66 replies to this topic

[Smiling Joe Hesketh]

Throwing this out there to the board. This is a modest proposal for discussion; I'm very interested to hear what others might have to add.

When the subject is broached of the individual who's done the most damage to major league baseball over the years, the usual names are offered as answers: The Black Sox, Charlie Comiskey, Cap Anson, Charlie Finley, Marvin Miller, Bud Selig, Ford Frick, Barry Bonds, etc.

Today, for your consideration, I'd like to offer up the name of another: Jerome Holtzman.

Who is Jerome Holzman, you may ask? He's the man responsible for the definition and introduction of the modern save rule, made part of baseball rules in 1970.

Today's save rule reads as follows from Rule 10.19:

The official scorer shall credit a pitcher with a save when such pitcher meets all four of the following conditions:

1. He is the finishing pitcher in a game won by his team
2. He is not the winning pitcher
3. He is credited with at least ⅓ of an inning pitched

4. He satisfies one of the following conditions:
A. He enters the game with a lead of no more than three runs and pitches for at least one inning
B. He enters the game, regardless of the count, with the potential tying run either on base, or at bat or on deck
C. He pitches for at least three innings

No more than one save may be credited in each game.

Why is Mr. Holzman such an evil man? Because these definitions of saves do not necessarily correlate well to either pitching well in relief situations, or have no direct relation to winning as many baseball games as possible.

The rule is so broadly written that the save statistic itself is pretty meaningless. A 3 inning save in a blowout is considered equal to pitching 1 inning with a 3 run lead which is considered equivalent to coming in with men on base in a 1 run game and getting 5 outs. Wes Littleton's effort in this game is considered as worthy of a save as Jonathan Papelbon's work in this one. While this is the same weakness that the wins statistic can have for starters, this absolutely defies common sense.

Even worse, the save rule assumes that pitching with a lead in a game is inherently more valuable than pitching in a tie game, which leads managers and organizations to place their best pitchers in roles which may not have the highest leverage, and to spend insane amounts of money to acquire players who have the "makeup" to be a closer. A cult mentality has developed around the closer, when in fact the job of getting traditional saves is either not as difficult as widely assumed or is of vastly overstated importance.

Two scenarios for your consideration:

On Friday night, the Red Sox were in a tie game on the road that went to extra innings. In the bottom of the 11th inning, the Red Sox were in a situation where giving up a single run would cost them the ball game. Terry Francona had a choice to make as to which relief pitcher to bring into the game in that situation. He decided to bring in his 5th best reliever in the person of Mike Timlin, instead of either his best reliever in the person of Jonathan Papelbon or his second best reliever in Hideki Okajima. Remember, a single run allowed costs the Red Sox the ball game. Mike Timlin is far more likely to give up a run than Papelbon. Yet Francona chose Timlin in that situation because he's mentally tied to the rather useless idea of the save situation, that somehow pitching an inning with the lead is more important than pitching an inning in a sudden death situation. He saved Papelbon for a situation that never arrived. In this situation if Timlin gives up a run the game is over. Papelbon could enter with a 2 or 3 run save situation on the line, give up a run, and still do his job as it's been defined. Does this really make a damn bit of sense at all?

Secondly, let us consider the curious case of Joe Borowoski. In 2007, Joe Borowski was a below-average major league pitcher by any reasonable evaluation. His ERA was 5.07, he allowed 96 runners in 65 2/3 IP, and allowed 9 home runs in 2007. Yet he also led the American League with 45 saves. He had 9 games where he entered and gave up runs yet still earned a save. He only had 13 saves out of 45 where he entered with a 1 run lead and protected that lead. Doesn't the "success" of a crappy pitcher like Joe Borowski in traditionally-defined save situations indicate that either A) it takes no special skills to become a modern day closer, B) finishing games for a save is not the most crucial job in a modern bullpen , or C) our save definitions are sadly out-of-date?

Mangers getting tied to the notion of using their closer only in save situations is costing teams ballgames. It cost Terry Francona a game in TB last week; it may well have cost Joe Torre a World Series in 2003. It's almost perverse how the notion of the closer can be inversely related to winning baseball games.

The modern closer job is a job description, not a skillset. Pitching a single inning with a 3 run lead simply isn't that difficult for the vast majority of major league pitchers. It's time for baseball to change the rules as to what constitutes a save in order to make the statistic more relevant to winning baseball games. Changing statistical measures is not new in baseball; in the mid-80s MLB got rid of the stat of Game Winning RBI when it became apparent that it measured nothing meaningful.

I would propose A) getting rid of the 3 innings to end the game portion of the save rule (no one should get a save in a blowout), and B) dumping the 1 inning with a 3 run lead portion of the rule. If managers and organizations are going to insist on the importance of the closer, make that closer's stats count. Smart baseball organizations need to break out of the "save him for a lead" mentality of the closer's job and use their best pitchers in the tightest situations. While this idea is hardly new (Bill James has been arguing for this for years), perhaps changing the definitions of a save will allow managers to use their ace relievers more wisely.

[TomRicardo]

I don't mind the 3+IP save rule. It is more about saving the rest of your bullpen. Also something like 9 runs is not that impossible in 3 innings if you march out some crappy relievers.

[Fratboy]

I would propose A) getting rid of the 3 innings to end the game portion of the save rule (no one should get a save in a blowout), and B) dumping the 1 inning with a 3 run lead portion of the rule. If managers and organizations are going to insist on the importance of the closer, make that closer's stats count. Smart baseball organizations need to break out of the "save him for a lead" mentality of the closer's job and use their best pitchers in the tightest situations.

I completely agree, but your proposal doesn't really solve the problem because

1) There aren't many 3+ inning saves to begin with.
2) It still allows for the situation on Friday night to happen.
3) There isn't an easily understood way for Joe or JoAnn Fan to evaluate relief pitchers (ERA or RAA doesn't quite cut it.)
4) And because of 3) there isn't a proper way to measure or incentivize relievers when used properly.

Granted, the Red Sox are lucky because they have TWO relief aces, and Okajima was frequently used as the Jamesian relief ace. I think James himself has a proposal in Goldmine 2008 about a change to the rule, and it goes along a sliding scale where you get a save for pitching 1 inning with a 1 run lead, 2 innings with a two run lead, and 3 innings with a 3 run lead. And maybe you get a save in addition to a win for being the pitcher of record in the tie situation, but that's the situation that needs a counting stat associated with it, or maybe the Hold stat needs to be better modified to account for it.

[RSN Diaspora]

I think the crux of the problem leads back to the ultimate issue for all relievers: there is no truly reliable metric for a relief pitcher, and as awful as the save may be (you make a pretty compelling case), it's as good as any other current measure of relief pitching. If you try to reduce the importance of the save, the only way to have a positive impact is to come up with a new statistic that is a better measure of a reliever's importance.

Simple economics dictates that pitchers will seek to perform at a level that gets them the best status and payday. Thus, while Hideki Okajima is a better pitcher than Joe Borowski, Borowski gets the saves and thus the closer's paycheck and status. Even casual fans know who their closer is and who others teams' closers are. But few of them could name more than one or two setup guys on their own team, and even more interested fans are going to know closers on other teams but only a handful of the setup guys. I know who J.J. Putz is, but I couldn't tell you who his primary setup guys are (yes, I know he's injured right now, but generally speaking...). So by eliminating the save or at least actively seeking to reduce its importance, you probably only open the door to another flawed metric becoming front and center.

You probably can't use wins and losses, since those aren't particularly common for relievers. Say ERA supplanted the save. This, IMO, would be even worse. Someone like Mike Timlin, who has a horrible history with inherited runners, could have a great ERA, but would you want him coming in to a two-run game with runners on? I think you would need to come up with some kind of stat that accounts for outs remaining in the game, the score, inherited runners, batters who face the reliever in question and make it to second base or beyond, and runs scored while the pitcher is on the mound. If you can find a way to weight all of that and come up with a way that fans and media members can calculate it, then we may (may) find relievers used in a more optimal fashion.

Edited by RSN Diaspora, 28 April 2008 - 08:42 AM.

[Worst Trade Evah]

The rules for Wins are as bad or worse as the rules for Saves. How many times does a guy get too much slack in the 4th because his manager wants him to get a win? These assignments of highly team and context dependent outcomes to individual performers are always a problem. A guy can pitch 9 scoreless, have some reliever give up some runs, who can then vulture the "win" if the offense comes back after all.

[Rudy Pemberton]

Would a change in the save rule change how Tito uses Papelbon? The team has avoided using him in tie games since his injury, for whatever reason. I don't think we are likely to see Papelbon pitching in extra innings on the road, unless the Sox have the lead. I think it's likely because they only want him to go an inning and prefer to slot the relievers in their most familiar spots, but not really sure. Dissapointing when Okajima and Papelbon don't pitch in a tight loss (or two, or three) but I don't really think the save rule was really a factor. Babying of relievers, perhaps?

On the subject of Bill James, it's fairly interesting that the team has carried 12 or 13 pitchers all year, and at least 12 for most of his tenure, isn't it? An extra bat could have been useful lately although the team has been in some bizarre circumstances with the flu and all.

[OttoC]

Rather than giving all the blame to Holtzman, perhaps MLB needs to bear some of the criticism. Holtzman simply used it as a reporting tool starting around 1960. MLB adopted it as an official stat in 1967. His original save rule said: "To earn a save, the reliever had to come in with the tying or winning run on base or at the plate and finish the game with the lead."

The original, official save rule said: "Credit a save to a relief pitcher who enters a game with his team in the lead if he holds the lead the remainder of the game, provided he is not credited with the victory. A relief pitcher cannot be credited with a save if he does not finish the game unless he is removed for a pinch hitter or pinch runner. When more then one relief pitcher qualifies for a save under the provisions of this rule, credit the save to the relief pitcher judged by the scorer to have been the most effective. Only one save can be credited in any game."

I don't find that an especially bad concept, although it is ambiguous when it says "he holds the lead for the remainder of the game" but then says "when more than one relief pitcher qualifies for a save."

[wibi]

I fail to see how "saves" would affect a managers decision as when to pitch a certain pitcher. I dont argue that managers arent set by the concept that their most valuable pitcher is used for a traditional save situation but I would argue that making a change to saves situation rules will take a while to populate down through the through processes before managers change how pitchers are used.

[Fratboy]

Rather than giving all the blame to Holtzman, perhaps MLB needs to bear some of the criticism. Holtzman simply used it as a reporting tool starting around 1960. MLB adopted it as an official stat in 1967. His original save rule said: "To earn a save, the reliever had to come in with the tying or winning run on base or at the plate and finish the game with the lead."

Did the team have to have a lead at the time the reliever entered the game?

[Kevin Youkulele]

I think you would need to come up with some kind of stat that accounts for outs remaining in the game, the score, inherited runners, batters who face the reliever in question and make it to second base or beyond, and runs scored while the pitcher is on the mound. If you can find a way to weight all of that and come up with a way that fans and media members can calculate it, then we may (may) find relievers used in a more optimal fashion.

WPA (or its more elaborated BP version, WXRL) tracks most of what you're asking for, but is not tractable for calculation by hand/in one's head. However, this is not a death knell; there's the example of QB rating in the NFL, and some ESPN broadcasts were including win probability information last year IIRC (although those may have been in the college WS). Still, it will take a long time for it to get into broad mainstream circulation given that everyone still shows AVG/HR/RBI when a guy comes to the plate and every once in a while they'll tell you what his OBP is.

[BCsMightyJoeYoung]

A very thoughfull, well presented post by SJH .. just about everything here I would agree with. To Jerome Holtzman, I think you could add Tony Larussa and Arbitration as the major culprits here. Modern Closer usage patters seem to have been set in stone with Larussa's usage of Eckersly and the rest of the A's pen in the late 80s early 90s. This led to huge Save totals over the next few years (Bobby Thigpen as Exhibit A) which were subsequently used by their agents to drive up the salaries of Closers and the establishment of the "Closer Myth". Now - I have no idea that it takes a special skillset to be a major league Closer. I would strongly suggest that it doesn't. But I don't know this - and neither does SJH or Bill James for that matter. Personally, I think that there actually may be a few pitchers out there that do wilt in high leverage sitiuations. But, finding them statistically is a difficult, if impossible, task.

It's kind of like the argument that it takes a special mind set to play in Boston or New York. Most folks on this board tend to believe this .. so why is the opposite theory applied when closers are evaluated?

But this brings up an interesting dichotomy. That being:

The RedSox are viewed as a progressive, forward thinking organization. You would think that, if the Sox truly believed that even the likes of Joe Borowski could make an effective closer, then they would not be using Papelbon in that role. This seems to me to be a perfect example of having a Competitive Advantage. Let the poor downtrodden tradition-bound Organizations invest fortunes in guys who arn't very good. So much the better for us.

Tito has shown over the years that he understands the idea of using your best pitchers when it really matters. His post season managing record is outstanding. So why doesn't he do this in Tampa in April? I think it's because Francona has two, fairly fixed, but different approaches to player usage - the Big Game theory in which the best/hottest players are used exclusivly - and the "All other games" theory where he resorts to more traditional player usage.

I think it comes down to this. Papelbon is a Closer because he's the most comfortable in that role .. AND .. he's still cheap. Whenever one of these things change then I expect he will become a starter or , more than likely, an Angel or a Met or an Indian.

[Fratboy]

I fail to see how "saves" would affect a managers decision as when to pitch a certain pitcher. I dont argue that managers arent set by the concept that their most valuable pitcher is used for a traditional save situation but I would argue that making a change to saves situation rules will take a while to populate down through the through processes before managers change how pitchers are used.

It would change usage patterns almost immediately, and yes, the save rule almost entirely dictates when a "closer" pitches (you can call this the LaRussian Relief Ace... good lord this is like a game of Mornington Crescent) except for when the closer needs to get some work in because he hasn't pitched in a week.

Saves are the primary way closers are evaluated, and the second that rule changes (if it's ever changed), is when usage patterns will change. I think the MLBPA and the owners would go over this thing with a fine-toothed come, and it would be a big deal, unless the rules committee or whatever it is had some kind of unilateral authority over it.

[jose melendez]

I guess my problem with this is that it makes comparrisons with past relievers on the save rule pretty much impossible.

Are there any examples of the definition of a statistic in baseball actually being changed?

The way it normally works, it seems to me, is that a new, better statistic is created. They didn't change the definition of batting average to include walks, they created OBP, etc. Have they ever changed the definition of wins, for instance? (Serious, not rhetorical question).

Maybe have a new metric called clutch saves or something.

[jodyreeddudley78]

I fail to see how "saves" would affect a managers decision as when to pitch a certain pitcher. I dont argue that managers arent set by the concept that their most valuable pitcher is used for a traditional save situation but I would argue that making a change to saves situation rules will take a while to populate down through the through processes before managers change how pitchers are used.

It affects how managers use their pitchers right now. In the Buchholz home run thread OCD SS made a great point that many pitchers are accustomed to their defined roles. If Paps comes into a game with a 1 run lead, he probably expects to close it, which is not always the situation with highest leverage. If Francona weren't restricted by clubhouse roles and keeping players happy, maybe he brings Paps into a game in the 8th with two runners on in a one run game, only to have someone elso close the game in the 9th. Th fact of the matter is, as long as there is a strict definition for what constitutes a save, pitchers are going to desire that role. Even if that role isn't always the most effective one. If you redefine or eliminate the stat I guarantee that the more intelligent managers/organizations will change their usage patterns. Like all good ideas, this will lead to a trickle down effect in MLB.

edit: what Frat said

Edited by jodyreeddudley78, 28 April 2008 - 09:15 AM.

[B H Kim]

It would change usage patterns almost immediately, and yes, the save rule almost entirely dictates when a "closer" pitches (you can call this the LaRussian Relief Ace... good lord this is like a game of Mornington Crescent) except for when the closer needs to get some work in because he hasn't pitched in a week.

Saves are the primary way closers are evaluated, and the second that rule changes (if it's ever changed), is when usage patterns will change. I think the MLBPA and the owners would go over this thing with a fine-toothed come, and it would be a big deal, unless the rules committee or whatever it is had some kind of unilateral authority over it.

Anyone who questions whether the save rule dictates closer usage should pay attention the next time the home team extends a 3-run lead to a 4-run lead in the top of the 9th. I'd venture that in close to 100% of those situations, the closer is warming up and almost immediately is told to sit down. I don't think I've ever seen that happen when a 2-run lead was extended to a 3-run lead. That said, shouldn't leaving the rule as is actually present an opportunity to the more forward-thinking teams? If there are most efficient ways to use a closer than in purely save situations, the manager that takes advantage of those situations should gain some marginal advantage over managers who remain tied to the save rule.

Edited by B H Kim, 28 April 2008 - 09:17 AM.

[Fratboy]

That said, shouldn't leaving the rule as is actually present an opportunity to the more forward-thinking teams?

Of course it does. Depending on your frame of reference:

1) The team: It allows us to leverage an excellent reliever and depress his salary because he'll never achieve the important counting stats that arbitration and the market use to set salaries.

2) The player: I ain't got no taco.

Okajima could close for nearly any team in MLB. But because of the way he's used by the Sox, he'll never make the big bucks unless he's installed as "closer" for a team. That's just the way it works. Setup guys, despite having the more difficult job, are not well compensated because there's no effective way to measure their contributions.

[Cuzittt]

On Friday night, the Red Sox were in a tie game on the road that went to extra innings. In the bottom of the 11th inning, the Red Sox were in a situation where giving up a single run would cost them the ball game. Terry Francona had a choice to make as to which relief pitcher to bring into the game in that situation. He decided to bring in his 5th best reliever in the person of Mike Timlin, instead of either his best reliever in the person of Jonathan Papelbon or his second best reliever in Hideki Okajima. Remember, a single run allowed costs the Red Sox the ball game. Mike Timlin is far more likely to give up a run than Papelbon. Yet Francona chose Timlin in that situation because he's mentally tied to the rather useless idea of the save situation, that somehow pitching an inning with the lead is more important than pitching an inning in a sudden death situation.

[...]

It cost Terry Francona a game in TB last week;

I assume the game in question is Friday's game... and while they lost, there is no guarantee that the Sox win the game even if Papelbon comes into the game in the 11th instead of Timlin. The Sox had only scored a single run in the prior 9 offensive innings... and would go on to score a single run in the next 18 innings. Jonathan Papelbon coming in does not make the offense better. So, while it is certainly possible that Papelbon coming into the game in the 11th prolongs the game giving the Sox offense more chances to score another run... it also may have just delayed the inevitable of bringing someone else in who will give up the winning run.

I think it is important to look back at the entire run of the bullpen in this game, not just in the 11th. Aardsma, Corey and Lopez have already been used. Timlin came into the game in the 10th, and got the final 2 outs.

Once you go to the 11th... you have the following pitchers available: Timlin (who is in the game), Papelbon (who was rested, but had not gone more than 4 outs the entire season), Okajima (who had pitched 2 nights prior, and has only one game above an inning pitched this season), and Delcarmen (who was coming off the flu and pitched not so well 2 days prior).

Now, if the question is who has the best chance to get through the inning unscathed, the answer may very well be Papelbon. It is certainly not Timlin. However, looking at this Papelbon, Okajima and Delcarmen... the best case scenario is that you may be able to get 5 innings combined out of them and have no bullpen for the rest of the weekend. That is the best case scenario. The worst case scenario is the Papelbon, Okajima and Delcarmen can only go one inning a piece. Assume they do so without giving up a run... who pitches the 15th inning?

Now, I can understand the sentiment that all you want to do is get to the next inning, and Papelbon can get you there. But, the question of who pitches once I make all these moves has to be thought of in this instance. You have a whole season to play and more importantly, you have 2 more games in the series... the next one started by a rookie, the second one started by a pitcher who had missed his previous start due to illness. It had to be considered probable that the bullpen would need to be used (possibly heavily) in the next 2 games.

Therefore, the decision to start the inning with Timlin had nothing to do with the save rule, it had everything to do with stretching the bullpen. Taking Timlin out after 2 outs after already using 3 other relievers in a game that is tied in the 11th inning... I'm not sure if any manager makes that move. More importantly, I'm not sure the managers are wrong in not making the move. Extra inning games are often tricky with regards to bullpen management.

All that being said, I don't want this to digress from the conversation being held. There are managers who do manage strictly to the save rule (I don't believe Francona is one) and/or the Win Rule. Which likely makes pitching strategy less than optimal. However, I don't think the rule is to blame... I think managers (who make decisions by the rule), Front offices (who pay more money to those that rack up large save/win numbers) and Players (who want "set" roles) are to blame.

[Rudy Pemberton]

Setup guys, despite having the more difficult job, are not well compensated because there's no effective way to measure their contributions.

Is the role of set up guy physically more demanding than that of closer? Is there a higher rate of injury for set up man than closer? It certainly seems like there is a lot more volatility in that role; the top set up guys change year to year which isn't as true with closers. If true, it could be because of who teams select to put in these roles or just the demands of the position (warming up a lot, more appearances, irregular work, etc.) but maybe something to think about.

I don't buy that there's no effective way to measure the contributions of setup guys, though, although here may be no effective way to predict it.

[Zupcic Fan]

I think it's just silly to make that point about the Sox losing the game in the 11th inning because of that decision.
The game was also tied in the bottom of the 9th and the bottom of the 10th, when he went with Corey and another pitcher who escapes my mind, but who I also had little confisdence in.
He could just as easily have put Papelbon in then.
At which point the end result would have been that Papelbon would have pitched the 9th and maybe the 10th. The sox would not have scored, and Papelbon would have accomplished nothing. We'd still be facing the 11th inning.

It's pretty easy after the fact to wait until Tampa scores and claim that if Papelbon had been pitching in that particular inning, the game would not have been lost.
Francona, however, does not have in his head at the time the certainty about which inning the run will score.

The Papelbon decision was much more arguable in the following game----but that of course has nothing to do with the save rule and is for another topic.

Edited by Zupcic Fan, 28 April 2008 - 09:46 AM.

[Worst Trade Evah]

It affects how managers use their pitchers right now. In the Buchholz home run thread OCD SS made a great point that many pitchers are accustomed to their defined roles. If Paps comes into a game with a 1 run lead, he probably expects to close it, which is not always the situation with highest leverage. If Francona weren't restricted by clubhouse roles and keeping players happy, maybe he brings Paps into a game in the 8th with two runners on in a one run game, only to have someone elso close the game in the 9th. Th fact of the matter is, as long as there is a strict definition for what constitutes a save, pitchers are going to desire that role. Even if that role isn't always the most effective one. If you redefine or eliminate the stat I guarantee that the more intelligent managers/organizations will change their usage patterns. Like all good ideas, this will lead to a trickle down effect in MLB.

edit: what Frat said

I guess this is basically true -- players like routine, and like knowing when and how they'll be used. In general, people would rather know than not know.

But players also follow the money. It wasn't that long ago that simply pitching in relief at all was the sign of a lesser pitcher, prior to the construction of the closer mythology -- and the money that followed it.

If someone sets up a Fireman Relief Ace Award with large cash prizes to the guys at the top of the WPA boards (or these get rolled into contracts, if possible), I guarantee players would take a whole new interest in the concept of leverage, and it would be easier for managers to do what they need. And players would quickly develop a sense of who gets used in which game situations -- usage would just track leverage instead of innings, and people intuitively know when the critical game situations are anyway. Roles change, incentives get realigned, and people adjust -- how many teams had LOOGYs 20 years ago?

A manager can try to buck the incentive system and impose his will, and that might work for some managers and some players. But it would be better if incentives were set up to align team and player interests.

Edited by Worst Trade Evah, 28 April 2008 - 09:55 AM.

[Smiling Joe Hesketh]

It's pretty easy after the fact to wait until Tampa scores and claim that if Papelbon had been pitching in that particular inning, the game would not have been lost.
Francona, however, does not have in his head at the time the certainty about which inning the run will score.

Francona damn well had it in his head about when to use his ace reliever. He made a deliberate decision to hold back his ace reliever until the team had a lead...a situation which never arrived. It's a stupid way to manage, to allow a crappy reliever to lose a game for you instead of using your best pitcher in the highest leveraged situation, one where a single run costs you the game. And he made that decision because he and other MLB managers have determined that a save situation is more important than one where the game hangs in the balance. And that's exactly what I'm arguing against.

I would argue that in extra innings, perhaps the most effective way to manage is to use your bullpen backwards: start with the ace guy and work your down down the list until you've reached Timlin/Tavarez/Snyder levels of pitchers. This would give your club the best chance to win a win game in extra innings; by using your best pitchers first in order to maximize the chances of both holding the other club off the scoreboard and allowing your team a chance to score.

[TheBenzingerGame]

"To earn a save, the reliever had to come in with the tying or winning run on base or at the plate and finish the game with the lead."

Did the team have to have a lead at the time the reliever entered the game?

If the team did not have the lead when the reliever entered, but he went on to 'finish the game with the lead,' I think that'd be considered a good ol' fashioned win, no?

[Zupcic Fan]

sjh: I don't agree with you about the reason why he did it. Granted, you guys know far more than I do about this kind of thing, but my guess is that managers shy away from putting closers into tie games (especially on the road) far more because they are watching their overall number of innings over a course of a year than because of the save rule. My guess is that their thinking is that if they keep throwing them into tie games, plus games that are more typical save games, that their total number of innings eventually will become a problem in September.

In the game you're talking about, I'm sure that you had the same feeling (at least I did) about Corey pitching in the tie game. Would you have put Papelbon in then? If so, you would have wasted an innning or two off his arm to no gain.

So in general, managers tend to avoid putting closers of that caliber into tie games until far later in the season and hope that their weaker pitchers can get through an inning or two.
And I don't think that's a bad thing. The way the Sox have been hitting the ball lately, the most likely result would have been that I would have had to stay up a bit later to see Timlin give up the run in the 12th or 13th.

I'm a bit more confused about how rarely Oki has been used, quite frankly. But then, I do remember that he seemed far less sharp to me as last year came to a close, probably for exactly the reason we're talking about.

Edited by Zupcic Fan, 28 April 2008 - 10:21 AM.

[MannysDestination]

Francona damn well had it in his head about when to use his ace reliever. He made a deliberate decision to hold back his ace reliever until the team had a lead...a situation which never arrived. It's a stupid way to manage, to allow a crappy reliever to lose a game for you instead of using your best pitcher in the highest leveraged situation, one where a single run costs you the game. And he made that decision because he and other MLB managers have determined that a save situation is more important than one where the game hangs in the balance. And that's exactly what I'm arguing against.

I would argue that in extra innings, perhaps the most effective way to manage is to use your bullpen backwards: start with the ace guy and work your down down the list until you've reached Timlin/Tavarez/Snyder levels of pitchers. This would give your club the best chance to win a win game in extra innings; by using your best pitchers first in order to maximize the chances of both holding the other club off the scoreboard and allowing your team a chance to score.

That may be your observance of Tito's rationale, but that doesn't make it what happened. I agree with Zupcic's point regarding Papelbon's usage, and also being able to lock down the lead if it was obtained. I also agree that pitching your pen backwards in extra innings seems better at first blush, but there's also the possibility that it turns into a 15 inning affair and towards the end you're forced using Tavarez in the most highly leveraged situation because you've burned all your other options.

Regardless, from a pragmatic point of view, the only way there is change in baseball is if there is some kind of clamor for that change. There is very little, if any, chance it would be changed because there are too many vested interests in it staying the same. Players like it, front offices like it, agents like it, fans like it (at least the vast majority) so.... why would they ever change it? Then there are implications for the multi-billion dollar fantasy sports industry (after all, it's one of the traditional 5x5 categories), and the hall of fame already has closers in it (which may be reason enough in the eyes of "traditionalists" for it to never change).

I think scrapping saves and replacing it with holds would be the most practical way of replacing the stat. Hell, to satisfy the players (union), you could even have the official scorer "rank" holds in a game on a tier system based on leverage... but the resistance to change in baseball probably prohibits even this relatively lateral movement. A more robust "hold" stat while removing saves would free managers to use RP in the highest leverage situations (especially if you can be granted a hold in a tie game).

[Kevin Youkulele]

I think scrapping saves and replacing it with holds would be the most practical way of replacing the stat. Hell, to satisfy the players (union), you could even have the official scorer "rank" holds in a game on a tier system based on leverage... but the resistance to change in baseball probably prohibits even this relatively lateral movement. A more robust "hold" stat while removing saves would free managers to use RP in the highest leverage situations (especially if you can be granted a hold in a tie game).

This is not a bad idea. Maybe relievers should even get holds for preserving one-run deficits, as these are also hi-lev (albeit not as much as ties).

[Eric Van]

I severely doubt the save rule will ever be retired or significantly changed. It has too much history attached to it. In the best-case scenario, it is superseded with a better stat so that the save becomes a relatively unimportant curio.

So there is no reason to wait. The sabermetric community should get together this year and reach a consensus on what constitutes a "rescue."

Rescue obviously needs to be based on WPA. (I know I'm stealing some of this from somebody, probably Tom Tango.)

We need to decide the following:

1) Which WPA values should we use? Should they be park-adjusted? Once we have a methodology for calculating WPA based on the actual hitters faced, should we use that?
2) How much WPA does a reliever need to earn a rescue?
3) Should the official scorer be given leeway in cases where a pitcher is close to the dividing line? For instance, a pitcher might be given a rescue if he missed by .005 WPA and had to retire Papi and Manny in Coors, or denied a rescue if he only qualified by .005 and got the bottom of the order out on a cold day with the wind blowing in hard.
4) Is there any logical reason to limit rescues to pitchers who enter with a lead, or with the score tied?
5) Is there any logical reason to only hand out one rescue per game?
6) Is there any logical reason to only give a rescue to pitchers on the winning team?

My answers:

1) WPA tables for each year should be calculated using Markov chain methodology based on the average offensive level in MLB of the last three years [we can debate the time frame]. One table shall be used assuming a neutral ballpark and conditions. In cases where the eventual actual offensive level proves to differ significantly from the level of the last three years (defined as differing from the average by more than twice [we can debate the multiplier] the standard deviation of offense over the last ten years [we can again debate the time frame]), the WPA tables shall be recalculated at the end of the year and all rescues re-evaluated. MLB / the sabermetric community shall each year designate the party responsible for the creation of the tables. [Obviously, Tom Tango currently.] [We also have to decide whether tables used to retroactively assign rescues for past seasons should be calculated with the same logic, for consistency, or based on the actual offense level of each season, for possibly better accuracy. I would vote for consistency.]

2) I would pick 1/9, i.e., .111, but this is highly subject to debate.

3) Yes, the scorer has leeway. Again, we have to determine how much leeway in terms of WPA points (we need to find examples of games with extreme conditions like the ones cited which were close to the agreed-upon limit; this can be done by folks retroactively assigning rescues to past seasons). The scorer shall note in detail the logic for any such decisions should a re-evaluation become necessary. Once we have WPA values that more accurately reflect specific game conditions, the scorer shall be instructed to consult those in leeway cases.

3) Rescues for guys who entered with the team trailing. If three innings of scoreless relief with your team down a run is as valuable as a scoreless 9th, why not reward it? What's the argument for making the scorer check the line score instead?

5) Multiple rescues per game. Why not?

6) Sure, rescues for pitchers who labored in vain. Again, why not?


Sox 2008 Rescues by the above rule:

Papelbon 4
Lopez 3 (2 in losses)
Okajima 1 (opening day, when Papelbon got a save but not a rescue)

[BCsMightyJoeYoung]

Sox 2008 Rescues by the above rule:

Papelbon 4
Lopez 3 (2 in losses)
Okajima 1 (opening day, when Papelbon got a save but not a rescue

The main problem with introducing new stats is that the average fan has to be able to understand how it's calculated. This has always been the problem with the NFLs QB rating. Unfortunately, any stat that accurately reflects a relief pitcher's value is , inherently, going to require more math than the average fan can tolerate. I mean, just look at the resistance to stats like OPS or OBP. Runs Created? Total Average? All this stuff has been around for decades and they are still not accepted.

I think ERA is just about as complicated as it can get before people start ignoring it.

Maybe they should just have an award at the end of the year that they can vote on - a "Relief Pitcher of the Year" award or something. That will generate some debate on reliever's usage and value I would think,. It's sort of what the NHL does when it gives separate awards for team defence - the lowest Goals Against Average and the Vezina which is voted on and goes to the best goalie.

[glennhoffmania]

Smart baseball organizations need to break out of the "save him for a lead" mentality of the closer's job and use their best pitchers in the tightest situations. While this idea is hardly new (Bill James has been arguing for this for years), perhaps changing the definitions of a save will allow managers to use their ace relievers more wisely.

This is what bothers me the most. I think we all consider the Sox to be a smart baseball organization, yet they continue to follow this closers strategy. I don't care if they don't change the rule- in fact, I'd prefer that they don't. Instead, let all of the other teams stick to this mentality while the Sox realize it's stupid and starting using their pitchers in the most appropriate situations. Yeah, the save rule sucks, but no one is forcing Francona to manage based on it.

[thrawnqq]

Maybe have a new metric called clutch saves or something.

How about a new metric called "Clutch Outs" which are outs that relief pitcher records with the game tying or go ahead run on base or at bat. A nice counting stat that would make a lot of sense to non-SABR types, and could have applicability at anytime during a game.

Todd

Edited by thrawnqq, 28 April 2008 - 12:02 PM.

[Kevin Youkulele]

How about a new metric called "Clutch Outs" which are outs that relief pitcher records with the game tying or go ahead run on base or at bat. A nice counting stat that would make a lot of sense to non-SABR types, and could have applicability at anytime during a game.

Todd

This certainly has simplicity going for it. What about self-induced clutch outs, i.e. the pitcher enters with a 2 run lead, walks a guy and is now facing the tying run, and then gets 3 outs to escape? Should you get credited for being clutch after being un-clutch enough to get into a tight spot?

Perhaps also "tying run on deck" is clutch enough?

[thrawnqq]

This certainly has simplicity going for it. What about self-induced clutch outs, i.e. the pitcher enters with a 2 run lead, walks a guy and is now facing the tying run, and then gets 3 outs to escape? Should you get credited for being clutch after being un-clutch enough to get into a tight spot?

Perhaps also "tying run on deck" is clutch enough?

I dont know, seems like it has the benefit of simplicity, either way. Maybe the pitcher has to enter the game in Clutch Out situation, and then all outs are counted from then on.

Thus removing the self-induced clutch situation.

[BG913]

I dont know, seems like it has the benefit of simplicity, either way. Maybe the pitcher has to enter the game in Clutch Out situation, and then all outs are counted from then on.

Thus removing the self-induced clutch situation.

Well, we track BB and IBB separately, as well as R and ER - no reason you couldn't track COs and SICOs (Self-Induced Clutch Outs) - which also would get pronounced "Sickos" and would describe the feeling you get watching a high SICO closer (remember Armando Benitez - my guess is he'd be the SICO king). A SICO rate would be SICO/CO...

[syoo8]

A relief pitcher cannot be credited with a save if he does not finish the game unless he is removed for a pinch hitter or pinch runner.

How could a relief pitcher be removed for a pinch hitter or runner? This makes no sense to me. By definition, the person with a save throws the last pitch for the last out, no?

[Eric Van]

The main problem with introducing new stats is that the average fan has to be able to understand how it's calculated. This has always been the problem with the NFLs QB rating. . . Unfortunately, any stat that accurately reflects a relief pitcher's value is , inherently, going to require more math than the average fan can tolerate.

I think the average fan would understand that if you have, say, 6 Rescues and you get 1 more, that makes 7. That's how the actual stat is calculated.

Would folks really have a problem with "the experts have a formula that determines who gets a rescue?" There are plenty of sports where expert judges hand out the points.

Besides, by the time this would become widely accepted (10 years? 20?), I would expect the current Win Probability to be shown onscreen in the score / count / pitch speed line or box. You may see that within the next few years. The WP is real easy to explain and understand intuitively, and telling folks that rescues are given out according to the impact on WP would be enough for most fans. Some would even learn the threshold number (11% or whatever) and be able to say in advance whether the pitcher was entering in a rescue situation or not.

[Fratboy]

The WP is real easy to explain and understand intuitively, and telling folks that rescues are given out according to the impact on WP would be enough for most fans. Some would even learn the threshold number (11% or whatever) and be able to say in advance whether the pitcher was entering in a rescue situation or not.

Something tells me that MLB will be reluctant to drop a pile of money in Tom Tango's lap to use WPA as a component of one of its official statistics. Not that he'd mind, of course.

Though WPA can be easily understood, the calculation of it is too obscure to be easily accepted by the powers that be and the public at large. "Oh, but nobody knows how QB rating is calculated," you say. MLB isn't the NFL, and when you hear people discuss quarterbacks, you're not really talking about the rating that much.

[thrawnqq]

MLB isn't the NFL, and when you hear people discuss quarterbacks, you're not really talking about the rating that much.

When they do, they ridicule it for being an unfathomably complex statistic.

[Eric Van]

Something tells me that MLB will be reluctant to drop a pile of money in Tom Tango's lap to use WPA as a component of one of its official statistics. Not that he'd mind, of course.

Though WPA can be easily understood, the calculation of it is too obscure to be easily accepted by the powers that be and the public at large. "Oh, but nobody knows how QB rating is calculated," you say. MLB isn't the NFL, and when you hear people discuss quarterbacks, you're not really talking about the rating that much.

MLB adopting the rescue would be 20 years down the road. That was the whole point of my proposal--screw MLB. The sabermetric community should get together now and define the stat to the absolute best of our ability, then track and disseminate it widely.

The next time a Borowski leads the league in saves but is 28th in rescues, you can bet cash money that Rob Neyer will do a column on the superiority of the new stat. Twenty years of that will go a long way to fix the problems you speak of.

[BG913]

I think the average fan would understand that if you have, say, 6 Rescues and you get 1 more, that makes 7. That's how the actual stat is calculated.

Would folks really have a problem with "the experts have a formula that determines who gets a rescue?" There are plenty of sports where expert judges hand out the points.

Besides, by the time this would become widely accepted (10 years? 20?), I would expect the current Win Probability to be shown onscreen in the score / count / pitch speed line or box. You may see that within the next few years. The WP is real easy to explain and understand intuitively, and telling folks that rescues are given out according to the impact on WP would be enough for most fans. Some would even learn the threshold number (11% or whatever) and be able to say in advance whether the pitcher was entering in a rescue situation or not.

A compromise might be to translate "WPA added" into the situations that drive it, and use the situations as the criteria, not the actual increase in WPA. My guess is you could arrive at a menu of baserunner/out combinations in the 7th, 8th and 9th that are all typically high WPA and establish those as rescue eligible situations...

So that a pitcher who enters a game from the 7th inning on in a game closer than 3 runs is "rescue eligible" if they do the following things (whatever those are - which typically result in a WPA over .111).

That way the fan, player and manager would know during the game whether a situation was rescue eligible...the inability to do that seems like one of the disadvantages of something more arcane that would slow its adoption. It's slightly more complex than the current save rules (roughly analagous to the infield fly rule, which is also situation dependent) but more useful and would actually incent managers and players to use resources most appropriately. And true diehards would love lording their understanding of the "rescue rules" over casual fans.

Though I still find myself emotionally attached to the "SICkO" , this is a lot more useful...

[Joe Sixpack]

How could a relief pitcher be removed for a pinch hitter or runner? This makes no sense to me. By definition, the person with a save throws the last pitch for the last out, no?

No, not according to the old rule. A relief pitcher could come into the game in a save situation (not in the 9th inning, obviously), get pulled for a PH, and still get a save despite not finishing the game.

[RingoOSU]

No, not according to the old rule. A relief pitcher could come into the game in a save situation (not in the 9th inning, obviously), get pulled for a PH, and still get a save despite not finishing the game.

But the losing team always gets to bat last, if the lead does not change.

Edit: Are you saying once the PH is replaced by a pitcher, he still gets the save?

Edited by RingoOSU, 28 April 2008 - 04:13 PM.

[Eric Van]

A compromise might be to translate "WPA added" into the situations that drive it, and use the situations as the criteria, not the actual increase in WPA. My guess is you could arrive at a menu of baserunner/out combinations in the 7th, 8th and 9th that are all typically high WPA and establish those as rescue eligible situations...

So that a pitcher who enters a game from the 7th inning on in a game closer than 3 runs is "rescue eligible" if they do the following things (whatever those are - which typically result in a WPA over .111).

That way the fan, player and manager would know during the game whether a situation was rescue eligible...the inability to do that seems like one of the disadvantages of something more arcane that would slow its adoption. It's slightly more complex than the current save rules (roughly analagous to the infield fly rule, which is also situation dependent) but more useful and would actually incent managers and players to use resources most appropriately. And true diehards would love lording their understanding of the "rescue rules" over casual fans.

Though I still find myself emotionally attached to the "SICkO" , this is a lot more useful...

I think this translation of WPA into the situations is desirable and can be promulgated alongside the actual WPA criterion.

For instance, pitching the 9th, it's a rescue situation when there is a:

1 run lead, always
2 run lead, if there's a runner on with less than 2 outs or 2 RISP with 2 outs
3 run lead, if there are 2 runners on with less than 2 outs, or a runner on 3rd with 0 outs
4 run lead, if the bases are full with less than 2 outs, or 2 RISP with 0 outs

Four out saves with a 2-run lead are always rescues even if you enter with nobody on base.

Two inning saves with a 3-run lead can be rescues if you enter with a runner on and 0 outs, 2 runners on and less than 2 outs, or the bases full (any # of outs).

And on and on endlessly. You need the WPA rule because there are too many combos, but when you do break it down it always appears to be quite logical.

People would eventually learn a lot of this . . .

[thrawnqq]

On comment that has made me think is the situation when a reliever comes in for a team that is losing 2-1, and pitches and inning or two, and keeps the score 2-1. That would seem to be a valuable outing, keeping the score close, but is it valuable enough to incorporate into a new relief pitcher metric?

[OttoC]

Are there any examples of the definition of a statistic in baseball actually being changed?

There are a number. The sacrifice/sacrifice fly rules went through tumultuous changes. In 1889, advancing a runner by fly or ground ball was counted as a sacrifice. The number of balls for a walk changed, being as high as nine. One year, bases on balls were scored as errors against the pitcher. One year, bases on balls were counted as hits. There were many, many rules changes before things settled down to what we now know as baseball.

[drtooth]

We can talk stats and rule changes all we want. What it will take (in the end) is a manager willing to think outside the box in regards to the modern day use of the bullpen. The problem here is that if a manager does think a bit outside of the box and has the horses to pull it off (see LaRussa with the usage of his pen in Oakland and St.Louis), he considered a genius and an innovator. If the manager tries to think outside the box and does not have the players, his is ripped by the media/fans and is fired without too much hope of finding future employment. I'm not sure many baseball lifers are willing to take that risk. You will need to have a manager with enough cache/job security to be able to make the drastic change as to the use of the pen and the closer's role.

Edited by drtooth, 28 April 2008 - 07:03 PM.

[thrawnqq]

Good point. And God forbid they bring in a 'closer' in the 6th, who saves the game in one of these high leverage situations, and then another pitcher blows the lead in the 9th.

[Eric Van]

On comment that has made me think is the situation when a reliever comes in for a team that is losing 2-1, and pitches and inning or two, and keeps the score 2-1. That would seem to be a valuable outing, keeping the score close, but is it valuable enough to incorporate into a new relief pitcher metric?

Here's a scenario:

Down a run, Joe Setup pitches a scoreless top of the 7th and 8th.

Home team scores 3 in the bottom of the 8th and Freddy Closer earns the easy save.

But who was actually more valuable? Joe Setup has a WPA of .121 and would earn a rescue. Freddy Closer has an .082 and would not earn one (although we might want to count WPA from c. 050 to .111 as something like "job well done").

[Joe Sixpack]

But the losing team always gets to bat last, if the lead does not change.

Edit: Are you saying once the PH is replaced by a pitcher, he still gets the save?

Hypothetical situation:
Pitcher A pitches the top of the 8th with his team up 2-1 and completes a scoreless inning. Pitcher A's spot in the lineup comes up in the bottom of the 8th and he is lifted for a pinch hitter. Pitcher B pitches the 9th and finishes the game.

Under that rule, Pitcher A could have been credited with the save.

[mabrowndog]

Here's a scenario:

Down a run, Joe Setup pitches a scoreless top of the 7th and 8th.

Home team scores 3 in the bottom of the 8th and Freddy Closer earns the easy save.

But who was actually more valuable? Joe Setup has a WPA of .121 and would earn a rescue. Freddy Closer has an .082 and would not earn one (although we might want to count WPA from c. 050 to .111 as something like "job well done").

In that scenario, Joe Setup gets the win.

[Eric Van]

In that scenario, Joe Setup gets the win.

Indeed. In this system you ignore W/L rather than adding them to rescues and whatever the opposite stat is. You can probably get a rescue and a loss as well: a few scoreless IP in a tie game, leave a guy on 1B with 2 outs, next reliever allows him to score.

"Rescue" is perhaps not the right term if it's going to include pitching well in tie games and when trailing.

[Vermonter At Large]

Well ... this is a very interesting discussion, but the notion of using WPA to determine the leverage of the situation to a three-decimal-point value, and re-calculate totals at the end of the year, etc really loses touch with reality. It doesn't have to be that complicated.

You don't need WPA to define high-leverage situations. That could be defined rather easily without the use of Markov chains, and you could add the high-leverage "rescue" as a separate stat, incorporated or in-fact replacing "hold", or incorporated into "save." Or you could just do away with all of them and establish a point system for relief appearances (sort of a degree of difficulty, using whole numbers (say, 0-5) and total them all up per IP.

Then, after doing this, we can sit back and watch the great experiment in human nature to see if managers ultimately change their tactics based on the change in statistics ...