1. Putting a RH reliever on the mound and when I bring in the LH reliever, stashing the RH reliever at 1B or in RF and swapping the two batter by batter. Yes, I think this is overkill and would never really advocate it for a real team, but it's something I've been experimenting with.

One problem with that is that you have to remove your starting RF or 1B, who is probably one of your best hitters.

There are a helluva lot of books/movies/TV shows/etc. etc. etc. in the world to potentially consume. As one way of making that manageable, I think many of us have decided that certain genres are not for us. I'm sure it's often an overgeneralization in the sense that there's probably something in the genre that the person writing off the genre would in fact enjoy. But all the same, it's a thought process that I think most of us utilize. James comes off grouchy in this quote because he phrases it in a way that's derogatory to the author, rather than the work. If you don't like horror movies, chances are you would say something like "I think horror movies tend to be poorly written." But if you instead said "I think the writers who write horror movies usually can't write", you'd really be making the same point, and I'd like to think that's what James is doing here.

Only it certainly doesn't come off that way. Generally speaking, I think fantasy movies and musicals are best appreciated by overaged teenagers and menopausic women, but if someone asked me about a specific movie from one of those two genres, I'd just say "I haven't seen it" rather than trashing it sight unseen. And in this case, the reputation of My Baseball Diary among those who've read it is such that it compounds the stupidity of James's comment.

There are a helluva lot of books/movies/TV shows/etc. etc. etc. in the world to potentially consume. As one way of making that manageable, I think many of us have decided that certain genres are not for us.

It reminds me of the way I react to Country Music. I always say I don't much care for it, then I'll say, but, of course, I love Hank Williams.

One problem with that is that you have to remove your starting RF or 1B, who is probably one of your best hitters.

(Setting aside the issue of whether this is against the rules.) That is true, but it's the late innings, and your starting RF or 1B will maybe get up once more if at all unless you go to extra innings, and I just don't think the marginal utility of that one PA is worth much. Especially if you have a competent PH available.

Stated more clearly: Anyone can hit anything in one PA. (Or in 50, but I digress.)

There are real gains to be had with liberal platooning.

The evolution of the game over the past 20 years or so tells us that this almost certainly isn't true.

Guy - what's your reasoning? The bullpen specialization? That doesn't kick in until the 7th inning.

Ray: My reasoning is that, through trial-and-error, teams have evolved over the past 20-30 years in the direction of expanding pitcher rosters at the cost of position players. Teams seemed to find that the advantage of the marginal pitcher was greater than an extra position player. This has happened even though it results in less platooning on offense (I don't know if anyone has quantified the decline of platooning, but certainly fewer position players makes platooning much more difficult).

Guy, I don't see that as a reason; I see it merely as a statement of what teams have done. One does not necessarily support the other.

Ray: My reasoning is that, through trial-and-error, teams have evolved over the past 20-30 years in the direction of expanding pitcher rosters at the cost of position players. Teams seemed to find that the advantage of the marginal pitcher was greater than an extra position player. This has happened even though it results in less platooning on offense (I don't know if anyone has quantified the decline of platooning, but certainly fewer position players makes platooning much more difficult).

I think you're saying--correct me if I'm wrong--that teams have all moved in this direction because they believe it is in their best interest, and we should take that seriously? That's true. Of course they may be mistaken about their best interest, so we don't want to assume that teams move teleologically toward more efficient strategies, since that may or may not prove true.

I don't see that as a reason; I see it merely as a statement of what teams have done. One does not necessarily support the other.

Yes, it does. Why else have teams added pitchers, if it didn't provide a competitive advantage? That isn't to say teams necessarily deploy resources perfectly. But MLB teams are, if anything, change averse. They won't change unless they need to. Yet something caused teams to evolve, over time, from 10-11 pitchers to 12-13 pitchers. Clearly, teams were being rewarded as they added pitchers. (Up to a point -- I don't expect to see this trend continue.)

And empirically, I think finding effective platoon partners is harder than it first appears. The real question is how many LHH starters can be replaced by a better RHH option against LH starting pitchers? My local team, the Nats, have 3 LHH starters. Espinosa is pretty crappy against LHPs, but Lombardozzi vs. LHP has about the same OPS+ and gives you a worse glove. Moore is probably a better hitter vs. LHP than LaRoche, but a much worse glove at 1B. Moore might also outhit Harper, but do you want to take PA away from Harper or let him learn to hit LHPs as well as possible? In all 3 cases, I doubt the Nats gain from using a platoon.

Is the arithmetic different for many other teams/positions? Maybe so, I haven't really looked. But my guess is there are at most a few opportunities that aren't already being exploited.

Yes, it does. Why else have teams added pitchers, if it didn't provide a competitive advantage? That isn't to say teams necessarily deploy resources perfectly. But MLB teams are, if anything, change averse. They won't change unless they need to. Yet something caused teams to evolve, over time, from 10-11 pitchers to 12-13 pitchers. Clearly, teams were being rewarded as they added pitchers. (Up to a point -- I don't expect to see this trend continue.)

because they mistakenly believe it gives them a competitive advantage?

I wonder if the word "evolution" isn't doing too much in your prior comment. Evolution does necessarily imply improvement.

Yes, it does. Why else have teams added pitchers, if it didn't provide a competitive advantage?

That's evidence, perhaps, that they _think_ this is better, but it's not evidence that it actually is better.

I reject this type of justification.

That's evidence, perhaps, that they _think_ this is better, but it's not evidence that it actually is better.
I reject this type of justification.

That's absurd. These are professionals operating in a highly competitive environment, with enormous financial rewards for those who succeed. When we observe changes in behavior, it is extremely likely that teams are being rewarded for that change. After 1993, SBAs steadily declined. This was almost certainly an evolutionary change, in response to an increase in HRs. Few if any teams sat down and calculated how many fewer SBA they should make. But over time, managers got tired of seeing CS followed by solo HRs, and they sent runners less often. In the NFL, teams are increasingly relying on the pass rather than the running game. This is pervasive, for the simple reason that it works. In the NBA, teams are taking fewer long-2s -- again, because it works.

Can you give me an example of a similar evolutionary change that was NOT productive, that was the wrong move and yet was implemented across an entire league?

That's absurd. These are professionals operating in a highly competitive environment, with enormous financial rewards for those who succeed. When we observe changes in behavior, it is extremely likely that teams are being rewarded for that change. After 1993, SBAs steadily declined. This was almost certainly an evolutionary change, in response to an increase in HRs. Few if any teams sat down and calculated how many fewer SBA they should make. But over time, managers got tired of seeing CS followed by solo HRs, and they sent runners less often. In the NFL, teams are increasingly relying on the pass rather than the running game. This is pervasive, for the simple reason that it works. In the NBA, teams are taking fewer long-2s -- again, because it works.

Your model seems to take for granted that businesses never make mistakes because it is never in their interest to make mistakes.

because they mistakenly believe it gives them a competitive advantage?

If it were mistaken, someone would have switched back by now, because they'd have understood that the competitive advantage was illusory and that there was an advantage to be gained by switching back. Yet this model of pitcher (and by extension, total roster) usage has remained remarkably steady for a number of years; basic starter and late-inning bullpen usage has barely changed since about 1998, after two decades of fairly rapid change.

It is hard to find position players who can be consistently productive in 150-200 PAs. It's less hard to find pitchers who can be consistently productive in 40-50 IP, and that group of pitchers help extend the life of your starters, who can work 210 innings per year instead of 240. There are a combination of factors that come together here, not all of which are necessarily on-field effects.

Adding pitchers can't go much further because of roster size limits, and because there are only so many innings to go around. But it strikes me that the current usage model does a relatively good job in distributing the highest-leverage innings to the best pitchers, without concentrating too many of them into two or three pitchers and not pushing tired arms when it isn't necessary. That's not necessarily saying that it couldn't be better, but there doesn't seem to be any incentive for someone to try to make it better.

-- MWE

These are professionals operating in a highly competitive environment, with enormous financial rewards for those who succeed.

They're herd and trend followers. That's why every team has a "closer" and a LOOGY now, too.

Can you give me an example of a similar evolutionary change that was NOT productive, that was the wrong move and yet was implemented across an entire league?

Overcoaching and overemphasis on defense in the NHL, which runs directly counter to the principle that teams with a distinct talent advantage (cough, Rangers, cough) should want a riskier game, not a more conservative one. The Rangers are near the top of the league in blocked shots and constantly harp on being "disciplined in your own end." Every other team preaches the same thing. It's stupid.

If it were mistaken, someone would have switched back by now, because they'd have understood that the competitive advantage was illusory and that there was an advantage to be gained by switching back.

This seems to make intuitive and logical sense, but it isn't really the way the sports (or the real) world works. There's more money and job security in being wrong like everyone else is wrong, as opposed to being alone in being right.

That's absurd. These are professionals operating in a highly competitive environment, with enormous financial rewards for those who succeed. When we observe changes in behavior, it is extremely likely that teams are being rewarded for that change.

Want to buy a credit default swap?

It is hard to find position players who can be consistently productive in 150-200 PAs. It's less hard to find pitchers who can be consistently productive in 40-50 IP, and that group of pitchers help extend the life of your starters, who can work 210 innings per year instead of 240. There are a combination of factors that come together here, not all of which are necessarily on-field effects.

That's what I suspect. While the bloating pitching staffs aren't necessarily the optimal employment of resources at the single game level, they are more beneficial over the course of a season (or beyond) for the team as a whole.

Guy, there are many factors involved here. For example, look what happened some years ago when the Jays tried to go back to a 4-man rotation. There was a mutiny.

And managers are risk-averse so GMs often placate managers, etc.

Teams do less than ideal things all the time, for a number of reasons.

In DMB I can switch my pitchers between 1B and the mound. Now, it's probably not a good strategy anyway, but if it were, trying to do it in the majors would be more trouble than it's worth. Or substitute any non-traditional strategy. You need to sell the players on it.

Guy, there are many factors involved here. For example, look what happened some years ago when the Jays tried to go back to a 4-man rotation. There was a mutiny.
And managers are risk-averse so GMs often placate managers, etc.
Teams do less than ideal things all the time, for a number of reasons.

Hey, no fair -- this is MY argument! Seriously, I think inertia is a powerful force, and that teams/leagues tend to be change-averse. (SBB's notion that MLB is driven by constantly shifting fashion trends is, of course, nonsense.)

So I have no trouble believing that inefficient practices can persist because no one identifies them, or because it takes time for someone to take the risk of changing it. And I believe teams may be too slow to take advantage of a new opportunity, as appears to be the case with the 3-point shot in the NBA. But in both cases, we are talking about a failure to act. What I don't believe happens -- or at least it's very rare -- is slow, steady, incremental change in a direction which is in fact counter-productive.

The 4-man rotation is actually a great example. Since teams are afraid to change, how did 4 days of rest ever displace 3 days rest as the norm in MLB? Making the switch meant being unorthodox (initially), and it meant you had to hire a 5th starting pitcher. The only plausible reason is that it worked -- it helped teams to win.

Part of the reason that bloated pens have won out over deep benches with platooning is that the now-prevalence of those pens kills a lot of the advantage platooning can give you. Want to bring out the lefty bat? - here's a fresh lefty arm to deal with him and now you've both lost your starter for the game and are unlikely to pinch hit again in the same at bat (and I was planning on using my LOOGY anyway). There's no similar *obvious* penalty to losing the ability to platoon by having the deep pen, you're still starting the guy you think is your best option in most situations... this despite extensive platoons being a more effective option than deep pens in a "neutral" league where teams otherwise do little of either.

Want to buy a credit default swap?

Cute, but not analogous. I'm arguing that teams, like financial institutions, respond to short-term positive and negative feedback. As teams added pitchers to the roster, the result was success more often than failure. So they kept it up, and others imitated the practice. Now in the case of some financial instruments, there was enormous tail risk. So the short-term profits led to a bad outcome. But what is the equivalent "tail risk" for MLB teams? Each season is self-contained, and this has worked for c. 30 years. Even if sometime in the future a change in the game makes the strategy stop working (e.g. a rule change that forces each pitcher to face at least 3 batters), then teams will simply change their rosters then. They won't retroactively lose a bunch of games they won over the prior decades. (Well, maybe the Mets could figure out how to do that.....)

#69 You're going to have to work to convince me that it's a better use of the roster than the Earl Weaver model.

Weaver's track record with health of staff matches up with anybody's. Further (and contrary to Mike's point) he had absolutely no problem finding position players who could be productive in limited playing time. Precisely because he wasn't looking at what a player could not do (Lowenstein for instance wasn't much of an infielder and couldn't hit lefties)

Weaver also let pitchers call their own game and that may well have contributed to the ability to handle heavy workloads and stay generally healthy. A successful practice that was resisted by catchers (in particular it drove Rick Dempsey nuts)

Yes, it's not something every manager (or management team can do). Doesn't mean it wouldn't still work (though there's no chance of getting the roster spot that went to the 5th starter)

As to why things evolved to the current state, I suspect it has a lot less to do with optimal roster use than man management. Position players really don't like to be platooned. Marginal pitchers like to pitch in roles where they can be successful.

Ray as I've mentioned before I knew the 4 man roster was dead when Odell Jones pitched a fit because the Expos AAA team went to a 4 man rotation for a while. Guys like Jones know that being perceived as a team player is vital in getting a major league spot. Not much separates the guys looking for the spots at the end of the staff. Yet Jones was willing to take a public stand on the issue.

This seems to make intuitive and logical sense, but it isn't really the way the sports (or the real) world works. There's more money and job security in being wrong like everyone else is wrong, as opposed to being alone in being right.

The point I was making is that for 20 years, from 1976 through about 1996, teams *were* switching, virtually every year. Since then, the pace of change has not only slowed, it's essentially stopped.

What happened during those 20 years that pushed teams to change, if we assume (I believe correctly) that sports teams tend to avoid change, all else being equal?

Mike Marshall breaking down after pitching 106 games in 1974 made teams aware that they couldn't keep loading innings onto their ace reliever, forcing teams to look for ways to split the load. The DH, after initially helping to drive starting pitching loads back up because teams realized they didn't have to pinch-hit for a starter, added some flexibility in the AL, where teams could remove a pitcher before things got totally out of hand because they didn't have to pinch-hit for the guy they brought in, and eventually the NL followeds suit (aided to some extent by the double-switch, which was rarely a big deal before NL teams started following the AL's lead on pitching changes). Teams started to realize that it didn't make sense to try to find five guys who could got 8-9 innings every fifth day when they could get a bunch of relievers who could go an inning or two three-four times a week. Expansion played a role, certainly. Money started to play a role beginning after the 1981 strike which essentially cemented free agency. There were a lot of factors - some on the field, some off the field - that drove change.

But the one most obvious change is this - teams started to bring relief pitchers into games at the start of an inning, rather than waiting for trouble to occur. Before 1975, it was relatively rare to see a pitching change at the start of an inning when the preceding pitcher had not departed for a pinch-hitter. By the end of the 20-year makeover, it was common. I think the biggest reason for that, quite honestly, is that teams have come to believe that when a relief pitcher knows when he is coming into the game, he can be better prepared (and will pitch better as a result).

One thing that has happened is that teams have become less likely to lose a late-inning lead, when you account for changes in offensive environment over the years - it's not that teams lose fewer leads overall, mind you, but that (a) run scoring has generally increased since the mid-70s and (b) late-inning games have tended to be a little but closer, so you would have expected to see the rate of lost leads going up, all else being equal. I don't credit all of this to the pitching, since the flip side of more pitching is fewer bats, and that in turn means that some hitters bat in key late-inning situations who would have left for pinch-hitters in the past.

Anyway, I think that on balance the changes to the game have made it easier for managers to manage their personnel and to find roles for more of their roster - you have fewer guys sitting around for days on end not playing. Even if there were no net on-field impact otherwise, that would probably be a positive from a management standpoint.

-- MWE

#69 You're going to have to work to convince me that it's a better use of the roster than the Earl Weaver model.

Weaver's track record with health of staff matches up with anybody's. Further (and contrary to Mike's point) he had absolutely no problem finding position players who could be productive in limited playing time. Precisely because he wasn't looking at what a player could not do (Lowenstein for instance wasn't much of an infielder and couldn't hit lefties)

Other than the well-documented Roenicke-Lowenstein pairing, who were his other stellar platoons, with each side delivering great offense (not a gotcha, a serious question)? Obviously he got fantastic work from the Roenstein duo, but how common was that?

Earl was a great manager, of that there's no doubt. I'm not sure that his success refutes the idea that stocking up on relievers helps the other guys stay healthier, and thus has benefits outside the individual game.






I think it's pretty damn inarguable that pitchers are more likely

#72 All I can tell you is that tabletop managers generally opt for the extra position players.

And a point I don't see you addressing is that your lefty reliever has to face the first batter. He may be hot death to lefties, but having to face a righty in a high leverage situation is a pretty big deal.

Now if you're talking a good pitcher (say prime Mike Stanton) who just happens to be left-handed fine. But somebody like John Candelaria at the end of his career (1991 splits: .138/.206/.207 vs LHP, .354/.392/.600 vs RHP -- not precisely typical, but he did have a career 149 points of OPS platoon split and was always murder on lefties) could be made to face more righties than lefties simply because he had to face the first batter.

Mike Marshall breaking down after pitching 106 games in 1974 made teams aware that they couldn't keep loading innings onto their ace reliever, forcing teams to look for ways to split the load.

Marshall's a bad example. It wasn't workload related and wasn't an arm issue.

There are plenty of other relievers who burned out in the same time frame though. I certainly wouldn't object to Wayne Granger as an example (the Reds burned through a lot of relievers)

Eck's probably the guy who pointed the way to cutting back what was asked of a closer. Even then, in his great 4 year run he was averaging 68 GP per 162 team games (79 IP). I think it's fine to say that you don't ask your closer to handle a Wayne Granger workload (never mind a Mike Marshall -- though Marshall himself believed that there was no problem handling that kind of a workload), but there's no real evidence that they couldn't manage somewhere around 100 or so IP. (basically ask the relief ace to pitch a little more frequently in tie games -- don't use them almost exclusively as a designated save accumulator)

#77 He didn't platoon all that often. Thanks to the Baltimore farm system he generally had good regulars at most positions. And when he did platoon it wasn't always a straight platoon. Bill James points out that he'd use a complex setup with 5 guys splitting 4 spots and role guys (say Curt Motton) popping up in very specific situations.

Bobby Cox in Toronto is probably a better example of a guy who got a lot of mileage out of some fairly limited players by platooning them. Mullinorg being the best known, but he had a pretty productive platoon situation at catcher too. And did some platooning in the outfield and at DH.

Incidentally, something that undermines my example. While Weaver's teams were constructed so that he generally had good options when pinch-hitting, he didn't get great results from his pinch-hitters. In his books he mentions plenty of cases where he had the right guy available and it worked, but overall his pinch-hitters hit (roughly) .234/.326/.335

And maybe this helps explain the change. Those are actually pretty decent pinch-hitting numbers (in particular, the walk rate). A lot of guys who are fine hitters overall can't handle pinch-hitting. Something that happens in real life but not in a tabletop game.

but there's no real evidence that they couldn't manage somewhere around 100 or so IP

People say this all the time, and also that there's no evidence that a 4-man rotation couldn't still work. But it's hard-to-impossible to "prove" that something wouldn't work. I can't prove that it would be a bad idea for the Cardinals to move Matt Holiday to 2B, but there's lots of evidence that points in this direction. Similarly, there is a mountain of evidence suggesting today's pitchers could not be as effective with increased workloads. We know that starters do better on 4 days rest than 3 days rest, and better on 5 than 4. We know that starters throw harder, and are more effective, when they become relievers. And we know this is especially true if they have short outings in relief. In short, everything we know tells us that the less often you use a pitcher (within reason), and the fewer innings you ask him to pitch, the more effective he will be. This is perhaps the most important truth that baseball has discovered over the past 3 decades -- vastly more important than the importance of OBP -- and yet many fans continue to miss this simple truth.

And please, don't tell me what pitchers used to do back in the day. Today's pitchers simply throw much harder than pitchers of earlier eras, and the result is much higher K rates (with no increase in BBs). There is little reason to think that you can significantly increase the workload of these pitchers without impacting their performance, and lots of reasons to think you can't. This isn't Earl Weaver's game.....

Those are actually pretty decent pinch-hitting numbers (in particular, the walk rate).

Yeah, I was going to say that doesn't look half-bad, considering it's (a) people who aren't generally good enough to be in the starting lineup, and (b) numbers primarily from the 1970s, when you had regulars who hit like that. Do you have a baseline to compare it to?

A lot of guys who are fine hitters overall can't handle pinch-hitting. Something that happens in real life but not in a tabletop game.

Although under my system of basically profuse platooning there would be far more PH situations and thus guys wouldn't be asked to PH once in a blue moon as they are now.

In DMB I burn through my bench pretty much every game. I PH frequently. I have guys switch positions in-game (the neat thing is Pujols got in a game at 3B in 2012 so I use him there often even though he's rated as a Pr fielder). I have one pinch runner who is kind of useless with the bat though he can field all three OF positions (Tony Campana) and one pinch runner/pinch hitter who hits better than Campana though he only plays CF and I have Adam Jones there who doesn't have to come out of the lineup (Jarrod Dyson).

When I say pinch runner I mainly mean pinch stealer.

Of course, admittedly a lot of the platooning is forced on me because I've got a lot of 250-450 PA hitters and I'm trying to keep within PT limits. Guys like Brandon Moss and Andy Dirks and Chris Nelson and Wilfredo Rodriguez and Chris Carter. But I really think the strategy inflates a team's OPS.

Weaver platooned Etchebarren and Hendricks for years behind the plate. He also rotated four outfielders (Buford, Blair, Robinson, and Rettenmund) into three spots; it was not a strict platoon, but a variation of one.

Later on, I remember him platooning Rich Dauer and Billy Smith at second base. I also seem to recall Terry Crowley being platooned in right field one year.

Late in 1982 Weaver effectively platooned Glenn Gulliver and Lenn Sakata -- Dauer would switch between 2b & 3b.

But the World Cup qualifier next month is at the Broncos stadium.

It's at the Rapids stadium, which seats 18,000 and change. US soccer chose it over a larger stadium so they could leverage the season ticket base to help fill the stadium with US supporters and build a good atmosphere.

The Broncos stadium is getting Gold Cup games this summer, because CONCACAF sets the venues for that tournament and wants as many briefcases of cash as it can get.

That's absurd. These are professionals operating in a highly competitive environment, with enormous financial rewards for those who succeed. When we observe changes in behavior, it is extremely likely that teams are being rewarded for that change.


Want to buy a credit default swap?

I thought of that, too, but the point of selling credit default swaps was the profit to be made selling credit default swaps, and not serving any larger, economic purpose (despite the brochures). The latter was incidental to the profit motive. It didn't matter if they worked; it mattered if they sold.

One thing that has happened is that teams have become less likely to lose a late-inning lead, when you account for changes in offensive environment over the years - it's not that teams lose fewer leads overall, mind you, but that (a) run scoring has generally increased since the mid-70s and (b) late-inning games have tended to be a little but closer, so you would have expected to see the rate of lost leads going up, all else being equal.

This hits on the key point. This is testable. Have bullpens actually become more effective?

He says they have indeed become more successful since the advent of modern reliever usage. Mike, can you provide a citation here?

As Mike points out, this isn't definitive evidence. It could be that shortened benches on the offensive side are at least in part to blame. But if bullpen aren't better at protecting leads, that would be strong evidence that the shift in reliever usage has not been a productive one. So it's the thing we should be checking.

And empirically, I think finding effective platoon partners is harder than it first appears. The real question is how many LHH starters can be replaced by a better RHH option against LH starting pitchers? My local team, the Nats, have 3 LHH starters. Espinosa is pretty crappy against LHPs, but Lombardozzi vs. LHP has about the same OPS+ and gives you a worse glove. Moore is probably a better hitter vs. LHP than LaRoche, but a much worse glove at 1B. Moore might also outhit Harper, but do you want to take PA away from Harper or let him learn to hit LHPs as well as possible? In all 3 cases, I doubt the Nats gain from using a platoon.

Espinosa is a switch hitter with a .699 career OPS against RHP and a .814 career OPS as a right handed hitter against LHP. So a pltoon with him makes little sense, and if anything he'd be the short side of any potential platoon not the long side. Moore very likely will replace LaRoche against some lefty starters though I doubt it will be a fulltime platoon with LaRoche sitting against all lefties. Obviously platooning Harper would be absurd. However you forgot about the Nats new CF, Denard Span, though he actually has a reverse platoon split over his career.

Have bullpens actually become more effective?

He says they have indeed become more successful since the advent of modern reliever usage.

Here are starter / reliever splits from B-R every 10 years since 1952. Reliever tOPS+ was lower (i.e., better) in 1982 than 1952, 1962, and 1972, but higher in 1992 than 1982. The 2012 reliever tOPS+ is back to what it was in 1982.

However, I think another way to look at this question is, has pitching improved in the late innings (whether split by leverage or not)?

MLB, 1952

                                               
I         Split    PA   BA  OBP  SLG  OPS tOPS+
     as Starter 71380 .253 .325 .363 .688    99
    as Reliever 23453 .251 .333 .360 .694   101

MLB, 1962

                                               
I         Split    PA   BA  OBP  SLG  OPS tOPS+
     as Starter 88531 .259 .324 .398 .722   101
    as Reliever 36023 .253 .332 .381 .713    99

MLB, 1972

                                                
I         Split     PA   BA  OBP  SLG  OPS tOPS+
     as Starter 103026 .244 .305 .356 .662    99
    as Reliever  36942 .244 .325 .347 .672   102

MLB, 1982

                                              
Split             PA   BA  OBP  SLG  OPS tOPS+
as Starter    111781 .265 .325 .399 .724   103
as Reliever    49323 .252 .322 .365 .687    93

MLB, 1992

                                              
Split             PA   BA  OBP  SLG  OPS tOPS+
as Starter    110453 .258 .321 .384 .704   101
as Reliever    50092 .250 .326 .363 .689    98

MLB, 2002

                                              
Split             PA   BA  OBP  SLG  OPS tOPS+
as Starter    123795 .266 .331 .427 .758   102
as Reliever    62820 .252 .332 .397 .728    96

MLB, 2012

                                              
Split             PA   BA  OBP  SLG  OPS tOPS+
as Starter    121548 .261 .320 .419 .739   104
as Reliever    62631 .243 .317 .379 .696    93

Espinosa is a switch hitter with a .699 career OPS against RHP and a .814 career OPS as a right handed hitter against LHP. So a platoon with him makes little sense, and if anything he'd be the short side of any potential platoon not the long side.

Good correction -- I misread his platoon split. On LaRoche, he may sit against some LHP. Whether that's a plus or not will depend on how well Moore fields.

But the point remains: there is no LHH on the Nats where there is a platoon partner who would clearly be an improvement. This is also true of the other team I follow, the Cardinals. You might like to sit Jay against LHP, but StL doesn't have a RHH OF who would clearly be an upgrade in CF. Can people come up with a long list of LHH on other teams who could easily be platooned? If Ray is right -- that basically half the starting players should be platooned -- identifying 20 or 30 of these guys should be easy. I'm skeptical. And if you can't do it now, I don't see how expanding the position player share of the roster -- adding a bunch of replacement-level AAA players -- is going to change the equation.

In DMB I burn through my bench pretty much every game. I PH frequently.

Does DMB apply a PH penalty to the player when you do this? Players hit much worse as PH than as starters.

More generally, it does seem theoretically possible -- I know this will sound crazy -- that not everything that works in DMB will work in MLB.

Bobby Cox in Toronto is probably a better example of a guy who got a lot of mileage out of some fairly limited players by platooning them. Mullinorg being the best known, but he had a pretty productive platoon situation at catcher too. And did some platooning in the outfield and at DH.

And yet Bobby Cox was one of the leaders in advancing the current bullpen usage strategies (if I remember my Dag correctly). It strikes me that if there's one guy who wouldn't be uncomfortable bucking conventional wisdom if there was an advantage to be gained in the platoon system, Bobby'd be it. But whatever success he had with Garth, Dane, Hosken and Jesse, I don't recall him trying to round up a similar gang in Atl.

Mike, can you provide a citation here?

I did a study about four years ago, which I presented at a conference in San Francisco (think that was summer/fall of 2008). I used Dave Studenmund's probability charts, which show the probability of a team scoring "x" runs in an inning in that offensive, to establish a baseline probability for the likelihood of teams blowing a lead, and then compared that to the actual percentage of blown leads. For example, a team trailing by 1 run in a 4.5 run-per-game environment (which was Detroit's ballpark in 2012) is likely to score at least one run 27.6% of the team, so that would, on average, be the probability that I assigned to a blown lead when a team playing in Detroit led by a run at the start of an inning. If the Tigers held 4 of those leads to start the ninth inning, they'd blow one on average (and that happens to be exactly what happened in 2012, from what I can see). I looked at this from 1960, which was the first point at which essentially every team had an ace reliever, through 2007. Through about 1975, teams lost leads in the late innings at essentially the rate one would expect based on the probability distribution. From 1975 forward, the rate at which teams lost leads in the late inning declined to a point where by 2007 teams were losing only about 90% of the leads in the ninth inning, and 95% of the leads in the eighth inning, that would have been expected based on the probability distribution. I did a second study focusing only on 1-3 run leads and saw the same trend, so it wasn't a factor of having larger leads with which to work.

Before 1975 virtually all of the high-leverage innings after the sixth (and before extra innings) were pitched by either starters or a single ace reliever. By 1990 virtually all of the high-leverage innings after the sixth had been transferred to the bullpen, usually to multiple pitchers. That specific change is by far the largest change in pitcher usage over that time frame, and I don't think you can fairly evaluate pitcher usage without accounting for that specific change.

-- MWE

And yet Bobby Cox was one of the leaders in advancing the current bullpen usage strategies (if I remember my Dag correctly).

Absolutely. Bobby Cox was a pioneer in transferring *low* leverage innings away from starters and to the back end of his bullpen. Cox would more or less routinely use three pitchers for one inning to close out games in which his teams estblished large leads after six.

-- MWE

#84 Boog Powell was also in the mix. Frank Robinson would play first against many tough lefties (62 starts at first from 1969-71)

And Tom, yeah not bad in context but remember the pinch-hitter was often somebody like Jim Dwyer (with the platoon advantage). Not good enough to start for the Orioles (particularly in the outfield -- check the stats for his 4th OF. They're almost always very good) still generally meant a guy who could hit. And Weaver constructed his team so that he didn't have to ask Dwyer (as an example) to try and hit a tough lefty.

It's funny though. As I check specific players I see that the guys Weaver talked about were pretty successful for him in years where he was frequently using them in specific situation. He talked about the need to spot Curt Motton carefully and in 1969 Motton went .286/.394/.536 in 28 PH PAs (Dave May and Merv Rettenmund were the frequent pinch-hitters who brought down the PH line in 1969.) Dwyer had a good year as a PH in 1991.

There's more money and job security in being wrong like everyone else is wrong, as opposed to being alone in being right.

This. The problem with being a crusader/pioneer is that you have to right nearly all the time to keep your job. The excellent book "Scorecasting" talks about a high school football coach who never punts, no matter where the team is on the field; the team has won numerous championships. If an NFL or major college coach tried that, the minute it didn't work he'd be de-boned and filleted right there on the field.

This is why you have pitch counts: pitchers get hurt as a matter of course, but if you've got controls in place, you can say, "Hey, don't blame me."

One way to see the improvement in bullpen performance is to look at scoring in the late innings. Here is the scoring rate (per 9 innings) in the 7th, 8th, and 9th, compared to overall scoring, for the three-year periods 1970-72 and 2010-2012:

7th inning:
1970-72: +.13 (i.e. slightly higher than average)
2010-12: -.07
Delta: -.20

8th inning:
1970-72: +.07
2010-12: -.27
Delta: -.34

9th inning:
1970-72: -.12
2010-12: -.59
Delta: -.47

You can see that today's bullpens are more effective in every inning, and the gap grows steadily in the late innings to almost half a run in the 9th.

Did Weaver over-rely on small-sample pitcher-vs-batter matchups? Such as if the hitter was 10-30 against the pitcher Weaver would think such a sample was representative?

Through about 1975, teams lost leads in the late innings at essentially the rate one would expect based on the probability distribution. From 1975 forward, the rate at which teams lost leads in the late inning declined to a point where by 2007 teams were losing only about 90% of the leads in the ninth inning, and 95% of the leads in the eighth inning, that would have been expected based on the probability distribution.

That's seriously impressive. Roughly how many team games per season does that work out to?

(And thanks for the very helpful summary of your study.)

If bullpen use had evolved into something more inherently effective, I'd expect to see better performance as the game goes on, and better relative performance in high leverage situations, as time goes on. I'm not sure I'm seeing it; in other words, I'm not seeing any better management of the ninth inning as opposed to the 7th, or any better management of high leverage sitiations as opposed to medioum and low leverage situations.

Seventh/Eighth/Ninth Inning tOPS+ (all National League data)

1972: 107/107/99
1982: 100/98/93
1992: 105/96/100
2002: 103/94/81 (far and away the most effective)
2012: 98/105/101

High/Medium/Low leverage tOPS+ (all NL)

1972: 108/102/94
1982: 108/102/94
1992: 98/95/95
2002: 98/102/98
2012: 98/105/101

The leverage numbers have improved since the 70s and early 80s, but aren't really any better than they were at the onset of specialization around 1992 (as the chart in 13 indicates).

Of course, with the multiplicity of modern roles comes increased risk of filling the roles with the wrong guy -- another disadvantage of modern usage. That's probably part of what's being reflected here. Bullpens are probably "better" than they used to be, but it's far tougher for managers to get the right guys in the right roles. The two things offset each other.