Bill James (almost) just discovered the old PAP/PAP3 system. How cute.

That's exactly what I thought. This isn't new, Bill.

New or not, would it work? I'm imagining the Randy Johnson/Curt Schilling Diamondbacks as virtually undefeatable.

Especially funny because Bill specifically ripped apart PAP in his and Neyer's book about pitchers.

Well, that's a whole bunch of assumptions that 1980s Bill James would rip to shreds. Here's one fun one:

I also think that pitchers would have liked it. A pitcher making 54 starts for a good team would have had a fair chance to "win" 30 games, and a hell of a chance to win 20. Frank Tanana might have gone 27-11 in 1983, as Bob Feller and Bucky Walters did in 1939 or 1940.

First, this ignores the actual culture of baseball at the time. Pitchers took pride in finishing what they started back then and would not have taken kindly to coming out after 90 pitches. Heck, pitchers today don't like to do that. Second, he's just plain wrong about the wins. A guy throwing 17 pitches per IP will last an average of 5.3 IP per start. So he's going to qualify for a win in about, 60 percent of his starts tops? So starting with 54, and dropping say 4 off the top (Bill's 'occasional 70-80 pitch outings), the pitcher in question would qualify for the win in about 30 games per year. Nobody's getting 30 wins (or 27) that way.

On the flip side, you're always eligible for a loss. So the overall result would more likely be good starters with 20-17 records, average starters putting up 15-20 records, and below average starters, yikes. Middle relievers would probably like racking up the easy wins, though.

More significant than that, you’re also reducing the stress per pitch, for an obvious reason. The most stressful pitches are those thrown when the pitcher is tired.

Sure, but 'tired' doesn't have a definition inscribed in granite. It could easily mean 'when a pitcher has not rested enough days after a start of 90 pitches'

Or it could just be untrue.

Maybe he sent Rany and Keith a nice fruit basket...

So, I love Bill and but I'm starting to wonder if he's found himself thisclose to something like 3000 opinions and now he's just dragging himself to it Biggio style.

That's one of those un-answerable questions, like whether Joe DiMaggio could still hit major league pitching with a 51-ounce bat or whatever it was. There is no way to test the theory.

You couldn't scientifically test the theory, but there's nothing except peer pressure to keep a team from going to a 10-man staff or a 4-starter rotation. Trying to determine the DiMaggio question is more like arguing who would win a superhero battle.

First, this ignores the actual culture of baseball at the time. Pitchers took pride in finishing what they started back then and would not have taken kindly to coming out after 90 pitches. Heck, pitchers today don't like to do that. Second, he's just plain wrong about the wins. A guy throwing 17 pitches per IP will last an average of 5.3 IP per start. So he's going to qualify for a win in about, 60 percent of his starts tops? So starting with 54, and dropping say 4 off the top (Bill's 'occasional 70-80 pitch outings), the pitcher in question would qualify for the win in about 30 games per year. Nobody's getting 30 wins (or 27) that way.


On the flip side, you're always eligible for a loss. So the overall result would more likely be good starters with 20-17 records, average starters putting up 15-20 records, and below average starters, yikes. Middle relievers would probably like racking up the easy wins, though.

Decisions:IP remains fairly constant for starting pitchers throughout the years. Thus this system would probably result in a similar number of decisions, but more of them would be losses (for the reason you mention).

Bill James (almost) just discovered the old PAP/PAP3 system. How cute.

Oh, it's much funnier than that. There's a longish essay in the Neyer-James pitcher book where he tries a takedown of PAP. He's now just (a) reinvented something he once knew existed and (b) reinvented something he thought was crap.

PAP never worked but "stress" (which was PAP divided by something) seemed to be much more in the ballpark ca 2000-2001. It was (badly and vaguely) laid out in an abstract. Of course a main problem with the way BPro used PAP and stress was an assumption that whoever was near the top must therefore be at risk. Whatever utility it may have had disappeared just as it was coming into prominence because teams cut back dramatically on high pitch-count starts (and high pitch-counts in general). No starters are "abused" today (or really the last decade) other than maybe a rare start here or there.

James also glossed over the fairly effective (seems to me) days of the 5-day rotation (or 4.5 man). The Braves kept it going for a good long while. I think that's still how I'd run one today -- you get 34-37 starts from your top 4 guys and limit your #5 guy to about 20 starts a year while those top guys pitch every 5th or 6th day just like now (only more 5-day starts and fewer 6-day ones).

@11: I can't remember exactly which dates at the moment, but I recall a whole lot of bizarre doings with the Mets staff over the last decade that made it seem like the FO was hell bent on getting a cratering Ollie Perez and an imploding John Maine into the rotation as often as possible. It seemed beyond the club to skip their worst starters and keep their best two or three (who were not struggling physically at that point) guys on 5-day rotations. When there were a lot of off days they were putting a healthy Santana out there once a week.

Well every team runs a 5-man rotation now. The D-Backs might have been the last team to come close -- they certainly did their best to squeeze a couple of extra starts out of Johnson and Schilling (and later Webb). The 2000 Braves also got 35 each from Glavine, Maddux and Millwood with the remaining starts spread among 3 pitchers. The 2001 A's were pretty similar too. But I don't think we've really seen it since the 93 Braves -- 142 starts from Maddux, Glavine, Avery and Smoltz.

In that year, Maddux and Glavine got 3 starts on 3 days rest, Avary and Smoltz just 1. Avery also got a start on 2 days rest but I assume he left the first one early. Anyway, for the team, 10 starts on less than 4 days; 112 on 4 days; so 40 on more than 4 days. The 2012 team had 72 on 4 days so 90 on 5+. Wow, I had no idea it was that spread out.

The flaw in the debate that a team can use a four man rotation (or three man) is the assumption that every pitcher can maintain their effectiveness and health on that rotation. Pitchers are human, and not every human recovers from physical exertion at the same rate. Some pitchers can recover and go every 4th day, and some can't. Say, for example, Justin Verlander could go every 4th day. What good is that if Max Scherzer and Doug Fister can't, and need that extra day? You would burn out your #2 and #3 guy in order to get more starts from your #1. The trick would be to identify and acquire four starters who can thrive in a four-man rotation.

I'm not saying a four man rotation couldn't work. I'm saying it's a lot more complex than just tracking pitch counts and innings.

Or a manager could do what Whitey Herzog did in 1982 and juggle his rotation so that individual pitchers get the rest they individually need. Joaquin Andujar pitched better on short rest and Bob Forsch needed the full 4 days. So Herzog moved LaPoint, Mura, & Stuper in-and-out so that Andujar could pitch on 3 days rest and Forsch got 4 or 5 days.

as one curmudgeon to another i would tell bill the key to quality curmudgeonry is consistency. you have to be irritated and disdainful and angry about the things established and then add as needed.

flipflopping really harms the credibility of the curmudgeon.

as with most things, it's the ability to sustain that is the key.

There's a longish essay in the Neyer-James pitcher book where he tries a takedown of PAP. He's now just (a) reinvented something he once knew existed and (b) reinvented something he thought was crap.

Not the first time, not even the most egregious flipflop of his career IMO. James wrote a cogent essay arguing against trying to come up with one big number that represents all of a player's value (which I've quoted many times and which with I wholeheartedly agree) - and then spent three years of his life developing one big number that represents all of a player's value (AKA Win Shares).

And I'm sure that when you point that out directly to James he'll defend taking both positions.

-- MWE

#17 I don't think the argument about the great numbers really applies though. It was written as an intro to the player ratings in the 1984 Abstract -- and the players are rated by what you might call a proto-WAR (two years rather than a single season, and the defensive system is based primarily on range factor and errors)

Right after pointing out the limitations of a "great number", he offers up a "so why do it?"

Maybe he sent Rany and Keith a nice fruit basket...

If this is anyone but Craig Wright, you're stealing my bit!

Actually James has written quite a bit more on pitching limits than simply taking apart PAP.

While the transition from four man to five man rotations was in its final stages he was a vocal proponent of the four man rotation. He also mentioned the bit about pitches when you're fatigued being likely more dangerous back in the early 80s.

And one of the first times he attracted national attention came when he wrote about Billy Martin consistently breaking every regular starter he had -- probably due to working them as hard as he did.

There's also the rather lengthy Teddy Higuera comment in the 1987 about how managers tended to test pitchers to destruction (not a truly prophetic piece in that he said that Higuera might be one of the handful who could handle the kind of workload that managers gave out -- his last healthy year)

Actually James has written quite a bit more on pitching limits than simply taking apart PAP.

While the transition from four man to five man rotations was in its final stages he was a vocal proponent of the four man rotation.

from the 1983 Abstract:

1) If I have a four-man pitching rotation and you are trying to persuade me to switch to a five-man rotation, what you are saying is that I should take eight starts away from my best starting pitcher, eight away from my second-best starting pitcher, eight away from my third-best starting pitcher, eight away from my fourth-best and give all 32 starts to my fifth-best starting pitcher.

2) Before I am going to do that, I want to see some real good evidence that I am going to get something back in exchange for it.

3) I have not seen any such evidence. Ergo

4) I wouldn't do it.

It wasn't quite the same as James' suggestion of going from a four-man rotation to a three with a 90-pitch limit, but the Rockies did perform a rotation experiment in 2012 by going from a five-man rotation to a four, with a 75-pitch limit (was it a target or hard limit, I forget?). Given all the factors involved with that staff in that location, I don't know if they were better, or worse, off for it.

To propose pitchers coming back on two days' rest after 80 to 90 pitches is just silly lab conjecture until you ask actual major-league pitchers what they think about their actual major-league arms' ability to withstand that.

Why does it bother folks that he seems to have changed his mind? That's what intelligent people do when the evidence says so.

I also think that pitchers would have liked it. A pitcher making 54 starts for a good team would have had a fair chance to “win” 30 games, and a hell of a chance to win 20.

Maybe if you eliminate the 5 inning rule for a starter to qualify for a win. They most certainly would not like a format where they can lose any game they give up the first runs, but can't win if they can't make 5 innings before the 90 pitch limit.

Though I guess you could have a setup guy pitch the first inning, then the starter look to cover 2-6 or 2-7.

James also glossed over the fairly effective (seems to me) days of the 5-day rotation (or 4.5 man). The Braves kept it going for a good long while. I think that's still how I'd run one today -- you get 34-37 starts from your top 4 guys and limit your #5 guy to about 20 starts a year while those top guys pitch every 5th or 6th day just like now (only more 5-day starts and fewer 6-day ones).

I agree. I remember the Royals doing this in the late 80s with good effect. You have four studs, and try to juggle the rotation so that the off days give them four days off. When you can't do that, use your long reliever as your fifth starter. These days, with 12 man staffs, it should be even easier since you probably have two long-men.

Why does it bother folks that he seems to have changed his mind? That's what intelligent people do when the evidence says so.

It's not clear that James is changing his mind based on the evidence.

-- MWE

The most stressful pitches are those thrown when the pitcher is tired.

Is there any actual evidence that any one pitch is any more stressful than any other pitch? (Aside from different pitch types like the screwball, etc.) It seems intuitively possible that you're more likely to get hurt when you're tired, but you could argue just as convincingly that you're more likely to get hurt when you're throwing hard at the beginning of the night. And it's not like relievers never blow out their arms -- if stress really went like the square of pitches thrown, you should basically never see middle relievers getting hurt.

Why does it bother folks that he seems to have changed his mind? That's what intelligent people do when the evidence says so.

I don't think anyone cares if he changes his mind -- that's great...

I think it's more the sort of "Hey, I was just thinking...." -- when the 'just thinking' is sort of like a draft treatment of Keith/Rany's writeup in a Prospectus about 7-8-9 years ago... Of course - others (I think someone mentioned Craig Wright for one) had also done some work on this, but my recollection of the prospectus piece was that it did reference some previous work, at least mentioning it in passing.

The other big thing is that James had specifically taken apart the piece done with virtually the same thinking -- pitch counts mattering more than innings and the whole idea that 'harm' grows exponentially past point x, where x = tired....

Again - this is just a mailbag answer, I guess - but it's common courtesy to acknowledge the thought's origins, especially when it can so easily be traced back via a critique he wrote...

I mean - after reading the blurb - I was immediately going to jump into the comments with my own little bon mot (I was going to go with "So Keith Woolner is ghostwriting the mailbag for this week?"), but 3 others already beat me to it, making the same connection.

It's just rude to go about it this way...

Why does it bother folks that he seems to have changed his mind? That's what intelligent people do when the evidence says so.

I'm not sure I'm "bothered", I find it funny. James is rather well-known for ignoring others' work -- i.e. not really being up on the current state of play -- and reinventing (or suggesting) the wheel on his own. That's an odd way to go about things but it's not uncommon in lots of fields.

But in this case, he's (on the fly) reinvented something he clearly was aware of 10 years ago or so. Of course he's got a lot going on but immediately half a dozen of us realized he was re-suggesting PAP and none of us wrote published essays on the topic. Moreover he's come to the opposite conclusion from before while giving no sign he realizes he's contradicting himself from just 10 years ago -- and as Mike said with no reference to evidence. (Of course the folks who created current pitching usage didn't have any evidence either.)

As to an earlier comment that essay wasn't the first time he'd written on the subject -- I didn't mean to suggest it was. But in this answer he's gone beyond something generic like "presumably pitches after fatigue are riskier", he's suggesting a more precise exponential treatment of pitches after a certain point -- i.e. a very specific approach.

2) Before I am going to do that, I want to see some real good evidence that I am going to get something back in exchange for it.

This is the dicey point. Putting aside the impossibility of testing this properly alongside that even a good observational analysis is at best quite difficult, if there's a benefit, it's that the pitcher's career lasts more seasons and possibly more starts but not necessarily more innings. There would seem to be financial benefit in that for the player (esp if it reduces early career injuries ... which I'm not sure it does) but there's not necessarily a lot of benefit in it for the team in the FA era.

Anyway, current usage is basically the combination of two theories. I think Brock Hanke could ride in here with some historical stuff but the first idea is "more recovery time = less risk" and that one's been operating throughout baseball history -- you see IP/start begin to down, you see an extra half or full rotation spot added and IP/start goes back up. But the more recent idea is also that "lower usage per start = less risk."* As far as I'm concerned, both of those are intuitively obvious and must be true in the abstract. The challenge is in optimising the mix -- i.e. obviously you can't have everybody pitch just 1 inning every 5 days. James and many others are essentially suggesting the bigger benefit is from limiting pitch count and therefore you can retain the overall value by cutting back on recovery time.

As to relievers -- they seem to provide evidence that recovery time is pretty darn important.

*While this is probably what drives strict pitch counts, it may not be what has really driven the change in usage. It certainly seems that the use of 3-4 relievers for the last 3 innings of the game is highly effective, at least in terms of run prevention. Obviously teams would love to find a way to keep 5th starters and/or the crappy bullpen pitchers off the mound as much as possible but I doubt there's a single team out there truly wanting the return of the complete game.

Or a manager could do what Whitey Herzog did in 1982 and juggle his rotation so that individual pitchers get the rest they individually need. Joaquin Andujar pitched better on short rest and Bob Forsch needed the full 4 days. So Herzog moved LaPoint, Mura, & Stuper in-and-out so that Andujar could pitch on 3 days rest and Forsch got 4 or 5 days.

I'm not saying a four man rotation couldn't work. I'm saying it's a lot more complex than just tracking pitch counts and innings.

I remember that. To duplicate its effect you'd have to keep one fairly elaborate chart, but it's not like you're doing parabolic calculus. I suppose the cost is whatever it takes out of your lesser pitchers to go irregularly. I also remember charting the Mets in the first decade of the 2000s just for fun, trying to get Santana on four days rest' Pelfrey too, back when he looked like a promising young pitcher, but giving John Maine and his increasingly tattered arm more than four days rest whenever possible, and skipping Opie when I could. There was one stretch where keeping everyone on four days rest meant Perez was out of the rotation for fourteen days. Instead, the Mets got him starts twice in that time, with predictable results.

Keeping everyone on, say, a 5 game rotation means they're getting irregular amounts of rest (sometimes four days rest, sometime five; occasionally six); no idea why teams think that sort of artificial 'predictability' is superior to giving their best two or three pitchers a very predictable four days rest throughout the season.


.

At 17 pitches per inning, 16.5 pitches per inning, that’s 5,000 pitches in a season, more or less.  Suppose that pitchers had been asked instead to pitch in a THREE-man rotation—but with strict limits of 90 pitches per start, and less than that for very young pitchers.   That’s 54 starts a season, 90 pitches per start MAXIMUM. . ..you’re actually REDUCING the number of pitches thrown in a season from about 5,000 to about 4,600

A couple of other people made the right note: Once a pitcher throws 50 pitches, he needs three days to recover. SO 90 pitches means I can't pitch in a 3-man, but could in a 4-man.

Even relievers that go 30+ pitches need an extra day (usually, I think). esseff and darren both note that the rest required is longer than the gap between games.

It's more like this:
0-20 pitches - 0 days rest
21-40 - 1 day
41-60 - 2 days
61-80 - 3 days
80-110 - 4 days
110+ - 5 days

There is some cumulativeness (like they used to have in LL - like, if you throw 20 back-to-back, you need 1 day off. If you keep it under 40 total, you can go again.)

Is there any actual evidence that any one pitch is any more stressful than any other pitch?

It's more that when you are tired, you deviate from proper fundamentals. So that's the "stressful" aspect. The more intense a situation, the more likely you are to burn more ATP, and get more tired earlier (regardless of pitches thrown).

Also, #31 is researchable and calculable. By someone better at Retrosheet than I.