Top ERA+ from 1950-2005 60% of games as SP, qualify for ERA title ERA+ >= 180 (when did Pedro excel?)
1950-1960 - 3
1961-1970 - 6
1971-1980 - 5
1981-1992 - 5
1994-2005 - 19

Um that's a bunch.

I'd agree that that a difference like that should be a red flag, but I'm also not sure that it's a given that great pitching talent necessarily has to be evenly distributed over time, and not randomly concentrated within a short time frame.

And of those 19 180 ERA+ seasons from 1994-2005, Johnson (6) / Pedro (5) / Maddux (4) and Clemens (3) accounted for 18 of them. That could be due to an imperfect metric, but it could also just mean that those were 4 pitching ############# who could have dominated in any lively ball era under almost any set of circumstances. It doesn't necessarily mean that the overall environment they played in was easier to dominate.

EDIT: an FDA-inspected and certified coke to Ray

Also, baseball performance is asymptotic - you can only get so low in RA/ERA, BA etc.

It strikes me that this statement is a fact that must be accepted, and from this Foghorns conclusion necessarily follows.

You got a Pedro once every 5 (sometimes 6 games); you got a Gibson once every 4.

Not really. If you'll look at a rather long post I made here, you'll see me detailing the 1967 starts made by the Cardinals. It wasn't a 4 man rotation. It was mostly a 5-man rotation, shortened when possible, so sometimes a little closer to 5-day. I didn't do 1968 the same way, but remember that Gibson had only 34 starts in 1968. That's not a 4-man rotation. You'd see Gibson just about as often, but he would pitch a complete game. Note that 34 times 9 is 306 and Gibson had 304.2 IP.

Yes, to put up dominant K/BB/HR numbers. The Moneyball Era offenses weren't set up to beat Pedro Martinez and Randy Johnson; they were set up to wait out Pedro and RJ and then destroy Scott Kamieneicki and Steve Woodard. And they got much more Kamienicki and Woodard than 1968 offenses did. The very same approach that made Pedro look more dominant than he (likely) would have looked in 1968 was tailor-made to destroy Kamienicki and Woodward, guys with meh stuff and command, prone to making tateriffic mistakes. A BP fastball in 1968 is a single to left or maybe a shot up the gap by Felix Millan; in 2000 it's a 400 foot blast by John Jaha.

Fair enough on the raw HR/K rates, but why couldn't Gibson lead his own league in K/9 a few times -- or even once? Answer: because he wasn't good enough to do that.

HUH? the goal of a pitcher is not to put up great K/BB/HR rates. It's to prevent runs.

Two more posts of mine, brought over from the 1968 Most Meritorious Player thread:

-----

Splitting up Gibson's year into three phases.

1. Early in the year, April 10 through May 28. Gibson made 11 starts and had 97.2 IP. The team was 6-5 in those games; Gibson came out of a few of those games late and had a personal 4-4 record. Gibson's RA was 2.12 and the team allowed 2.18 R/G. But the Cardinals only scored 2.09 R/G in these 11 games. In fact, Gibson was pitching quite well but suffering from lack of offensive support.

2. The streak, June 2 through July 30. For Gibson: 11 starts, 11 CG, 99 IP, 3 runs. Took a run at Drysdale's scoreless inning streak record, but lost the streak after 47 innings to a 1st inning WP (against Drysdale, on July 1). But he wasn't done with the streak, not yet. So his RA, and the team's runs allowed per game, were 0.27. And NOW he was well-supported. In those 11 games, the Cardinals scored 4.18 R/G. So naturally, they won all 11.

3. The rest of the season, August 4 through September 27. For Gibson: 12 starts, 108 IP. He got the decision in all 12, and was 7-5. His RA was 1.92, with the team allowing 2.00 R/G. But he was back to being not all that well-supported, with 2.83 R/G scored for him. Included in here is his worst performance of the year, giving up 6 runs to the Pirates on August 24. But there are also four more shutouts, and a 0-1 loss.

Overall, the Cardinals were 24-10 in Gibson's 34 starts; his personal record was "only" 22-9.

Two things to note: one is that he did allow a little more than his share of unearned runs, so it doesn't look quite as good by RA as it does by ERA (but it's still way out there, even by RA). The second is that it doesn't look quite as good by DIPS theory as by RA - he did have BABIP working in his favor that year.

-----

Cardinal team record when Gibson allowed:

0 runs: 13-0
1 run: 8-3 (one of the three losses was 0-2, with a run allowed in relief)
2 runs: 0-0
3 runs: 2-4
4 or more runs: 1-3

Joe Hoerner was pretty effective per inning, somewhat surprisingly making DanG's list since he only had 49 IP. He wasn't exactly overworked. In 1967, the position of "bullpen ace" or most trusted reliever was shared between Hoerner (LHP) and Willis (RHP). In 1968, Willis had a bad year, although he still got some of the high-leverage outings. Of course, neither one saw much action in Gibson's starts.

Dal Maxvill batted .253. With a laughable lack of power. Which was well above his career raw BA, and which made him, in a 1968 offensive context, an unsung hero.

HUH? the goal of a pitcher is not to put up great K/BB/HR rates. It's to prevent runs.

No. That's the goal of the pitcher+defense. The goal of a pitcher is to put his defense in the best position possible to prevent runs. That happens if you limit the balls in play and the walks and home runs. Of course, the pitcher has to find a working balance. He doesn't want to (say) limit walks so much that he ends up grooving meatballs. But we're getting far afield here. The point is that BB/K/HR/ERA+ tell a lot more about talent level than simply using ERA+ does. And since we're talking about shifting eras and who was more dominant within their era, talent level matters.

That happens if you limit the balls in play and the walks and home runs.

It happens if you limit walks and home runs. Balls in play are on average very good events for a pitcher. So far this season, the AL is batting 289/367 on balls in play. A hypothetical all-BIP pitcher with average results on BIP would be the 8th best in batting against in the league this season and a solid Cy Young candidate.

EDIT: Obviously, if you pitched entirely to contact, you'd give up lots of well-hit balls in play (plus a bunch of homers). So the job of the pitcher is to limit walks and limit good contact.

The point is that BB/K/HR/ERA+ tell a lot more about talent level than simply using ERA+ does. And since we're talking about shifting eras and who was more dominant within their era, talent level matters.

Talent level? Where the hell are you going? What does that have to do with whether or not Gibson's 1968 season is as good or better than Pedro's 2000 season?

I get the claim, Ray, but I'm not sure you can make it cross-era. There was a secular change in the offenses' conscious propensity to walk, strikeout, and hit home runs between 1968 and 2000, which is why we see a third fewer balls in play in the park now than we did in 1980. The TTO approach has overtaken the game. We have to take that into account when we're projecting peripherals. It could simply be that John Briggs weakly grounded a Gibson pitch to short that John Jaha would have missed.

And that development makes sense given the incentives. With elite starting pitchers throwing so many fewer innings now, the effective cost to an offense of "punting" those innings in favor of making up the loss by pummeling Kamieneicki/Woodard is much lower now than in 1968. The atmosphere is simply better for an elite pitcher to put up dominant numbers now than it was pre-late 90s. At the very least, there is a very strong prima facie case.

No.

Yes. The goal of the pitcher is to strike everyone out. Everyone. But he doesn't have all that control. Given that periodically someone makes contact with the ball, he likes for a fielder to make a play on it. But ideally, he'd like to strike out every hitter on three pitches.

The true four-man rotation is as rare a thing as the 300-inning man (uncoincidentally). To stay in a four-man rotation all season long, somebody needs 40 starts, and that's been done 88 times in the lively-ball era. The Platonic ideal of it was the 1966 Dodgers, who gave just 8 starts to their fifth man – and it could only be done with great pitchers and few rainouts. Several teams in that era approached that ideal: the great A's and Orioles teams in the early '70s, the insane White Sox and Tigers teams from the early '70s (though they were verging on a 3-man, or "two guys pitch as often as they can" rotation), a team or two from the Cubs and Phillies also-rans of the late '60s, the Twins in the late '60s, the early-'70s Indians and Rangers. Other clubs tried to be like them, but injuries and personnel weakness made it impossible.

Before that narrow window, slow travel and rain and doubleheaders made 4-man rotations impossible. They'd be more feasible today, but common sense about the human arm seems to have prevailed :)

It happens if you limit walks and home runs. Balls in play are on average very good events for a pitcher.

Not compared to a strikeout, as you know.

EDIT: Obviously, if you pitched entirely to contact, you'd give up lots of well-hit balls in play (plus a bunch of homers). So the job of the pitcher is to limit walks and limit good contact.

I thought DIPS is still good law as far as showing that other than home runs there is really no such ability to limit "good contact." Once a ball is in play, all bets are off.

I thought DIPS is still good law as far as showing that other than home runs there is really no such ability to limit "good contact."

It is not. See Allen and Hsu et al, "Solving DIPS", among other things.

Also Tango's "DIPS Bands", MGL's mini-study on injured pitchers' BABIP. I could go on here. Also search for threads on this site about DIPS/BABIP with "GuyM" commenting.

In Gibson's era, a lot of good starters tried mainly to induce weak contact, whether or not they should have been doing so. Mel Stottlemyre, the still-living Dick Bosman :) , Claude Osteen – interesting how all those guys became pitching coaches, actually. It must have been interesting to face the Pirates and go from Bob Friend, who was a good-control weak-contact type, to Bob Veale, who was an extreme strikeout pitcher and walked a lot of batters. And by interesting I mean stressful.

And of those 19 180 ERA+ seasons from 1994-2005, Johnson (6) / Pedro (5) / Maddux (4) and Clemens (3) accounted for 18 of them. That could be due to an imperfect metric, but it could also just mean that those were 4 pitching ############# who could have dominated in any lively ball era under almost any set of circumstances. It doesn't necessarily mean that the overall environment they played in was easier to dominate.

It's not just those guys. Kevin Brown had a career ERA+ of 127 in a pretty long career. That's the same ERA+ as Tom Seaver. Tim Hudson (126), John Smoltz (125) and Mike Mussina (123) are in the same range. Andy Pettitte has a better ERA+ (117) than Fergie Jenkins. It's just a lot easier to move up past the "average" when the average comprises a bunch of bad pitchers pitching hundreds of innings a year.

Two notes:
1) Gibson's ERA from innings 1-6 was 0.88. If you go back in time and limit Gibson to a 6-7-inning pitcher, I think it's highly likely his ERA+ gets into the 300 range. 6.3 innings per start gets him to Pedro's 217.
2) By bbref, Gibson faced 265 batters in Close & Late situations, Pedro 110.

It strikes me that this statement is a fact that must be accepted, and from this Foghorns conclusion necessarily follows.

Not until "dominate" is defined.

I think what can be stated is that with smaller IP numbers and larger run environment the variance in ERA+ is greater. But, and this goes with the post about Gibson's FIP/lgERA in 1969 - ratios are not (necessarily) the best way to compare, especially across eras.

Logically (for Gaelan):

Who had a better season?
Better === Helped his team win more games in a 162-game season
win more games === probablistic assertion of P(win|Gibson starts) - P(win|replacement starter starts) X Nstarts(Gibson) as compared to the similar quantities for Pedro. For the record, Pedro had 29 starts, Gibson 34.

Warts and all WAR and WAA do a pretty good job of this estimation. Pedro 2000 is +1 WAA and 0.3 WAR above Gibson (the difference probably indirectly due to the extra innings/starts)

And I really like Tango's spreadsheet of pitcher / teammate BABIP. There is far, far more variation than you would ever see of a random number. Work by Tango and others (some guy who calls himself Pizza Cutter?) has found that it takes something like 1500 BIP to get a sample that measures pitcher talent to an r of .5, and that's where the "true gap" comes from.

Does a true gap in BABIP correlate with strikeouts? IOW, does inducing weak contact correlate with missing bats? You'd kind of think it would.

Two knuckleballers in the top 4 on Tango's spreadsheet is interesting. It makes sense that contact would be weaker against a good knuckleballer.

How much of a pitcher's value then comes from BABIP?

How much of a pitcher's value then comes from BABIP?

It depends on the pitcher. A gap of .005 in BABIP translates to a pretty significant difference in RA/9 (0.12, IIRC). For Tom Seaver, the plurality of his greatness came from BABIP prevention. For Sid Fernandez, it was what made him a professional ballplayer in the first place. For Andy Pettitte, it has little to do with his career. For Greg Maddux, it was maybe 10% of the story. For Glendon Rusch, it's what prevented him from being a star.

Does a true gap in BABIP correlate with strikeouts? IOW, does inducing weak contact correlate with missing bats?

It does. I wish I had a bibliography for all this stuff. It's disputed how strong the correlation is, but there's definitely one there. "Missing bats" is a skill that appears to show up in both numbers.

baudib, in #116 you say what Gibson would do in innings 1-6. In 2000, Pedro Martinez had an ERA of 1.42 in innings 1-6. What ERA+ would that be?

I remember, long before Sean made it easy, figuring approximate BABIPs for pitchers, and noting that Gibson's teammate Dick Hughes had one of the lowest career numbers ever. Small sample size: he only threw about 300 ML innings. But for guys with >300 innings, Hughes's .226 BABIP is still lowest among available records. (If you move the threshold to 200 innings, Neftali Feliz has overtaken him with a BABIP of .219.)

One of my tentative conclusions is that the late-'60s Cardinals could play some defense :)

One of my tentative conclusions is that the late-'60s Cardinals could play some defense :)

Yes, but who? Maxvill and Flood, sure, they were great defenders (even if Flood couldn't throw). But who else? In HoM discussions, defense kept being brought up as an argument against Brock. Shannon was an outfielder playing 3B. Maris was good, but he was old by then. Cepeda? Did he care about defense?

Dick Hughes was one lucky sonofagun. One glorious season as a 29 year old rookie, the best regular season pitcher on a WS champion, and just some spare bullpen change outside that one season. The possibility exists that he just wasn't ever really that good in the first place.

@114 - wow, that's almost 10 years old? Does BBREF use the rough 40/30/20/10 luck/pitcher/fielder/park ratio to create bWAR (for pitchers)? As I understand it, fWAR is pure dips.

Does BBREF use the rough 40/30/20/10 luck/pitcher/fielder/park ratio to create bWAR (for pitchers)?

They don't, which I think is the right call. Those numbers are a broad population mean, not a rule for every player-season. B-R does a separate park adjustment (which would include the park effect on BABIP), a separate but rough defensive adjustment (based on the quality of the team's defense on the whole), and assigns all the of the rest of the credit for run prevention to the pitcher. This definitely does attribute some stuff to the pitcher that's probably a function of luck, but that's what we do everywhere in baseball analysis (a dying quail and a frozen rope are both worth .47 runs).

B-R WAR for pitchers is imperfect and rough, particularly in its defensive adjustments, but I much prefer that to the fWAR methodology for pitchers, which is just entirely flawed.

In the referenced MGL study in MCoA's link in 114, MGL makes a point that I've been making for a long time:

This is further evidence that it is not that pitchers, in general, have almost no control over the ball when it leaves the bat and stays in the park. It is that MLB pitchers are specifically and partially chosen on the basis of not being hit hard by major league batters. If you get hit hard by major league batters, you don’t pitch in the major leagues.

and as I have also noted before, Bill James pointed out in Win Shares that when you evaluate players against a criterion used to help select the group in the first place, you are inevitably going to conclude that the criterion has little or no effect on player performance.

-- MWE

Bill James pointed out in Win Shares that when you evaluate players against a criterion used to help select the group in the first place, you are inevitably going to conclude that the criterion has little or no effect on player performance.

Bill James pointed out in a book released simultaneously with Win Shares that Voros' discovery was brilliant and that he could kick himself for not realizing it earlier.

This is further evidence that it is not that pitchers, in general, have almost no control over the ball when it leaves the bat and stays in the park. It is that MLB pitchers are specifically and partially chosen on the basis of not being hit hard by major league batters. If you get hit hard by major league batters, you don’t pitch in the major leagues.

and as I have also noted before, Bill James pointed out in Win Shares that when you evaluate players against a criterion used to help select the group in the first place, you are inevitably going to conclude that the criterion has little or no effect on player performance.

I fully agree with the point you're making, Mike. However, I am not seeing how it's relevant to whether major league pitchers have control over the ball when it leaves the bat and stays in the park. We're only talking about major league pitchers (specifically, pitchers who belong in the majors - not someone like Jamey Moyer or John Smoltz or Kevin Brown at the end). So it appears that among that group, MGL is in fact saying that pitchers have little/no control over the ball when it leaves the bat and stays in the park. Which MCOA seems to be saying the opposite of.

@127. Well clearly they are both wrong. And i think they are both bounds on the true value, much like ERA and FIP are bounds (100% pitcher credit vs. 0%)


Ideally we could correct hitting stats in the same way. Ideally. Not necessarily practically. With a few minutes more thought - its not symmetric. Batters face a wide, close-to-league-average distribution of fielders, but pitchers have a ~constant defense behind them.

No it wasn't. It may have been easier to put up high ERA+ numbers. But not to put up dominating K/BB/HR numbers.

I have to agree with Ray here. Pedro's league wasn't easier to dominate, as other people have pointed out with the listing of the quality of bats. Yes it's easier to put up higher era+, but it's not easier to dominate. Pedro was the dominant pitcher, but like Gibson with Koufax, Drysdale, Marichal and others, Pedro had his peers also having dominance such as Maddux, Clemens, Randy, Brown, etc.

Pedro put up elite reliever like rate numbers, while still pitching in a starter role. My knock against Pedro has nothing to do with doubting his performance, it's doubting whether it's right to compare it across eras, and assume his rate numbers translate, when he was pitching in an era where he knew going into the game, he only had to pitch 7 strong or 8 innings at the most. Where he consistently missed 2-4 starts a year, even in his prime years. My knock against Pedro isn't his dominance, it's his fragility.

Number of times Pedro had over 33 starts in his career? 0 Number of times Randy Johnson had 35 4 (along with 4 34, and 3 33) Maddux (9 times 35 or more starts, another 10 times with 33 or more) Schilling 4times with 35 starts... Pedro had 33 starts three times in his career. Every year you could pencil him in for a couple of missed starts. That is a major impact on his value as an ace. It also means he is taking off for aches and pains that other players are playing through.....Mind you, a some of that was called from the front office, but it does affect his value to the team.

1) Gibson's ERA from innings 1-6 was 0.88. If you go back in time and limit Gibson to a 6-7-inning pitcher, I think it's highly likely his ERA+ gets into the 300 range. 6.3 innings per start gets him to Pedro's 217.

That has been one of my points all along.

I thought DIPS is still good law as far as showing that other than home runs there is really no such ability to limit "good contact." Once a ball is in play, all bets are off.

Take a look at the FIP-ERA deltas for Greg Maddux during his 4 consecutive Cy Youngs. He's obviously an extreme example, but it's the clearest case I know of a pitcher consistently and decidedly beating his FIP and creating "bad contact."

I thought DIPS is still good law as far as showing that other than home runs there is really no such ability to limit "good contact."

This is a funny sentence. "Other than home runs."

It may have been easier to put up high ERA+ numbers. But not to put up dominating K/BB/HR numbers.

Ray, of course it was easier to put up dominant (relative) K/BB/HR numbers. I'm absolutely stunned that you would think otherwise.

First off, this

Also, baseball performance is asymptotic - you can only get so low in RA/ERA, BA etc.

should be acknowledged as true or likely true.

Now, let me get league "averages" out of the way. In 2000, the AL struck out 1.67 times per every walk. In 1968, the NL struck out 2.22 times per walk. In 1985 (to pick a league/year at random), the ratio was 1.58. Another random year, NL 1934, 1.28. AL, 1907: 1.55. So it's not like the 2000 AL K-BB numbers are in some otherworldly stratosphere.

Here's the list of single-season, all-time K-BB leaders.

Out of the top 35, seasons, you have one year by Fergie Jenkins (1971), two seasons by Cy Young (1904-05), and 32 seasons that came before 1890 or after 1994.

It's almost all outstanding pitchers, but it's not ALL Pedro-Maddux. No. 1 is Bret Saberhagen, No. 2 is Cliff Lee. No.5 is Curt Schilling -- three contemporaries who appear before Pedro. Other recent pitchers on the list: Ben Sheets, Carlos Silva, Roy Halladay and Randy Johnson.

Go into the 50s and you start getting people like Jon Lieber, Don Haren, David Wells, Brad Radke and Javier Vazquez. You don't get guys like Tom Seaver or Steve Carlton, or Carl Hubbell, or Lefty Grove or Robin Roberts or Warren Spahn or Bert Blyleven anywhere near the top 100.

BBref doesn't list "HR+" type data, but I'm pretty certain the list would be a bunch of guys from the sillyball era, and deadball guys who gave up 1 HR a year.

The top 100 K-BB single seasons, by decade:

Pre 1890:32 (a bunch of guys)

1890-1899: 0

1900-1909: 4 (Cy Young 3, Mathewson 1)

1910-1919: 1 (Walter Johnson)

1920-1929: 0

1930-1939: 0

1940-1949: 0

1950-1959: 0

1960-1969: 7 (Marichal 2, Koufax 2, Merritt, Perry, Kaat)

1970-1979: 1 (Jenkins)

1980-1989: 1 (Eckersley)

1990-1999: 15 (Maddux 4, Reynolds 2, Schilling, Smoltz, Pedro, LIMA!, Wells, Johnson, Swindell, Brown, Reed, Saberhagen)

2000-2009: 32 (Schilling 5, Wells 4, Halladay 3, Haren 3, Radke 3, Pedro 2, Santana 2, Sheets, Silva, Johnson, Lieber, Sabathia, Maddux, Beckett, Vazquez, Shields, Mussina)

2010: (3 Halladay 2, Lee)

Strikeouts are one of the most fascinating events in sports. It's so uneven. Hitters just keep striking out more and more and more, and nobody cares, because a strikeout is just any other out. And yet strikeouts are so hugely important to pitchers, and the more they strike out, the better they are.

The list in #137 is really striking. Rick Reed? Greg Swindell? Jon Lieber? They did something that Tom Seaver and Lefty Grove never did?

I feel like there's some grand insight that we haven't had yet that will resolve this.

If you get hit hard by major league batters, you don’t pitch in the major leagues.

Which, as we used to say, should be intuitively obvious to even the most casual observer.

The list in #137 is really striking

Listening to as much 1970s and '80s baseball on the radio as I did, I was always struck by how a plate appearance seemed destined to doom from both perspectives as soon as it began. If a pitcher got ahead, he wouldn't dare throw a strike, and if he got behind, he would have to give up and toss a changeup belt-high; meanwhile the batters seemed to start taking pitches as soon as they got ahead 2-0, and as soon as they were behind 0-1 they'd shorten up and try to punch it the other way. Maybe it was just the announcers I was listening to (Richie Ashburn in particular was a kind of Eeyore of the booth), but it always seemed like everybody just ought to give up and go home.

In the '90s and '00s, it seemed as if a batter would swing for the fences on any count, and a pitcher would throw a strike, particularly trying for a strike with a hard breaking pitch, on nearly any count. And it's probable that the results got distributed asymmetrically. Some pitchers started striking out a lot more than they walked, and some batters started striking out and walking a lot more than they used to. It's an impression, at least; it may not fit the evidence.

1) Gibson's ERA from innings 1-6 was 0.88. If you go back in time and limit Gibson to a 6-7-inning pitcher, I think it's highly likely his ERA+ gets into the 300 range. 6.3 innings per start gets him to Pedro's 217.

But what happens if Drysdale, McLain, Marijal, Perry, et all are all pitching 6.3 inn per start? He stiull gets 300 ERA+?

it's easier to put up higher era+, but it's not easier to dominate.

I find this contradictory, pending further clarification. Of course putting up a higher ERA+ is dominating (in one form).

Thanks baudib. Great stuff.

I also do not agree that a pitcher can't/won't throw 300 innings today. If he throws 24 shoutouts, he's going to, and pitchers get to throw when they are throwing shutouts (in most cases)

But what happens if Drysdale, McLain, Marijal, Perry, et all are all pitching 6.3 inn per start? He stiull gets 300 ERA+?

Exactly.

Ray, what are your thoughts on #137?

So it appears that among that group, MGL is in fact saying that pitchers have little/no control over the ball when it leaves the bat and stays in the park. Which MCOA seems to be saying the opposite of.

Well, I cited the MGL study because of what it shows - that injured pitchers have high BABIPs - not how MGL characterizes it. The study does not show "little/no control", and neither do any of the the other studies which have considered the topic over a sufficient sample.

What I think MGL is missing is that the difference between a .300 and a .320 BABIP may look like "not much control", in fact it translated to a difference of nearly 0.5 RA. It's a large difference, like the difference between a 5 and a 7 K/9. I can understand why you'd characterize it as "little control" given that it's only .020 points on a 1.000 scale, but in terms of actual baseball impact, it's huge.

Work by Tango and others (some guy who calls himself Pizza Cutter?) has found that it takes something like 1500 BIP to get a sample that measures pitcher talent to an r of .5, and that's where the "true gap" comes from.

Are you referring to this article by Pizza Cutter/former BPro author Russell Carleton (I think he works for an MLB team now)? I linked the republished fangraphs version since I don't think the Statspeak archives exist anymore, sadly. He doesn't explicitly say what you mentioned, but there have been a few other versions/similar studies, so maybe you mean one of those. Just curious as to where you got that exact number.

Interestingly, Carleton actually tried to address this issue of pitcher/batter responsibility and DIPS in a bunch of other articles all over over the place (a few are at his BPro search-thingy).

But what happens if Drysdale, McLain, Marijal, Perry, et all are all pitching 6.3 inn per start? He stiull gets 300 ERA+?

Wow. Yes, of COURSE he has a 300 ERA+! Obviously. Probably higher.

Because if Drysdale, McLain, Marichal, Perry et al. are all pitching 6.3 innings per start...then each team has the remaining innings eaten up by the Scott Kamienieckis and Steve Woodwards and the guys worse than those guys, the equivalent of whom weren't even in the majors in 1968. Every team uses 20-30 pitchers a year instead of 13 and the disparity between the best and worst pitchers grows larger, and bad pitchers are pitching as many innings (as a group) as the good pitchers (as a group). Just like 2000.

When you start limiting the innings pitched by the best pitchers in the league and start shifting a huge chunk of innings to worse pitchers, the overall effect is not to make the league pitching better, but much worse.

I think someone earlier asked about how BABIP was correlated with other stats. I did a regression with BABIP as the dependent variable and BB%, SO%, HR% as the independent variables. For the denominator I used batters faced with SH and IBB taken out. BBs included HBP but took IBB out. I used all pitchers from 2002-2011 with 1000+ IP. The r-squared was about .09, meaning that those variables explained about 9% of the variation across pitchers in BABIP. The standard error was about .012. So how well a pitcher does on the DIPS stats does not seem related to BABIP. Is this what other research has shown?

The sign on all the variables was negative. That sounds good for SO%: as you strike out more batters your BABIP goes down. But your BABIP also goes down if your HR% goes up (which seems odd since that should mean you are getting hit harder) and it also goes down as your BB% goes up. The t-values for HR% & SO% were -2.28 &-2.59. The SO effect seems small. If you improve by one standard deviation (SD) in SO%, your BABIP falls .003.

I also tried to determine how much a one SD change in BABIP, SO, BB, and HR per 9 IP would affect runs allowed. I used the run values from the FIP ERA formual (HR 1.3, BB .3 and SO .2) I assumed 25 balls in play per 9 IP and .5 runs for each non HR hit. Here is the runs per game saved by each 1 SD improvement

BABIP .15
BB .20
SO .28
HR .25