My theory regarding today's newspaper columnists is that they are pansies. Guys like Grantland Rice and Ring Lardner turned a column every day - none of this "Sundays and Tuesdays" nonsense.

I'll grant him Spahn but Marichal withstood that level of usage for just 10 years. He had fewer career innings than Mussina.

What about Roy Halladay? Or is he usually too efficient to ever get pushed to a high pitch count.

My other theories regarding today’s pitchers is that they are pansies

Am currently trying to envision Franklin Pangborn as a Kansas City Royals pitcher, standing on the mound in his inimitably supercilious manner.

EDIT: On that ending note, I saw the Avengers last night. I don't know much about superheroes but it had a lot of Robert Downey Jr. in it so I enjoyed it.

Edit function is being unruly for me!

There are certainly pitchers today -- many of them -- who could throw far more pitches. But, herds and office politics being what they are, no one wants to buck the trend. They aren't looking to do the smartest thing, they're looking to preserve their positions. If a pitcher gets hurt, they want to be able to fall back on the tried and true "we were very careful and did things the same way everybody else does them" excuse.

But, herds and office politics being what they are, no one wants to buck the trend

But what exactly should they do? Keep pushing guys harder and harder until they actually find the point where their arms blow up? That seems like a dumb strategy. This isn't an engineering experiment where you can simulate and test countless iterations before you actually implement something. You get one chance with a guy, and everything you do is irreversible. Yes, some guys are going to be able to do what Verlander is doing, but it's not like they're even remotely easy to identify.

Keep pushing guys harder and harder until they actually find the point where their arms blow up?

Yeah, but guys' arms blow up at 2012 levels of work. Closers' arms blow up at 2012 levels of work.

The first step to making the process more efficient (*) is to realize that you aren't going to prevent pitchers' arms from blowing out.

You get one chance with a guy, and everything you do is irreversible.

Is this actually true with modern sports medicine? It strikes me that you get more than one chance with almost everybody -- Tommy John surgery is a speed bump today -- and that reality should be leading to increased use, not decreased. The pitchers should have been pampered in the 50s, 60s, and 70s (**) -- not today.

(*) Defined to include not underusing pitchers.

(**) Understanding that there's a financial component to this, and that arm injuries cost more today.

Yeah, but guys' arms blow up at 2012 levels of work. Closers' arms blow up at 2012 levels of work.

That doesn't mean that burning through them quicker couldn't make things worse. Just because a guy's arm is likely to blow out at some point doesn't mean it makes sense to say "eff it" and not even try to minimize DL time.

And yes, I'll say that TJ gives you more than one chance with a guy, but you've lost a year of value and development, and maybe another year while he works back to his peak effectiveness. That's still a lot of lost time/value.

I've always imagined Livan Hernandez wakes up, pitches batting practice to high school kids, throws a shutout in a beer league in the afternoon, pitches a complete game and gives up seven runs for his MLB team that night, and then throws another 75 pitches in his sleep.

In Jerry Reuss's day, pitchers never slept.

#9 - And that still doesn't address my question "what exactly should they do?". Yes, the system is far from perfect, but we need realistic alternatives, and not just to dump on it.

You get one chance with a guy, and everything you do is irreversible.

That's what SHE said!

#12 They used to use the system Calvin's father advocated for bridge load testing. Build the bridge. Drive progressively bigger loads over until the bridge falls down. Weigh the trucks and rebuild the bridge.

Bill James (who is now on the "not pushed hard enough these days" side) wrote that they established as normal the load that will destroy all but a handful of pitchers.

As I've said before, Jack Morris is kind of a bridge pitcher. The last guy to be given loads that are destructive to almost all pitchers. They've almost certainly gone too far on the side of being risk adverse, but it's surely worth noting that Morris is a HOF candidate precisely because so much of his competition broke under the load. And that the first generation of "babied" pitchers has an awful lot of very impressive careers.

My other theories regarding today’s pitchers is that they are pansies

Am currently trying to envision Franklin Pangborn as a Kansas City Royals pitcher, standing on the mound in his inimitably supercilious manner.

And after a few of those Joe E. Brown / Ray Milland / Ronald Reagan triple pump windups, we'd notice that every baserunner seemed to be stealing home standing up.

-------------------------------------

I'll grant him Spahn but Marichal withstood that level of usage for just 10 years. He had fewer career innings than Mussina.

And Spahn also had the benefit of an Uncle Sam "vacation" during WWII, which meant that he threw all of 125 innings between 1943 through 1946, and didn't begin his career in full until he was 26.

Checking Reuss...
his last season never cracked 77 pitches.
1989: 105 peak,just once over 100
1988: 100 peak, reached just once
1987: no pitch info, 33 max batters faced
1986: 28 most batters faced
1985: 36 most batters faced
1984: 39 most batters faced
1983: 41 peak (40 reached twice), 10 IP once, 9 1/3 another time
1982: 36 peak
1981: 39 peak
1980: 36 peak (10 IP once)
1979: 39 peak, 79 pitches in the one start recorded (31 BF)
1978: 35 peak
1977: 38 peak
1976: 38 peak, 94/83/78 pitches in 3 starts with pitch counts (27-31 batters), 10 IP once
1975: 44 peak, 117/107 pitches in 2 starts recorded (33/34 BF) 10 1/3 IP once
1974: 50 peak, 40+ 3 times - 13 IP in one start (the 50 BF one)
1973: 43 peak, 40+ twice
1972: 38 peak
1971: 40 peak, 99 pitches in one start (31 BF)
1970: 38 peak
1969: 27 peak (1 start)

So he did go to extras a few times with a 13 IP game his one time over 10. What a wuss compared to guys from the 60's and earlier who regularly went 12+! :)

His one year with a lot of pitch counts he averaged 3.19 pitches per batter. So for that 50 game that would work out to 160 pitches (roughly). Randy Johnson did officially reach 160 pitches once (1992, an 18 K game in 8 IP vs Nolan Ryan - http://www.baseball-reference.com/boxes/TEX/TEX199209270.shtml) but no one else has since to the best of my knowledge.

Most pitches per GS 2009-2012, minimum 300 IP

Most pitches per GS 2000-2011, minimum 25 GS

Prior, Zambrano, Wood... good times to be a Cub fan.

I'd also like to add we don't know how long Verlander can stand up to his load. Every pitcher is healthy until he isn't.

My global "solution" would look like this:


1) heavy instruction at all levels to umpires to call a rule book strike zone
2) minimizing breaking stuff, going as far as banning them, prior to age 18
3) teaching every pitcher a straight change
4) routine long toss, year round (the one story I truly believe about the "old days" is that guys threw a lot more growing up than they do today. They may have actually pitched less but thrown more as games of catch were long and happened often)
5) heavy instruction on throwing strikes
6) somehow (I have no plan) putting in limits on innings/pitches at younger ages. I think a lot of damage is done in summer leagues/high school/college and that this damage is mostly invisible to us. We think it's usage in pro ball that is hurting arms when there are almost always a ton of pitches on arms before getting to that point.

7) as said above: recognize that a lot of pitchers will still get hurt and not reacting to each injury as if it is a crime of pitching coach/manager.

2) minimizing breaking stuff, going as far as banning them, prior to age 18

There is some research at least, that suggests curveballs are not more damaging on the arm than fastballs and that most the danger is fatigue(both too many pitches in a single game and in a single season).

Is post #1 disappearing?

This isn't only about fear of injuries, it is about a change in strategy. There are very few pitchers who are actually more effective in the 8th than the lefty (with platoon advantage) throwing 96 and the closer in the 9th throwing 98+. The manager is trying to get team wins, not decisions for his starting pitcher. For some reason it took 90 years to figure out.

Sure it's great theater when Jack Morris and his wonderful moustache go back out for the 10th, but it may not be great strategy.

would like to see fangraphs have an X-axis of "pitch number" and Y-axis of "FB mph", "pitch value" etc. averaged over a pitchers' yearly starts/career.

There are very few pitchers who are actually more effective in the 8th than the lefty (with platoon advantage) throwing 96 and the closer in the 9th throwing 98+. The manager is trying to get team wins, not decisions for his starting pitcher. For some reason it took 90 years to figure out.

It's interesting to me that they were figuring this out right when offense was exploding. Should we deduce that without the advent of the modern bullpen, the silly ball numbers would have been even sillier?

This isn't only about fear of injuries, it is about a change in strategy. There are very few pitchers who are actually more effective in the 8th than the lefty (with platoon advantage) throwing 96 and the closer in the 9th throwing 98+. The manager is trying to get team wins, not decisions for his starting pitcher. For some reason it took 90 years to figure out.

Chicken or egg......do we have 8-man bullpens because relievers are generally more effective than starters after the 6th inning, or because many starters physically can't pitch beyond the 6th.

Chicken or egg......do we have 8-man bullpens because relievers are generally more effective than starters after the 6th inning, or because many starters physically can't pitch beyond the 6th.

Not really chicken or egg. If it was starter fatigue, why would we see relievers (esp. short relievers, esp. the top 3 guys) being substantially much more effective than starters? Wouldn't we simply see more multi-inning relievers? In the old CG days, relievers generally sucked and they were expected to go multiple innings and spot start -- i.e. they pitched like starters and bullpens were staffed mainly by bad starters.

Using this as a proxy (needs tweaking, some retrosheet would be good) for short reliever: 90%+ games in relief, 50+ IP, IP < 1.2*G:

no occurrences prior to 1961 (Bill Henry)
28 qualifying seasons from 1961 to 1979
33 qualifying seasons from 1980 to 1989
245 qualifying seasons from 1990 to 1999
655 qualifying seasons from 2000 to 2011

That's NL only by the way.

Meanwhile, IP/GS for selected years:

1965 6.4
1975 6.5
1985 6.3
1995 6.0
2005 6.0

That's not a very big change, not big enough to explain such a massive shift in bullpen strategy, especially given the small shift from the 90s to 00s (which might look a bit more dramatic if I'd chosen different years for all I know).

The main shifts in starter usage have not been IP/GS. The main shifts have been (a) going from 37-38 starts per full season to 32-33 and (b) dropping CGs partially balanced by virtually requiring that the starter go at least 5. NL only:

1965 1,246 (125 per team)
1996 1,829 (130 per team) (to avoid strike year)
2005 2,196 (137 per team)

Of Jerry Reuss's 547 career starts (average 6.5 IP/start), an even 100 of them didn't last 5 innings. Reuss did last long enough that there aren't particularly good modern comps for his career, in part because any starter with more than 3200 IP (Reuss had over 3600) had a better ERA+ than Reuss's 100. But not ridiculous comps are Moyer (over 4000 IP), Wells (3400 IP) and Rogers (3300 IP). Starts lasting less than 5:

Moyer 80 out of 632, 6.2 IP/start
Wells 62 out of 489, 6.5 IP/start
Rogers 66 out of 474, 6.3 IP/start

So for Reuss, about 2 out of every 11 starts lasted less than 5; for the modern guys, it's more like 1 out of 8. Wells has the same IP/start, just fewer starts, but only 54 CG vs. Reuss' 127.

It could be that the decisions to leave Reuss in for 9+ innings were correct on those days given he was dealing but I don't see any reason to think that David Wells couldn't have completed just as many games if he'd been given the chance. He wasn't given that chance as often because of the strategic shift to short relievers. The strategic shift to short relievers is what requires the starter to go at least 5.

By the way, it probably is fair to say that starters today are "trained" to last only 6-7 IP ... because, most of the time, that's all the team needs/wants given they are going to use more effective short relievers to close it out. But it's also fair to say that today's starters are "trained" to pitch at least 5. Anyway, yes, starts over 110 pitches are more rare but I'll bet you starts under 90 are more rare too. If it takes you 105 pitches to get through your 5 innings today, you're throwing 105 pitches.

It's interesting to me that they were figuring this out right when offense was exploding

Complete games were on the decline well before 1993, so I disagree that these came in together. The number of pitches it takes to get through TTO lineup may have accelerated the trend, I'll admit.

But I don't think it is chicken/egg at all. It is primarily driven by the decision on how to use bullpens -- see Sutter, Bruce. That's when the CG totals started downhill, when the closer role was really defined. Maybe it really isn't about effectiveness, maybe its all because of Jerome Holtzman and his stupid Save stat. But if it's more effective to have your starters go nine and have fewer bullpen arms, then someone should exploit that market inefficiency. I'm not holding my breath.

Maybe, because of bullpen strategy, pitchers are not trained to complete games. But the strategy was first.

It's evolutionary; bullpens have to be adapted, and adapt, as the strategy of consciously endeavoring to get to them becomes widespread.

Which means we likely have bullpen pitchers who would have been many-pitch starters in the pre-TTO/Moneyball era, and a much smaller spread between the quality of a "middle reliever" and a starter. This evolution is also likely a big factor in the depressed offense of the past few years -- the reward for a strategy of having your hitters stand at the plate spittin' and scratching themselves has diminished significantly.

Meanwhile, IP/GS for selected years:

1965 6.4
1975 6.5
1985 6.3
1995 6.0
2005 6.0

That's not a very big change, not big enough to explain such a massive shift in bullpen strategy, especially given the small shift from the 90s to 00s (which might look a bit more dramatic if I'd chosen different years for all I know).

The main shifts in starter usage have not been IP/GS. The main shifts have been (a) going from 37-38 starts per full season to 32-33 and (b) dropping CGs partially balanced by virtually requiring that the starter go at least 5.

I think using averages obscures the main shift in starter usage, because the main shift has been to reduce the variability in pitches thrown from appearance to appearance.

Before: starters were pulled from starters when they "didn't have it."
Now: I would guess that the data show that the pitch counts for most starts cluster fall between 95-105, with far less variance than before.

Before I pull lots of data out, here is a small sample of pitcher seasons from the Dodgers' Roth dataset and from B-R:

Year   Pitcher G GS   Pit IP     BF IP/GS BF/IP PIT/BF STDEV-PIT MIN P MAX P AVG Pit/GS
1956  Newcombe 38 36 3353 260.7 1024  7.2   3.9   3.3   26          19 137  93 
1962  Drysdale 43 41 4556 313.0 1282  7.6   4.1   3.6   23          60 163  111 
1963    Koufax 40 40 4454 310.9 1210  7.8   3.9   3.7   31          16 165  111 

2002   Johnson 35 35 3991 260.0 1035  7.4   4.0   3.9   13          89 149  114 
2005 Carpenter 33 33 3395 241.6  953  7.3   3.9   3.6   12          60 120  103 

The number of starts has declined, but the average pitch counts are not all that different. What is most different is the standard deviation in the pitch count distribution across each pitcher's starts, from 23-31 down to 12-13.

I am not saying that this is a representative set, but just food for thought. My hypothesis is that starters are being now deployed to eat a target number of innings/batters/pitches each game, rather than having the length of any given start being dictated by their specific performance in any particular game.

By the way, it probably is fair to say that starters today are "trained" to last only 6-7 IP ... because, most of the time, that's all the team needs/wants given they are going to use more effective short relievers to close it out. But it's also fair to say that today's starters are "trained" to pitch at least 5.

IP/GS, 1988 NL and 2010 NL vs NL starts

  
  IP/GS Stat    1988          2010
Games Started  1,879*        2,340**
          Avg    6.4          95.9
       Median    6.7          98.0

        StDev    1.9          15.9

   5.0-7.0 IP  1,021 (54%)   1,699 (73%)
     0-4.2 IP    287 (15%)     366 (16%)
      7.1 IP+    571 (30%)     275 (12%)

The share of starts less than 5 IP has not changed, and the growth in starts of 5-7 IP has come at the expense of starts of 7.1 IP or more.

Anyway, yes, starts over 110 pitches are more rare but I'll bet you starts under 90 are more rare too.

Yes and yes.

P/GS, 1988 NL and 2010 NL vs NL starts

  
   P/GS Stat    1988         2010
Games Started  1,879*       2,340**
          Avg   94.1         95.9
       Median   95.0         98.0

        StDev   22.0         15.9

          Min    6.0          5.0
          Max  167.0        132.0

 Count 90-110  719 (38%)   1,324 (57%)
   Count < 90  742 (39%)     649 (28%)
  Count > 110  418 (22%)     367 (16%)

*1988 NL starts with pitch count data recorded in B-R, out of 1,938 total. The starts with no pitch count data are essentially identical in average BF and IP per start.
**All 2010 NL vs NL starts have pitch counts recorded in B-R.

"Pitchers are sissies now" - Kid Nichols, 1948

Great work on this thread. Fascinating.

But I don't think it is chicken/egg at all. It is primarily driven by the decision on how to use bullpens -- see Sutter, Bruce.

I was going to say the same thing. These guys don't get to have their closer fetish and eat it too (wait a minute, that didn't come out right). The same guys who obsess over complete games are usually the first ones to pop off about closer mentality. There's a disconnect there. Either there's something special about these guys that makes them better bets at the end of a game than other pitchers (I would call it simply "Being A Better Pitcher") or there's not.

[32]

Correction:

IP/GS, 1988 NL and 2010 NL vs NL starts

   IP/GS Stat    1988          2010
Games Started  1,879*        2,340**
          Avg    6.4          95.9 5.9
       Median    6.7          98.0 6.0

        StDev    1.9          15.9 1.5

   5.0-7.0 IP  1,021 (54%)   1,699 (73%)
     0-4.2 IP    287 (15%)     366 (16%)
      7.1 IP+    571 (30%)     275 (12%)

That is a great rant by Nichols in the link in 33. My favorite part: "Pitching wasn't the only job pitchers had in the Nineties. The day after we had pitched a game it was our duty to stand at the gate, and afterwards to count the tickets. I remember counting 30,000 tickets one day at the Polo Grounds in New York." I have no reason to think he's lying about that, but has anyone heard stories like that before?