Will Carroll writes:

Will Carroll:I’m not sure any statistical tool has ever had more of an effect on the game than PAP. If you look at the usage stats, we just don’t have high PAP starts any more because it measured it so well and pointed it out to the general public.

Tango is write to ask what evidence is there that PAP actually measures pitcher abuse. As far as I can tell, there is none.

Carroll writes that PAP "pointed it out to the general public" the dangers of high pitch counts. I've never heard PAP discussed in a baseball telecast. I've never read about PAP in a mainstream publication. My non-geeky friends have heard of OPS. Some even know that Derek Jeter doesn't turn a lot of batted balls in the shortstop area into outs. NONE have heard of PAP. When pressed to name a mainstream publication that PAP has appeared in by a poster in another thread, the only example was a passing reference in 2004 by Alan Schwartz in the New York Times.

No "statistical tool has ever had more of an effect on the game"? As a poster suggested in another thread, how about the save?

I think managers try to limit their pitchers to 100 pitches for the same reason that PAP starts its abuse points after 100. 100 is nice, round easily remembered benchmark. Really, 100 is so arbitrary. Are 105 pitches really 125 times more abusive on a pitcher than 101 (as PAP would predict)?

To me, this is a classic case of the third factor problem with correlations. Since PAP was developed, pitch counts have become more important. But is there any evidence that PAP caused this? I think not. Causation does not equal causality. I think there is the same root cause to each phenomena. The baseball community-- managers and sabrmetricians alike-- were becoming increasingly aware of the damage that high pitch counts could do to pitcher's arms. As a result, pitch counts were lowered. Similarly, sabrmetric types also thought high pitch counts could do damage. As a result, they created a statistic that might quantify and measure it.

Has there ever been a statistic where the rhetoric matches up so poorly with the evidence as PAP? I remember being excited by the first article on it, but subsequent articles always seemed more like retrenchings than actual research. They'd look in different places, invent new metrics, but the effect it was supposed to measure always seemed right at the limits of detectability.

If you were an infinitely wise GM, how would you even use a PAP score? You'd get infinitely more valuable information from a scouting report about changed mechanics or a medical exam. A few percent increase in the predictability of wildly unpredictable injuries is just too small an advantage to chase.

gone!

Let's broaden this discussion: is there any evidence (as in a rigorous study, not anecdotal evidence) that high pitch counts really do lead to more injuries? I don't think I've ever seen any, except Woolners's PAP studies, which to me weren't very convincing at all.

Is there anything else?

I'm pretty sure if you test anything against the 70s pitchers, it comes up with nothing. They were freaks. There's never been another period like it in the history of baseball.

Ignoring dead ball guys

6 of 12 300 game winners were 70s pitchers (was 6 of 9 before Clemens, Maddux and Glavine) plus another 6 with 250+

the top 4 in IP, 10 out of 15 with 4500+

the top 4 in starts, 9 out of 16 with 600+

the CG list is so full of deadball guys I'm too lazy to work through it

before that group of pitchers, only one guy had ever K'd 3000 batters; 10 of those guys did it.

They were just freaks.

If you want historical, likely cases of "abuse" -- Feller, Roberts, Koufax spring immediately to mind and those are the guys who actually survived it pretty well. Even in the 60s-70s, you have suspect cases like Jim Maloney, Gary Nolan, Sam McDowell, Vida Blue (301 K in 312 IP at age 21! remained durable until 30 but K-rate was never the same again), Tanana (K-rate collapsed between ages 23-30). That was off the top of my head. But young pitchers always have been and always will be fragile so who knows if all those IP and K were the reason.

Imagine how good the Big Red Machine would have been if Maloney and Nolan had steyed healthy.

Chances are PAP per se never had any predictive power for injury. In Woolner's article, he shifts to "stress" which, if memory serves, was PAP3/pitches, for his look at injuries. This is partly because of course the guys who survive "abuse" are the ones who rack up high raw PAP scores -- and to survive "abuse", you have to be successful enough to pitch for a long time.

"Stress" had some predictive power in Woolner's article. I poked around a bit, nothing at all formal, and "high" stress for young pitchers (up to 25) did seem a pretty good predictor of substantial injury (around 50% if I recall, for stress over 40??). But I had no base injury rate to compare it to and besides "stress" came onto the scene just as pitch counts were dropping -- by 2002-2003 or so, nobody except Livan had "high" stress scores with any regularity. But you look at those lists in the 00s and yes, you see Livan and the Unit and Zambrano; you also see Wood, Prior, Schmidt, Burnett (just before his injury), Eaton (just before his injury), etc.

By the way, the big shift occurred between 2000-2001. You've got about 30 guys with stress scores over 40 in 2000, including at least 11 of 60+. In 2001, you get about 9-10 guys of 40+ and only 2 of 60+. By 2003, you've got only 11 over 30. In 2005, Livan was #1 at 125; Zambrano #2 at 45; Prior #3 at 36 and nobody else over 30 and only 9 over 20. In 2007, no pitcher topped 40 and only 6 topped 20.

As to why. Well, other than maybe the 70s, it's almost always been the trend that pitchers throw fewer innings meaning fewer pitches. Pitch counts were obviously more influential than PAP. But also sometime in the early 00's, MLB had an insurance company study injuries and a confidential report was put together (some of the findings made the press). I'd guess that report contained information on pitcher injury rates that have been a major push behind the further change.

is there any evidence (as in a rigorous study, not anecdotal evidence) that high pitch counts really do lead to more injuries?

The problem that you have with such a study is that we don't have the depth of detail on injuries that we have today for any prior period in baseball history, and therefore it's very difficult to draw a meaningful relationship between injuries and pitch counts.

-- MWE

But also sometime in the early 00's, MLB had an insurance company study injuries and a confidential report was put together (some of the findings made the press). I'd guess that report contained information on pitcher injury rates that have been a major push behind the further change.

I wonder how much of this has to do w/ the contract Kevin Brown signed w/ the Dodgers. Before that deal, I think most FA pitchers would sign for 2 years, maybe 3, as insurance and risk sort of naturally limited contract lengths. After that deal, it seemed that FA contracts were generally 4+ years; getting an FA ace for 2 years was suddenly a quaint thing of the past. Plus, insurance, esp. post-9/11, got prohibitively expensive for pitchers, so I wonder if the lower pitch count limits are as much a reaction to these new FA conditions – ie, now teams are erring on the more extreme side of caution because the market conditions have changed so drastically since 1999, and are still struggling to make adjustments to best ameliorate risk.

Looking at some data at B-R there is pitch data per start from 1988 on. I looked at Curt Schilling for fun (one of the youngest in the majors in '88). In his first year as a starter in the majors - 1992 age 25 - had games with pitch counts of 138, 134, 131, 128, 125, 120 with just 26 starts that season. In 1993 he hit 134, 132, 127, 120, 131, 123, 124, 123, 121, 126, 139 (11 out of 34 starts at 120+) plus in the playoffs 135, 131, and the big one - 147 for his game 5 shutout (99 for strikes - almost made the 100 mark in just those!) over his 4 games (99 in the other).

Wow. 1993 he was 26 and pushed that hard. PAP would suggest his arm should've fallen off at that point and he did have injury problems from the looks of it in '94/95/96 before 2 more healthy seasons, 2 more with minor injuries (24 & 29 starts), 2 more healthy, etc.

With the pitch totals for 1988 on it would be possible to check careers for guys based on PAP. Plus, this all occurs during the era of the modern 1 IP closer with setup men (LaRussa made it very popular with his A's teams in the late 80's) thus there shouldn't be too much adjusting needed. Anyone up for doing the digging?

FYI: For fun look at Randy Johnson as all his starts are listed. His second ML start was 130 pitches. 7 over 120 in '89 as a rookie, 14 over in '90 including two over 150 (!), 12 in '91, 18 in '92 including a 160 pitch effort over 8 IP, 21 in '93, 14 in '94, 18 in '95 (another 160), and 4 in '96 before he finally got hurt, 14 in '97, 18 in '98, 17 in '99. Oy vey.

If I was running a MLB team I'd want to know why the 70's produced so many 'freaks'. Why did so many guys achieve the 250+ win level? How did their arms survive as long as they did? Was the 4 man rotation a factor? Was it going from 4 man to 5 man during their career? Was it high counts young then lower as they aged? Was it the reverse? Was it the lack of closers at the start of their career then having them there just as their arms were weakening?

Lots of items out there. One figures there must be a reason for the survival rates. Also, if you are a MLB team you want 6-6 1/2 solid years then the guy goes to free agency so how do you maximize the value of those years? Once a free agent how do you know his arm isn't about to fall off? Who is a future Randy Johnson (abuse is secondary) and who is a future Kerry Wood (aka arm-fall-off boy)?

The problem that you have with such a study is that we don't have the depth of detail on injuries that we have today for any prior period in baseball history, and therefore it's very difficult to draw a meaningful relationship between injuries and pitch counts.

Absolutely right. Moreover, the practice of injury diagnosis and treatment has hardly remained constant over time -- e.g. MRIs, arthroscopic surgery, Tommy John surgery, etc.

The relationship between limiting pitch counts and preventing injuries is a reasonable conjecture, but that's all it is. (And, of course, an equally reasonable conjecture can be made that at some point, limiting pitch counts works to prevent pitchers from building resistance to injuries.) The PAP theory is a wonderful theory, but it was never more than that, and has never been tested against rigorous, comprehensive empirical data.

If I was running a MLB team I'd want to know why the 70's produced so many 'freaks'.

If you look at baseball from 1950 to now, the 70s had the lowest HR rates, look at a few BBREF team pages, many teams had several lineup spots where the pitcher could get away with throwing BP and there would be little or no risk of giving up a HR. Pitchers were still "expected" to complete games, if a pitcher was pitching well and his team had the lead- he'd stay in- pitcher especially "stoppers" were not automatically pulled after 7 or 8 in favor of the set up man and closer (which didn't exist yet).

The combination of things meant the following:

1: Pitchers wanted to pick up complete games
2: Pitchers would "pace" themselves, going all out only against power threats and when the situation demanded- ie: men on base.
3: Obviously some pitchers could not pace themselves, too hittable unless going fullbore at every batter- those guys obviously were not the elite and did not rack up 250+ ip.
4: Batters- more batters now go by the work the count approach, just seems like a lot less Al Olivers today- low walks low Ks- Garvey, Etc.

So what you had was a scattering of guys who were both good enough to get many batters out at less than max effort, and disciplined enough to manage their own in-game workload that way. Voila, 250+ 275+, 300+ innings.

I suppose someone who just doesn't give up HR (ie a ground ball pitcher) might be able to still pace themselves like that in today's game.

Batters- more batters now go by the work the count approach, just seems like a lot less Al Olivers today- low walks low Ks- Garvey, Etc.

This is commonly assumed and asserted, but there is scant evidence to support it. Walk rates in the 1970s were virtually identical to those of the 2000s.

MLB estimated pitches/9 innings, 1970-2007:

1970 146
1971 143
1972 142
1973 145
1974 144
1975 145
1976 142
1977 145
1978 143
1979 144
1980 143
1981 142
1982 144
1983 144
1984 144
1985 144
1986 146
1987 147
1988 143
1989 144
1990 145
1991 145
1992 144
1993 146
1994 149
1995 149
1996 149
1997 149
1998 148
1999 150
2000 151
2001 147
2002 147
2003 147
2004 148
2005 146
2006 148
2007 148

The relationship between limiting pitch counts and preventing injuries is a reasonable conjecture, but that's all it is. (And, of course, an equally reasonable conjecture can be made that at some point, limiting pitch counts works to prevent pitchers from building resistance to injuries.) The PAP theory is a wonderful theory, but it was never more than that, and has never been tested against rigorous, comprehensive empirical data.

That's my impression, too. So how do we explain the widespread belief that limiting pitch counts is somehow preserving arms? Is is just a case of everybody being ultra-conservative?

So how do we explain the widespread belief that limiting pitch counts is somehow preserving arms?

I don't know that we need to explain it, beyond the recognition that in every human endeavor, widespread belief in things without underlying factual proof is commonplace. Human beings, baseball management, commentators, and fans included, are basically smart and rational, but that doesn't mean that we're anything close to unsusceptible to folly.

If I was running a MLB team I'd want to know why the 70's produced so many 'freaks'. Why did so many guys achieve the 250+ win level? How did their arms survive as long as they did?

The simplest explanation might be that you get whatever you expect to get. If you demand eight complete games from Chuck Dobson, nine from Wayne Simpson, 10 from Ferguson Jenkins, and 11 from Gaylord Perry before the 1970 All-Star Break -- or nine from Bob Johnson after the break -- you exert a Darwinian selection pressure on all of them. Only the freaks survive, and they can cope with freakish demands almost indefinitely.

That isn't quite a tautology, I hope, either ... it's like, what if you took all the Olympic swimmers from some decade and made them swim the English Channel. Many would drown. Some would become great Channel swimmers. Then you'd wonder why there were so many Channel swimmers from that decade among its Olympians, since most Olympians before and since can't swim the Channel.

The tough part is getting injury data, but here's a simple question:

Have there been fewer pitcher injuries from 2001-2005 than there were from 1996-2000?

widespread belief in things without underlying factual proof is commonplace.

productive outs say hi

If I was running a MLB team I'd want to know why the 70's produced so many 'freaks'. Why did so many guys achieve the 250+ win level? How did their arms survive as long as they did?

You can also look at where the arms are coming from. Let's look at baseball. In the 60's, pitching was dominant. So, you had to have good pitchers to win, and kids saw good pitchers and wanted to be them. There could have been a migration where kids that were the best dropped hitting and fielding to go to pitching, and organizations could have made a concerted effort to push kids that excelled at both to pitching. (Just look at all the guys that are drafted out of high school or college that do both, and are the best on their team at both).

Another thing to think of is where the athletes that played baseball are coming from. In the 60's, and early 70's, baseball was by far the dominant sport. If a kid was a great athlete, there was probably external factors pushing them toward baseball. The NBA was miserable, the NFL wasn't the superpower it is today, hockey was pulling no Americans, and soccer wasn't popular like it is today among kids. Great athletes could have been picking baseball to play. Contrast that to today, where great athletes (especially those on the low end of the SE spectrum) are probably picking football and basketball because 1. it's a much much likelier road to a full ride scholarship and 2. the financial payoff if you do get lucky could be much much higher.

What my whole long rambling post is trying to say is, that today baseball may not be getting the quality athletes they were in the 70's and, the high quality athletes they are getting are all becoming hitters. There's really no way to test it, but it is something interesting to think about.

The tough part is getting injury data, but here's a simple question:

Have there been fewer pitcher injuries from 2001-2005 than there were from 1996-2000?

That's one damn "tough part," isn't it. If anyone has ever compiled a comprehensive database of "pitcher injuries" (however defined and -- most challengingly -- accurately measured), it's been kept secret from the world at large, including Will Carroll.

My own entirely subjective impression is that there wasn't any meaningful change in pitcher injuries between those two periods, nor has there been any that I've been able to detect since I began closely following baseball in the mid-1960s. But I have no hard evidence for this.

Estimated pitch counts for nowadays are too low, I'm fairly certain. When I've tested Tango's estimator against recent games for which we have actual counts, it's almost always low.

Estimated pitch counts for nowadays are too low, I'm fairly certain. When I've tested Tango's estimator against recent games for which we have actual counts, it's almost always low.

Well, I don't believe Tom has found that to be the general case.

Regarding Post #12, would it a) be possible and b) be worthwhile to determine whether or not pitch counts are similarly distributed across each of the nine innings over that period.

I guess my Darwinian explanation only works if Fergie Jenkins flourishes under any workload, while Wayne Simpson can flourish under a light workload but be destroyed by a heavy one. That's the assumption most organizations make, I would guess. If Wayne Simpson is destroyed by any workload, it really doesn't matter one way or the other: such guys were not meant to pitch in the majors anyway.

But if there are Hyper-Fergies in the world who thrive under heavy workloads but break down under lighter ones, then my entire point is invalid, or at least much more complicated than it was :)

a) be possible

It would be possible if you had hit, walk, and K data by inning -- I don't know, does retrosheet have that?

b) be worthwhile to determine whether or not pitch counts are similarly distributed across each of the nine innings over that period.

I'm not sure it would demonstrate anything, but it would be interesting.

Get right on it!

I'm not sure it would demonstrate anything, but it would be interesting.

Get right on it!

I never progressed beyond figuring secondary average leader boards throughout history during car trips as a kid. But if anybody does this and anything comes of that, I'll demand credit for the idea.

Batters- more batters now go by the work the count approach, just seems like a lot less Al Olivers today- low walks low Ks- Garvey, Etc.

This is commonly assumed and asserted, but there is scant evidence to support it.

From 1977-92(in the NL) there were 38.0 PAs per game, from 1993-2006 there were 38.6 PAs per game
From 77-92 8.4% of PAs resulted in a BB, from 93-06 8.7% of PAs resulted in a BB
From 77-92 14.5% of PAs resulted in a K, from 93-06 17.1 % of PAs resulted in a K.
More Ks, more walks, more pitchers per PA, even if only slight.

From 77-92 1.0% of PAs resulted in an intentional walk
From 93-06 0.8% of PAs resulted in an intentional walk (even with Bonds skewing the numbers)
An IBB is a low stress event from the pitcher's POV- those pitches shouldn't even count against a pitch count.

From 77-92 1.2 % of PAs resulted in a SH
From 93-06 1.1% of PAs resulted in a SH
A SH attempt tends to be a low stess event for the pitcher's arm- and those are down too.

Quite simply, every thing have ever seen tells me that pitch count estimators:
A: Over estimate how many pitches were thrown in the 60s and 70s;
B: Can't account for the variability that goes into pitches- a pitcher doesn't work as hard throwing an IBB, throwing to someone who is trying to hit a Sac bunt;
C: Can't account for the variability that goes into pitching to a Mario Mendoza as opposed to Neifi Perez- I'm serious, Mendoza had 4 HR, 46 XBH in 1456 AB - 3.2% of PAs, Neifi had more than 2X the XBH rate- Neifi sucks- but you have to pitch to him- he can hurt you if you threw BP- Mario and many others couldn't (doug Flynn and the traveling all glove not bat patrol).


I'm pretty much convinced that

A: A pitcher has to throw more pitches to complete a game now, 148-150, as opposed to 30 years ago - probably around 140
b: On AVERAGE the pitches thrown now are more stressful than 30 years ago- more threat of teh XBH or long ball, slightly less IBBs, slightly less SHs AND the pitcher knows he's not going 9, no incentive to holding back anything for the sake of getting a CG .

I don't think anyone has brought this up, but isn't there a difference between throwing a lot of pitches when you are 21 as compared to 31? A 31 year old who has been around a while and never had serious arm trouble would seem a better candidate than someone who is new to the big leagues.

I'm pretty much convinced that

A: A pitcher has to throw more pitches to complete a game now, 148-150, as opposed to 30 years ago - probably around 140

You might be right, of course, but I'm not anything close to as "convinced" as you are based on the evidence we have. And even if you are right, you're talking about a change of 8-10 pitches per game, or about 1 pitch per inning. This is a change that's utterly dwarfed by the change in innings per year that top starters are compiling.

b: On AVERAGE the pitches thrown now are more stressful than 30 years ago- more threat of teh XBH or long ball, slightly less IBBs, slightly less SHs AND the pitcher knows he's not going 9, no incentive to holding back anything for the sake of getting a CG .

Probably true, but again probably not nearly to the degree of magnitude that seems to be commonly assumed, and certainly not nearly to the degree of magnitude of the rate of change in top starter workloads.

JPWF: smaller parks mean that extra base runners score more frequently as well, leading to longer innings. As I mentioned on Tom's blog, it's my belief (although without data) that pitchers of today put more of themselves into each pitch during a start, whereas in the 1970s they regulated their effort more in an attempt to stay fresh throughout a longer expected start.

and certainly not nearly to the degree of magnitude of the rate of change in top starter workloads.

Hell, managing according to the save rule pretty much accounts for the rest of it.

Hell, managing according to the save rule pretty much accounts for the rest of it.

Um ... no. It doesn't.

Um ... no. It doesn't.

Strict 5 man rotations too,

The leaders in GS have less GS than before too.
No team has a 4 man rotation anymore
When 5 man rotations came into vogue the 5th man tended to get skipped a lot when off days came up
- that's less likely now, team instead of skipping the 5th guy still let him pitch and give the 1 & 2 & 3 an extra day in between starts rather than keeping them on a 5 DAY schedule.
Elite Sps are not expected to go 9 or even 8, 7 seems to be the "ideal" now.
3 run lead after 7? In comes the #1 set up man to register a "hold" and then comes the closer to register the "save"

Whereas before someone may have 36 GS and average 8 per start (288 ip)
a pitcher of roughly equal accomplishment today may have 32 GS at 7ip per start (224 ip)

those are the IP leaders

Jim Kaat was once asked why he pitched so quickly. His response was something along the lines of, after 2 hours my arm turns into a pumpkin.

And that's the other big change in the game, but one that's rarely mentioned. Games take a lot longer than they used to, and its not just about more offense. Its about longer commercial breaks between innings. Longer time between pitches. And other factors. My guess is that its a lot harder to pitch 125 pitches in three hours than in two.

My guess is that its a lot harder to pitch 125 pitches in three hours than in two.

There could be something to that.

At some point your arm just might be trying to go into "shutdown and repair" mode, and that might be dictated by the amount of elapsed time since you started throwing really hard, or by your adrenaline levels or something, which are going to wane after 2+ hours no matter what.

I'm a lazy SOB now, but I used to exercise :-) and breaks are good, but breaks that are too long are not.

Causation does not equal causality.

Yeah it does.

That's one damn "tough part," isn't it. If anyone has ever compiled a comprehensive database of "pitcher injuries" (however defined and -- most challengingly -- accurately measured), it's been kept secret from the world at large, including Will Carroll.

One more thing to work on. I think Sig Mejdal has one - Fantasyland mentioned that. He's now working for the Cardinals.

Sig Mejdal's data was assembled for STATS, so somebody should have access to it besides him. Then again, maybe he only gave a summary of the data. Shoot.

Even if you have good injury data, there's the problem that pitchers realize they can throw harder if their pitch counts will be lower. And, perhaps coincidentally, my hazy memories of the 70s say the average pitcher today throws much harder than he did in 1977.

22: http://www.hardballtimes.com/main/article/the-long-and-the-short-ofplate-appearances/

I found Tango's basic estimator to be quite good in re: 2006 data. (if you're interested in the follow-up articles, click here.)

my hazy memories of the 70s say the average pitcher today throws much harder than he did in 1977.

I have a lot of books from the 80's that make it seem like a 90 MPH fastball is a big deal. Now 95 is nothing special. Not sure if its pitchers throwing harder or just faster guns.

If guns are faster, then there's a weird conspiracy or a common modesty that requires nobody report speeds much faster than 100 mph. That is, the top speed seems the same as it was 30 years ago, and while it's possible Richard and Ryan are the two hardest throwers ever by x mph, it seems unlikely.

You know, we all have this mental image of a maverick willing to put his job on the line applying the latest research, and I think a lot of people see pitch counts as the "progressive" wave of the future (I get that impression strongly when I read Carroll on the subject). But with the actual research that's been done, and the softness of the correlation between pitch count and injury, wouldn't the current state of the research suggest that a manager should agressively try to squeeze, say, an extra inning per start out of the starter?

It's an odd fact of life that "progressive" doesn't always equate to progress.

I have a lot of books from the 80's that make it seem like a 90 MPH fastball is a big deal. Now 95 is nothing special. Not sure if its pitchers throwing harder or just faster guns.

In the Neyer/James book on pitchers there are many comments about how radar guns speeds are faster today.

The same book also has many stories about old time pitchers pitching famous games into extra innings, hurting their arms, and then never being the same again.

Whether or not a specific causal mechanism is identified it strains credulity to suggest that there is no relationship between pitching more often with less rest and injuries. In fact no serious person could suggest otherwise. The only question is the specifics. This is, of course, an important question and it is on this score that PAP, taken as an abstract principle, fails.

[Cross-posted at The Book blog:]

While the relationship between injury risk and pitch counts is interesting, it is also largely moot at this point. And that’s because we have figured out that, at least in today’s game, the less you use a pitcher, the better he will generally pitch. Very short relievers (4 outs and under) are more effective than long relievers. Relievers in general have a big advantage over starters. Starters are more effective the first two times through the order than after that. Starters pitch better on 5 days rest than 4, and better on 4 days rest than 3.

The most important changes in the game over the past 20 years have exploited this fundamental truth: universal use of the 5-man rotation, reduced IP for starters and the death of the complete game, expanded pitching rosters, and shorter appearances for the best relievers.

The starter/reliever gap, as well as the advantage of long rest, was documented in The Book. If you doubt that less use and shorter appearances matters for relievers as well as starters, consider the following data. Baseball Crank compiled the IP and ERA for teams’ 2 best relievers in (roughly) 5 year intervals. I’ve converted ERA to ERA+, to make it easier to compare over time:
Year IP ERA+
1920 217.5 97
1925 249.1 93
1930 233.1 97
1935 253.2 101
1940 196.5 101
1946 212.1 108
1950 219.3 97
1955 201.7 115
1960 200.2 113
1965 190.1 125
1970 178.2 121
1975 186.4 116
1980 200.9 127
1985 182.2 125
1990 163.7 138
1996 148.1 138
2000 146 135
2004 144.9 138
The trends couldn’t be more clear: teams use their best relievers less, and get a much better performance out of them as a result.

It may be that the original motive for reducing pitch counts was avoiding injury (though I’ve never seen clear evidence for that). But even if it now became clear that reducing starter pitch counts did NOTHING to reduce injuries, the game would hardly change at all. Starters would still get pulled after about 100 pitches, because a decent reliever is simply a better pitcher at that point than all but the very best starters (especially with the reliever often having platoon advantage). Any team that pushed their starters for more innings would quickly find that they won fewer games, and would stop.

The central advantage of relievers over starters, and short relievers over long relievers, is higher K rates (presumably because they can throw harder). I think it’s clear that a team that tried a 4-man rotation today, or even a rigid 4-day-off rotation, would see a decline in Ks. So too a team that asked its best relievers to routinely get 6 or more outs. And given today’s HR rates and high BABIP rates, pitchers who stop striking out hitters won’t win a lot of games.

Yes, a few things would change on the margins: some young studs might be allowed to pitch the 7th more frequently, and veteran aces might pitch the 8th more often. But starters would still average about 6.0 IP, complete games would still be rare, and teams would still carry 11-12 pitchers.

it strains credulity to suggest that there is no relationship between pitching more often with less rest and injuries. In fact no serious person could suggest otherwise

I completely agree.

Let's say that in the pool of professional pitchers there are 1% who can pitch at the major-league level like Gaylord Perry: insane workloads forever, with no harm. There are another 10% who can pitch MLB ball pretty durably in four-man rotations: the Lew Burdettes of the world. There are another 20% who can be major-league effective for a long while on a strict five-man rotation with limited pitch counts: the Byrds and Suppans. There are another 30% who can relieve at the major league level, with varying degrees of durability and effectiveness. The other 40% or so are filler and dross -- either effective at very short length or durable at very low effectiveness, guys who will flame out suddenly or never get very bright. If you try to work any guy from a lower tier at the rate that a guy in a higher tier can work comfortably, he will get hurt, bad.

If you are committed to a four-man rotation and complete games, your chances for success are dependent on finding a Perry and a few Burdettes.

If you can consider a five-man rotation, your chances of successful recruitment of pitchers go way up. If you have a lot of bullpen jobs, you can draw from the middle 30 and even dip into the bottom 40 of pitchers for LOOGYs and mopup men. If you happen to have Perry or Burdette, so they start 32 games a year for you instead of 38, they pitch seven innings instead of nine: it's not that great a disadvantage over the 38-start 20-CG workload, especially if you have decent relievers.

Much of the critique of c2007 pitching-staff strategies is premised on the idea that we are defining those tiers too conservatively; guys relegated to a lower tier could succeed at higher workloads. Indeed the Perrys and most of the Burdettes always can, of course.

But by defining the tiers so conservatively, you increase your chances of finding effective pitchers in the world's pool of decent arms, and of getting an effective mix of outings from them. That offsets the minor advantage of getting more innings out of a few pitchers -- especially since there's no way of telling which pitchers you should be getting more innings out of without running the risk of burning them out. Quality does not correlate with durability.

Again, this model fails if the relation between workload and injury is non-linear in some weird way for a lot of pitchers. If the problem now is that pitchers are resting too much and throwing too few pitches, then Gaelan is wrong, and I am wrong too, and we should start increasing our expectations of pitchers so as to keep them healthier ...

Edit: GuyM in #43 comes at this a much different way, setting injuries entirely aside, but I agree with his central point about effectiveness. Shorter workloads are more effective, and there are a lot more pitchers in the pool who can throw one effective inning than can throw nine -- in all eras.

If guns are faster, then there's a weird conspiracy or a common modesty that requires nobody report speeds much faster than 100 mph. That is, the top speed seems the same as it was 30 years ago, and while it's possible Richard and Ryan are the two hardest throwers ever by x mph, it seems unlikely.

I hate to say it, but in that game where Ryan threw 100.8 to set a record the gun used was either wildly inconsistent or else nobody really knew how to use it. Just look at the top velocity by inning:

http://baseballanalysts.com/archives/RyanScoreboard.jpg

You can believe Ryan threw only 87 in the first and gained 13 MPH through the night, or we can just accept that we have no idea how hard he was actually throwing at that time.

You can believe Ryan threw only 87 in the first and gained 13 MPH through the night

Nothing about Nolan Ryan would surprise me :)

It may be that the original motive for reducing pitch counts was avoiding injury (though I’ve never seen clear evidence for that).

Will Carroll begs to differ. ;-)

The central advantage of relievers over starters, and short relievers over long relievers, is higher K rates (presumably because they can throw harder).

It isn't just K rate. As my research here showed, it's HR rate as well, though BB rate is worse for relievers.

The issue, of course, is that while each individual pitcher performs better in shorter stints, it isn't an unambiguous positive to use every pitcher in shorter stints, for two reasons:

- Those extra roster spots for pitchers don't come out of thin air; they're a directly proportional reduction in position player roster spots, which isn't insignificant at all. The dramatic reduction in platooning has cost teams hundreds of platoon-advantage ABs per season.

- Those extra pitchers don't come out of thin air either; there are, by definition, at least 10%, and probably closer to 20% of pitchers in the majors today who are the quality of pitchers not good enough to have made the majors 20-30 years ago. Rate stat improvement is real and valuable, but the absolute magnitude of innings worked by inferior pitchers is real and significant as well. This is a partial explanation of the improvement in ERA+ by top pitchers; ERA+ is a zero-sum stat.

The shift in roster deployment and pitching staff usage that's occurred over the past 15-20 years clearly has positive elements, but it's by no means entirely positive.

Baseball Crank compiled the IP and ERA for teams’ 2 best relievers in (roughly) 5 year intervals. I’ve converted ERA to ERA+, to make it easier to compare over time:
Year IP ERA+
1920 217.5 97
1925 249.1 93
1930 233.1 97
1935 253.2 101
1940 196.5 101
1946 212.1 108
1950 219.3 97
1955 201.7 115
1960 200.2 113
1965 190.1 125
1970 178.2 121
1975 186.4 116
1980 200.9 127
1985 182.2 125
1990 163.7 138
1996 148.1 138
2000 146 135
2004 144.9 138
The trends couldn’t be more clear: teams use their best relievers less, and get a much better performance out of them as a result.

Do you have a link? This looks like ########. Teams in the '20s did not have 2 relievers each throwing 120 IP completely out of the bullpen.

Furthermore, the pitchers who were throwing completely out of the bullpen back then were nearly always the worst pitchers on the team. Relievers today are often failed starters, but also pitchers who have an inconsistent/non-existent 2nd/3rd pitch, who have mechanics that indicate they may not last long as starters, who end up in the pen due to roster circumstances and are too successful to move to the rotation, etc.

I do agree that the trend since the '70s or '80s shows that relievers are more effective in their shorter stints. I don't think comparing the data from the '20s/'30s/'40s shows anything, unless I'm misunderstanding something.

Chicken/48:
Here's the link: http://www.baseballcrank.com/archives2/2006/02/baseball_changi_1.php. It does sound like some guys counted as relievers in the 1920s/30s may have done some starting as well. So feel free to ignore that data if you're so inclined. My real interest is the trends since 1960, which I think are very clear: the top relievers are throwing 70-75 IP instead of 95-100 IP, and they are MUCH more effective relative to league average.

Steve/47:
It took you 17 minutes to respond. You're slowing down.....

the top relievers are throwing 70-75 IP instead of 95-100 IP, and they are MUCH more effective relative to league average.

No one disputes this. The issue which you seem uninterested in considering is that those 25 innings, once the cascade is completed, are being given to pitchers not good enough to make the staff in the earlier era, and the roster spots those extra pitchers consume used to be able to be used to deploy flexible platoon arrangements (offensive and/or defensive) that simply can no longer be done.

Nothing happens by itself. Every change has a consequence.

If the consequence is that Scott Seabol and Einar Diaz are now at AAA, then it's a little wonder the change was made.

No one disputes this. The issue which you seem uninterested in considering is that those 25 innings, once the cascade is completed, are being given to pitchers not good enough to make the staff in the earlier era, and the roster spots those extra pitchers consume used to be able to be used to deploy flexible platoon arrangements (offensive and/or defensive) that simply can no longer be done.

I have considered it, I just don't think it's nearly as important as you do. The innings are being given to relievers who only have to get 3 (or fewer) outs and often enjoy a platoon advantage. The fact that these pitchers wouldn't have made the roster in your good old days is irrelevant; today, used in this way, they are more effective pitchers than about 90% of starters when facing a lineup for the 3rd or 4th time. And the exact opposite factor reduces the value of your extra position player: pinch hitters are at a huge disadvantage, compared to hitters who play the field (also demonstrated in the Book). So to oversimplify just a little: use pitchers less, they do better; use hitters less, they do worse. The result: teams should use roster spots for 50 IP pitchers over 150 AB hitters, and they do.

I know you believe that current usage patterns continue due to blind obediance to orthodoxy. But teams have arrived at the current equilibrium through a steady evolution, after much trial-and-error learning. If it's not optimal, I bet it's very close....

So to oversimplify just a little: use pitchers less, they do better; use hitters less, they do worse.

That isn't oversimplifying a just a little. Using hitters in a manner that increases their proportion of platoon-advantaged PAs doesn't make them worse.

I know you believe that current usage patterns continue due to blind obediance to orthodoxy.

The inflexible attempt to have every roster, with its inherently unique blend of strengths and weaknesses, fit precisely the same usage pattern is pretty much blind obedience to orthodoxy, yes. The smartest team should look for competitive advantage by differentiating itself, accepting the risk associated with being different instead of avoiding it.

But teams have arrived at the current equilibrium through a steady evolution, after much trial-and-error learning. If it's not optimal, I bet it's very close....

The point is that the evolution should always be seen as continuing. At every point in history one could say (and people commonly did) that it's all been tried before, and finally what we have now is the ultimate pinnacle. Then 10 years later some new wrinkle had developed.

"It must be optimal because it's current" has never proven out. I'm more skeptical than you that we have arrived at the end of history.

GuyM, in your opinion, is this why relief stints of longer than an inning are disappearing, not just for the closers but the middle relievers? I seem to recall Whitey Herzog saying something to the effect that he didn't like to go through too many pitchers in a game because it increased the likelihood that he would use one who didn't have their stuff that nite.

I'm more skeptical than you that we have arrived at the end of history.

C'mon, Steve, you're better than that. Even if I'm right that roster construction is basically optimized, that's only for the current game. It's easy to think of changes that might shift the equilibrium point: new stadiums, influx of new talent from abroad, rules changes, a new pitch (the gyroball!), a less lively ball, a changed strikezone, or even (I suppose) medical advances that allow pitchers to throw harder over long appearances. All I'm saying is that in today's game, with current talent pool and technology, I very much doubt that a 10-man staff is optimal. So yes, roster construction may change, but it will be precipitated by an external shock not just a tactical adaptation to current conditions.

I should take back what I said about the strict 4-day rest rotation. I think that might work, IF managers used a quicker hook when the team has a large lead. It may be possible to get 35-36 starts out of the best starters, if you're very careful with their workload and trust your pen to take a 5-run lead after 5 IP. And that shifts some starts from your worst starters to your best. But I don't think you will see teams increase total season IP very much for their best starters.

I also think we may see teams experiment with 4 starters and a bullpen game for the 5th game. A mix of relievers is almost certainly better than almost all 5th starters, if you don't pay too big a price in relievers being unavailable the next day. However, that would lead a team to need more, not fewer, pitchers on the roster.

Bamberger: I think the general answer is yes, pitchers are just better over 1 IP than 2+ IP. The other factor is that once you've got 12 pitchers on the roster, you need to give them all work.

I think the problem is that analysts try to examine something like PAP with the same lens as they would, say... win shares, player A's value relative to player B, etc.

There likely is an ultimate answer to player A vs. B -- I have no doubt, as we increasingly get better data and better metrics on defense, we're getting ever closer to THE right answer in that regard.

Handling of pitchers, though, is a horse of another color. I know I've read studies indicating that back-to-back-to-back high PC outings yield poor results at the end of the string, but PAP still seems more focused on maintaining the health of a pitcher.

It's further complicated by the fact that, unlike OPS+, win shares, EqA, or whatever -- which tells us virtually the same thing about Barry Bonds as it does Carl Crawford -- I don't think any PAP advocate would say one size fits all perfectly. We know there are freaks of nature. I look at Carlos Zambrano and I marvel - and fear for his health less as he moves past his mid-20s.

Ultimately, we don't need any onfield numbers to know that throwing a baseball 85+ MPH overhand, to say nothing of the contortions necessary to produce good movement, is bad for the arm and shoulder. A baseball team needs a pitcher - and no one's going to advocate signing strictly submariners or sidearmers.

It's simply a matter of protecting your investment (generally speaking, that 18,19,20,21 y.o. you just gave a few million to and expect to anchor your rotation for 5-6-7 years).

It's anecdotal, not proof -- but a few weeks back in a Dusty thread, someone brought up Jim Leyland. Leyland had a reputation as an arm shreader going back to his days in Pittsburgh -- and while there are certainly PCs I would question under Leyland's reign in Detroit, it's a far cry from what we saw with Drabek, Smiley, Tomlin, Neagle, and company in Pittsburgh.

When I have a talent like a Mark Prior; when I've already guaranteed him 10 million; when he's my property for at least 6 years -- I err on the side of caution.

Whether that line of reasoning was a large part or small part of any work done by Rany, Keith, or Will is irrelevant to me - I want my team to protect its young pitchers.

I had thought that was a Weaver quote, but then again, I tend to automatically think any insightful managerial quote as coming from him. I'd have loved if he did a reprise of 1983 and did some playoff color commentary.

GuyM/48, the data you presented doesn't really tell the full story. 200 innings with an ERA+ of 127 in 1980 for the top two relievers, compared with 144 innings with an ERA+ of 138 in 2004 does not necessarily mean less value. The relevant question is the functioning of the entire staff.

If you look at pitching staffs as a whole, peformance stays fairly constant through 50-75 pitches (depending on the club) and then trails off after that. It is true that the defined role of the closer allows him to pitch better over the first 25 pitches than other pitchers do, but the effect of this defined role is to create less optimal conditions for the other members of the staff.

I have wondered if the workouts pitchers have today are a contributor to pitchers not being able to go as long into games. Lifting weights was discouraged until the 80's and weight lifting does not usually increase stamina but can add a couple MPH's to your fastball.

#60

This depends on what regimen you're doing. Weight lifting / resistance training can increase stamina just as it can increase strength. It just depends on what you're doing. Many endurance athlete lift weights / do some form of resistance training. They just don't do what a strength athlete does.

I had thought that was a Weaver quote, but then again, I tend to automatically think any insightful managerial quote as coming from him.

Bamberger was Weaver's pitching coach.

Perhaps, but Bamberger Daly wasn't!

I seem to recall Whitey Herzog saying something to the effect that he didn't like to go through too many pitchers in a game because it increased the likelihood that he would use one who didn't have their stuff that nite.

(note to self: check IP per relief appearance from 1957-onward).

Szym,

I thought you were responding to the end of post #56. I just read post #55 a few seconds ago.

This depends on what regimen you're doing. Weight lifting / resistance training can increase stamina just as it can increase strength. It just depends on what you're doing. Many endurance athlete lift weights / do some form of resistance training. They just don't do what a strength athlete does.

I know that there are different ways to train, I am not sure what the average MLB pitcher does but my guess is they do whatever they felt got them to the majors. Even people like Kerry Wood seem to go years thinking they just had bad luck and don't need to change a thing with how they prepare for a game or their mechanics. I am not even sure he has still changed anything.

If preserving leads is one way to measure bullpens, the modern bullpen is doing okay. Aft 6 = winning pct with a lead after 6, etc...




Peak periods seem to be around 1987-1990 and 2001-2003 (I am curious about what happened around 1957. Was that year a fluke or did something happen in 1958 that bumped up the trend towards more certain victories?) The differences is percentages from year to year seem small, the range is less than 4% and that includes '57. The correlation with runs per game is pretty weak. I thought that leads might be easier to hold in higher scoring years, but there doesn't seem to be much difference. The 5 year moving average trendline for the 9th is pretty flat, but the other two lines undulate a bit like a sinewave. I'll leave it to others to interpret that.