        This is incorrect. That is not evidence that teams would actually win more games over the course of a season using an alternative strategy to the strategy that they now employ. That is data used to derive an HYPOTHESIS which PREDICTS that teams would win more games over the course of the season by changing their strategies. No one has ever, to my knowledge, whether this is, in fact, true. It is predicted to be true, but we don’t know it.

        Mine isn’t a Panglossian world view. I am simply (and correctly) stating that it is most conservative to accept the null hypothesis until there is evidence presented to refute it. The burden on refuting the null is not mine. It is on those who are asserting that bullpens are used incorrectly.

        Vote -1 Vote +1

      • Jason says:

        NadavT,

        Sorry for spelling your name incorrectly!

        Vote -1 Vote +1

      • noseeum says:

        @jason, I expect that this is most certainly not a hypothesis based on SABR assumptions. Rather it an observation of actual performance.

        Baseball has been played for many many years with statistics gathered for the vast majority of those years. Using your closer exclusively in a save situation is a relatively new phenomenon. It didn’t really take over until around the mid-80s. So we have decades of statistics where a team’s relief ace WAS NOT used exclusively for saves.

        I don’t have any links for you, but I have to think this research has been done, comparing team performance based on relief ace usage. You’re assuming it hasn’t been. I’d be shocked if that were the case.

        Vote -1 Vote +1

      • Jason says:

        Nooseeum,

        If any research has been done, I’d love to see it. I’m not as confident as you are that it has been done though. Dave Cameron isn’t aware of any such study as far as I can tell. Surely, he would cite the study rather than making an argument from plausibility if he was aware of such a thing. Also, it is actually a really difficult thing to study correctly. How do you do it? There are a lot of variables that need to be controlled for and it will be difficult to find the appropriate natural experiment in the data. I think the only feasible way to study it would be to model it through simulation.

        Vote -1 Vote +1

    • CircleChange11 says:

      Managers do this because losing games in the 9th if you’re not pitching your ‘closer’ will get you fired.

      Doing the same thing everyone else does puts the onus on the players to perform.

      It still comes back to teams having less time to come back from a blown lead in the 9th as they do for a blown lead in the 7th.

      Also, many bullpen situations in the 7th/8th innings feature relievers that have platoon split advantages.

      Vote -1 Vote +1

      • bstar says:

        Finally someone says this. Joe Girardi brings Mariano Rivera in the seventh inning with a one-run lead. Rivera blows the lead; the Yanks come back and forge ahead by one. Going into the ninth, they bring in Suboptimal Choice X, who gives up a game-winning HR. I wouldn’t envy Joe Girardi in that situation. Even his own players are going to be asked questions about that decision. Negative distractions hurt a baseball team.

        Vote -1 Vote +1

      • Sam Samson says:

        Yes, most fans would question the losing manager in this scenario — just as many fans questioned the first teams who emphasised the prioritisation of OBP over homes runs and RBIs. That didn’t make those teams wrong, though. It was a matter of educating those fans (the hard way), by doing it over a larger number of games, and showing them it worked.

        This whole discussion seems to me to be informed by an attitude of disdain for “the saber crowd.” As for “No one has ever, to my knowledge, whether this is, in fact, true. It is predicted to be true, but we don’t know it” — well isn’t the writer advocating putting the principles of LI into practice and therefore testing them in the real world?

        Vote -1 Vote +1

      • bstar says:

        There wasn’t a fan revolt when the A’s starting winning with OBP teams. Not a good comparison.

        Vote -1 Vote +1

    • jcxy says:

      “In my lifetime every single time a “closer by committee” has been tried it has failed”

      Isn’t there a selection bias here though? Teams that employ the “closer by committee” generally have a plethora of mediocre or below average pitchers. Should we be surprised that a bunch of meh RPs performed very meh?

      Vote -1 Vote +1

  22. bowie says:

    I think this will happen as soon as a few teams publicize the fact that they evaluate their relievers this way, and state that they don’t care about saves anymore. Then their local broadcasts will start using that stat, BB2N and baseball journalists will start referring to it, and eventually it will become commonly understood.
    Sort of like how people started to look at OPS or OBP instead of just BA, or how the W-L record has become a less popular measure. These are small, incremental steps, and they happen gradually. It doesn’t come from a decision from MLB.

    Vote -1 Vote +1

  23. Bip says:

    I don’t think of Don Mattingly as a very progressive manager, but whether intentionally or not, he has seemingly implemented the SD/MD bullpen management strategy. The Dodgers closer who starts the ninth inning when it’s a save situation is Javy Guerra (19.5 K%) and the guy who comes in when the leverage is highest is Kenley Jansen (44.0 K%).

    Vote -1 Vote +1

    • Except when the highest leverage occurs in the 9th and he uses Guerra.

      Vote -1 Vote +1

    • cable fixer says:

      Actually, the team that jumps to my mind in terms of this sort of usage is the Lidge-era Phillies teams. I, for one, loved seeing madson in the 7th and 8th instead of in the 9th. Of the many reasons Lidge got the chance to blow so many saves, Madson’s performance in high leverage spots earlier in the game has to number among the top few.

      Vote -1 Vote +1

  24. Stevesaxon says:

    Is there any reason the cutoff for SD/MD is 6%? I get that any cutoff chosen is completely arbitrary, but I feel as if the public would be more into the stat if the cutoff was, say, 5%. If it were 5%, then you could explain it to someone as, “If his team is 5% more likely to win after his appearance, then he gets a SD. If his team is 5% more likely to lose after his appearance, her gets a MD.” It’s a minor nitpicky point and may break the scaling of SD/MD to saves, but I feel as if 5% is a more intuitive number than 6% is. Of course, I may just be blowing smoke out of my ass as well.

    Vote -1 Vote +1

    • Tangotiger says:

      6% aligns itself to saves more. That was the principal reason.

      But, yeah, anything you choose is arbitrary.

      Vote -1 Vote +1

  25. Tangotiger says:

    I agree with two points above from the commenters:

    1. The ease of explaining a save situation compared to a SD or MD
    2. The requirement to show the Win Expectancy (WE) graphic, if not all the time, at least at critical junctures

    Speaking only for myself, using WE is what led me to WPA and LI (and eventually SD and MD). It’s just a natural progression.

    I think if viewers see the WE chart (occasionally, not all the time), then they’d be able to better appreciate when a relievers comes into the 7th inning with the runners on base and up by 1 run, and escapes with three outs and no runs (and exits the game).

    Right now, he gets a “hold”, which compared to a “save”, hardly seems to reflect much value.

    Vote -1 Vote +1

    • jcxy says:

      I’m on board with this but for a small reservation with one of the general premises–which is that we know when the high leverage situations are going to occur. Some have suggested the win probability graphs being included on telecasts, etc etc. And that would solve some of this (and, fwiw) increase my enjoyment of the game.

      But that misses the aspect of getting RP warm and ready without having them throw too much in the bullpen. Use the Mariano Rivera situation someone suggested. Let’s say it’s a 1 run game in the 6th and CC is on the hill. He walks someone and then gives up a bloop single. 1st and 3rd no out. Even the most casual fan understands this is a very high leverage situation and, if I or Giradi had powers of clairvoyance, I/Girardi would want Mo in. But is he ready? If CC gets out of it, am I willing to have him start and stop getting warm a number of times in that game? I am committed to using him that day, even if there aren’t any high leverage chances the remainder of the game?

      Now…am I worried about the effect of lots of bullpen pitches in game 5 of a playoff series? Not really. What about on a random Wednesday in Baltimore in June? Would the added bullpen pitches decrease effectiveness and potentially raise injury risk if used over the course of the season? I know it’s anecdotal and mostly correlational, but we can all think of many situations where RPs complain about having gotten up too many times in the pen and it hurting their performance outcome both short- and long-term (Broxton and Torre, Lidge’s declining FBv after his 110 bullpen pitches in the All-Star game).

      For me to fully embrace a new bullpen strategy, I’d want to see that these worries are unfounded or that you could maintain some of the division of bullpen labor.

      Vote -1 Vote +1

  26. jfree says:

    I suspect that this hasn’t happened because the “closer” is actually a somewhat unique relief pitcher. Many relief pitchers have the mental ability to pitch in leverage situations. But a closer also has to have the ability to do so without any significant platoon/situational splits.

    “Leverage” and WPA can’t measure the capability of the opposing coach to pull batters in off the bench. That’s not really a coaching option earlier in the game but in the 9th inning, it is.

    I’d like to see a team with the nuts to get rid of their high-paid “closer” (presumably when they blow up) – and then put a different system for managing their bullpen. Realistically though, it will be virtually impossible for them to simply institute a SD/MD system using the current 1-inning relievers. Because even if they pay the middle relievers who do have those “closer” skills, some other team will simply rob them and slot those guys into a traditional system. My guess is they’d have to run with 2-inning relievers – who might not have the per pitch shutdown capabilities of the traditional closer

    Regardless, it won’t happen until some MLB team actually puts their money into paying for shutdowns rather than saves.And that’s about as likely as a herd animal deciding to march to their own drummer (confusing metaphor).

    Vote -1 Vote +1

  27. philosofool says:

    I just don’t understand what’s so hard about WPA. Team’s probability of winning when the reliever leaves minus the probability of winning when he enters isn’t some kind of insane idea. The public should actually find this pretty easy to grasp, and so should your typical jock. These probabilities are calculated as simple frequencies–it’s fifth grade mathematics, except that the numbers are pretty big and the arithmetic makes you want a calculator.

    Vote -1 Vote +1

    • vivalajeter says:

      It’s not nearly as easy as you’re trying to make it, especially the part about “These probabilities are calculated as simple frequencies”. What’s simple about it? How do you measure the chance of losing the game with Pujols hitting instead of a regular hitter? What about the person who’s on deck? What about the stadium you’re in? I’m sure there are a lot of smart people that would calculate it differently. With millions of dollars on the line, which method do you use?

      Vote -1 Vote +1

      • philosofool says:

        These problems exist for *every metric*, including Shutdowns and Meltdowns, and are well recognized by every smart person who thinks about baseball.

        They are still simple frequencies. If 9 games have been playing in the state in which the pitcher enters the game and 6 of those were won, the win probability in that state is .666….

        Vote -1 Vote +1

      • vivalajeter says:

        “If 9 games have been playing in the state in which the pitcher enters the game and 6 of those were won, the win probability in that state is .666″

        Let’s set aside the fact that just because a pitcher enters a tie game with a runner on 3rd and 2 outs in the 9th, that all similar games have the same expected odds of winning (obviously it also depends on the hitter, etc). Instead, let’s focus on the frequency. How far back are you looking to see how many games were played in a similar state? In a given situation, let’s say the pitching team won 71% of the games over the last year. But they won 72% over the last 3 years, 70% over the last 5 years, etc. Which number do you use? If you’re using an arbitrary cut-off of 6%, that can make a huge difference in whether someone gets a shutdown. And if this is the main metric we’re using in arbitration, you’d better be sure it makes the most sense.

        Vote -1 Vote +1

    • bstar says:

      I can see fans in the stands picking up their calculator taped under their seat as they first get there. Right.

      Vote -1 Vote +1

  28. Sandy Alderson says:

    Your analysis falls apart once you consider Frank Francisco!

    Vote -1 Vote +1

  29. Pete says:

    Regardless of whether fans are willing to adopt Shutdown/Meltdown into their boozy jargon, front offices absolutely should place more value in stats like these and their market activities should reflect that.

    Vote -1 Vote +1

  30. PKJ says:

    Couldn’t you say the same thing about W/L, RBI, batting average, and a whole multitude of BS stats?

    Vote -1 Vote +1

    • Max says:

      No. Part of the way the game is played has been shaped around the save, unlike the other stats you mentioned. It is making an actual impact.

      Vote -1 Vote +1

      • Keystone Heavy says:

        That may not be totally true. I would be willing to bet that runs scored, RBIs and BA DO have an impact on how a manager runs his team. There may be managers that fall into a sort of self fullfilling prophecy by putting inferior hitters with more RBIs in the past into lineup spots they shouldn’t have. Or managers who unjustly give less playing time to players who have (for example) .200/.350/.450 type lines. Or Managers who factor in the W/L record of his pitchers when deciding who will pitch on opening day.

        Not that anything I said isn’t 100% skeptical, I just suspect that it is pretty likely.
        Of course I don’t

        Vote -1 Vote +1

  31. Wily says:

    Minor nitpicking, but Soriano made the vast majority of his money once he was FA eligible. He signed a 2 year $9 mil deal in which the second year ($6.1 mil) bought out what would have been his first year of free agency. Then the deal with Tampa for $7.25 mil was when he could have been a FA but elected to accept arbitration. He’s more in the Mike Adams ballpark of money earned if we’re talking service time.

    Vote -1 Vote +1

  32. noseeum says:

    Unless you can get the Yankees PA to play “New York, New York” after a 7th inning shutdown, I’m not sure I can get on board with this plan.

    :P

    Vote -1 Vote +1

  33. Something about supporting a COMPLETELY contextual stat seems to go against my saber beliefs. Why, only the the case of RP, should context suddenly become important? In almost every other situation, we saber folk try to isolate out all context and judge value in a vacuum. Why should we then judge RP by WPA, the most context dependent stat there is??

    Vote -1 Vote +1

    • Newcomer says:

      The simple answer would be that relievers are intentionally placed into a given context. The players that start the game will play in whatever context comes up, but in the case of relievers (and pinch hitters), the manager selects the player according to the context.

      I don’t think Dave believes SD/MD is THE way to evaluate relief pitcher performance. But it provides a counting stat alternative to the Save. With relievers, the manager can decide in which situations he can deploy each of his pitchers, and a reliance on the Save is causing managers to select sub-optimally.

      Granted, this is not as sub-optimal a usage as it would seem from the amount of hand-wringing in some sabermetric circles. From some comments, it would seem that closers ONLY come in when there is a three-run lead in the 9th, which is not true.

      Vote -1 Vote +1

  34. Alex says:

    Arbitrators (you mean agent & teams) are allowed to use advanced stats, but because arbitrators are not “baseball insiders”, using advanced stats would require a SABR 101 course first. There are time limits of presentations and such a course would not allow a case to be presented.

    As far as the SD/MD replacing the save, we need two things: technology in the dugout and better broadcasts. In today’s world of technology, there is no reason for laptops, iPads, etc. to be banned in the dugout, as they are today. I would even go as far as to recommend a uniformed “researcher” in the dugout (it could be some type of intern to save money) to look things up at the manger’s request. The bench coach could help with this as well.

    The other thing we need are better broadcasts. The biggest problem right now is that broadcasts are not equipped for it and the broadcasters are uneducated. I am a Jays fan and notice this frequently about their games. During the first game of the season, a player’s OBP, SLG, OPS and HR were brought up when he came to bat. Since then the stats displayed are AVG/HR/RBI and OBP/OPS sometimes. The pitching side is a bit better, showing the standard stats (holds are not displayed), but K, H & BB have been replaced by K/9 and K/BB ratio. I cannot recall at the moment if H/9 are displayed. GB% is also displayed. Further, displaying 2012 pitching stats this early is absurd. They should display the prior year’s or career numbers solely or in conjunction with the current year’s until at least May 1. The problem is when they are explained incorrectly. Twice now, the broadcast team has explained GB% as a percentage of outs, not batted balls. Education of broadcasters is the most critical component here.

    This brings me to my final point: broadcast displays. I explained above what is displayed in a typical Jays broadcast and it is very poorly done The batter graphics do not always appear and I noticed several times tonight where they failed to mention, in text, who was even due up the next half-inning.

    My solution for the SD/MD problem is live win probability as part of the display. Express it as a percentage of win probability for the team that owns the feed (from the perspective of the Yankees on YES, but the Red Sox on NESN for example). Place that next to the score and you could introduce SD/MD quite easily.

    Vote -1 Vote +1

  35. Eric M. Van says:

    I find it amazing that no one has yet adopted live WE in the game status box. It’s trivial to understand — the only thing that needs to be explained is that it’s the theoretical odds of the average team winning against an average opponent, and ignores who is actually pitching, at bat, or will be doing so subsequently (and the fact that home teams usually play a bit better than road teams). And anyone who has ever followed a game with FanGraphs live (or a copy of The Book handy) knows it’s addictive. My casual-fan roommate, who basically only watches if he wanders into the living room in the late innings of a close game, has been known to ask me what the WE is, or was before a lead was blown, because I often talk in those terms.

    Of course, putting the WE up will force every color man in baseball to explain why it has just gone down after a successful sacrifice bunt. :)

    Prediction: once someone pioneers this, it will be ubiquitous within 3-5 years, and a few years after that you can introduce SD/MD.

    Vote -1 Vote +1

  36. pft says:

    I don’t think any stat should be killed. However, I think MLB should consider sanctioning some new official stats that are not based on estimates or adjustments, no matter how valid they seem.

    Shutdowns and meltdowns are based on WPA which assumes league averages apply. I don’t agree MLB should make them official stats, anymore than they should make WAR, UZR, OPS+, RC, etc official stats.

    Vote -1 Vote +1

  37. CircleChange11 says:

    I really don;t care if casual fans know about SD/MD or not. I just want teams, managers, and organizations to know about such things.

    I’m sure there is plenty of information that teams use in NASCAR that the casual fan doesn’t know about, and probably doesn’t need to know about.

    When it comes to relievers just knowing the number of appearances without allowing a run is an improvement.

    Knowing what % of inherited runners score is an improvement.

    The “save” and “hold” are stats that have a use, it’s just that there’s stats that have a better use.

    Again, I don’t care if the rest of the baseball world knows about SD/MD … as long as those making baseball decisions do. What the casual fan knows is of far lesser importance.

    By the time SD/MD comes along, sabermetrics will have a new stat that’s replaced SD/MD. Not unlike how the baseball world finally caught on (a little bit) to OPS and now there’s wOBA.

    I would not be surprised at all if the trend of TV broadcasts sort of went the way of video game presentation, where players are possibly rated on the 1 to 99 scale (like The Show, 2K, madden, etc) … where on the TV screen it would show Pujols as a 87 Contact, 85 Power, 74 Fielding, 93 Discipline and 90 Overall … talk about stuff the fan is already familiar with.

    It would not be difficult to convert sabermetrics into these type of ratings. IMO, it’s missing the boat not to.

    Vote -1 Vote +1

  38. Nathan says:

    Kind of an aside, but something Zac’s great post led to… I wish we could convince some of the broadcasts to start showing a real-time win probability graph like FG does. That would be sweet. I think showing a graph like that after every AB or maybe even after every “major” event (XBH, run scored, end of an inning) would be a cool way to get fans slowly into the concept of looking at these probabilities.

    As mentioned in the original original article, people understand and like this presentation with things like the World Series of Poker. So there is reason to believe the comprehension is there, it just needs baby steps. Making the win probability graph into some fancy animated graphic with highlights of the key events in the game to the given point over-layed would draw some eyeballs, no?

    Vote -1 Vote +1

    • bstar says:

      Why would the average fan want to look at a graph instead of just watching the game unfold? We’re the ones looking at graphs and enjoying them. To think Joe Sixpack is gonna give a crap is a big leap.

      Vote -1 Vote +1

  39. Snowblind says:

    Another solution is to see more teams develop more quality starting pitchers who can actually reliably go 7+ innings, and swing baseball back towards the model of “a reliever is a failed starter”. Wouldn’t need to measure and value reliever contributions if we had a generation of better, smarter, stronger pitchers who don’t flee after 6 innings / 100 pitches.

    Vote -1 Vote +1

    • CircleChange11 says:

      There’s a shortage of pitchers that can get modern, good hitting lineups out three times through the lineup on a consistent basis.

      There is not, however, a shortage of hard throwing fastball-slider guys that make great short relievers.

      You’re idea that it’s a “team development” issue is naive. Pitchers don;t “run out of gas” or “throw lesser pitches” in the 7th inning or the 3rd time through the order … it’s that batters have acclimated to them, their speed, their movement, etc. Even great pitchers like Halladay, Verlander, etc follow the pattern. The more a lineup sees a starting pitcher, the better they do against them (in a single game). A new reliever is most often a better option than a starter going through the lineup a 3rd or 4th time. That is a sabermetric supported statement.

      I’m generally a big fan of baseball from the 60s through 80s, but if you watch some of those games, starters (even more than today) are left in the game for far too long. They get away with it because 7-9 hitters in those eras were nothing more than ground ball slappy types that would allow the pitcher to “take an inning off”. As linear mechanics have been replaced by rotational mechanics and astroturf has been replaced by smaller ballparks, there really are very few “weak-ass” hitters in each lineup.

      Orel Hersheiser essentially explains the same thing as a big difference between when he pitched and now.

      A starter that goes 7 strong innings each start makes the managers job very easy. If there were more guys out there like that, managers would use them … damn near abuse them. Those guys aren’t “developed” IMHO.

      Vote -1 Vote +1

  40. birtelcom says:

    Fangraphs has Heath Bell with 31 Shutdowns in 2011. When I look at Heath Bell’s 2011 games on baseball-reference, it shows him with 43 games with a WPA of .06 or more. I know there are some slight differences in the way fangraphs and b-ref calculate WPA, although I’m not sure what those differences are. The difference in the number of Shutdowns for Bell assigned by fangraphs vs. b-ref seems very large but appears on close inspection to be the result of very small differences in WPA calculation methods. It looks like Bell has a large group of games in 2011 where he is just below 6% WPA (5.3% to 5.7%) on fangraphs and just above 6% (6.1% to 6.2%) on b-ref. What is creating this difference, and how should it be resolved? After all, one important condition for having a stat widely accepted is likely a consensus on how the stat is calculated and what the numbers are — 31 Shutdowns on one major site and 43 on another, for the same player season, is never going to result in mainstream acceptance for the concept.

    Vote -1 Vote +1

  41. 801braves says:

    How does this metric account for a reliever that pitches multiple innings but his team scores a bunch of runs? A reliever that comes in a tie game and pitches 3 innings, giving up three runs, but his own team scores 6 will see the team’s winning percentage increase by more than 6%. Is this a shutdown?

    Vote -1 Vote +1

    • 801braves says:

      or a reliever comes in with his team down 2, pitches for 2 innings, and his team scores 4. Again, the WPA goes up by more than 6%, but the pitcher didn’t do any better of a job than the guy whose team didn’t score any runs in a similar situation. It seems this stat works well in some instances but pitchers can still be dependent on their offense

      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>

Current day month ye@r *