1. The Shift

            Discussing what is called the Ted Williams shift, which is played against David Ortiz, Carlos Delgado, Ryan Howard and many other big-time left-handed power hitters, and was played against Cy Williams in the 1920s.   I have always known there was something wrong with this concept, but I have finally been able to put my finger on what it is.

            Suppose that you visualize the baseball field as a green surface, and the areas covered by the fielders as black voids—wedge-shaped black voids, for the infielders, and circular black voids, for the outfielders.   The size of the black voids depends on how hard and how high the ball is hit.  If the ball is hit hard and not too high, the black voids are small, and much of the green field is not covered.   If the ball is hit higher and/or more softly, the black voids expand, and the green area shrinks.

            Looking at the field prospectively, however (before the ball is hit) the field is a mix of covered areas, redundantly covered areas, and uncovered areas.  If a ball is hit into the gaps at about 225 feet and not too hard, either the center fielder or the wing fielder can get it.   That area is redundantly covered.   If the ball is hit hard into the gaps at 350 feet, neither the center fielder nor the wing fielder can get to it.   That area is uncovered. 

            What John Dewan always says, when I question the wisdom of the shift, is “How can it not work?   If you put the fielders where the ball is most often hit, how can that not increase your chances of making the play?”   But the problem with the theory is this:  that the traditional alignment of the fielders maximizes the areas that are covered and minimizes the areas that are redundantly covered, and this minimizes the areas that are uncovered.  As Craig Shipley says, “They put those fielders there for a reason.” 

            When you employ the shift, what happens is that you increase the areas that are redundantly covered, thus increasing the areas that are uncovered.   You thus increase the areas where the ball becomes a hit, or (especially) an extra base hit.   That’s why the shift doesn’t work:  it increases the areas of redundant coverage, thus inevitably increasing the areas of no coverage.      

            (Five minutes after I wrote this, the Red Sox lost a double play by playing the shift against Adam Dunn.    Runner on first, ball chopped to the shortstop in the second base position.   Nobody left to cover second base; double play becomes an out at first.   Exactly what I was saying.  The shift created redundant coverage on the second-base side of the base, but inadequate coverage on the other side, thus losing the play.) 

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

2.  Chasing Randy

            How many active pitchers have more wins than Randy Johnson had at the same age?

            Somewhere between 80 and 100.

            Most of these pitchers, of course, are young.  Felix Hernandez of the Mariners currently has 46 career wins, which is 46 more than Randy Johnson had at the same age, and Chad Billingsley has 44 wins, with is 44 more than Randy had at that age.   You might infer, then, that every pitcher in that age range who has a win, has more wins than Randy.    Tyler Clippard has more wins than Randy had at the same age, as does Joba, Johnny Cueto, Clay Buchholz, Michael Bowden and Matt Cain. 

            As we move in to the age range 24-26, there are still a very large number of pitchers who are ahead of Johnson’s pace, including not only the obvious (Greinke, Justin Verlander, Jered Weaver) but also a fair assortment of the unlikely.   Ambiorix Burgos still has more wins than Randy Johnson did at the same age; I know you don’t believe that, so look it up.   At the start of the season, not only did Scott Baker have more wins than Randy Johnson did at the same age, but so did his teammate, Jesse Crain.   So did Tom Gorzellany, Rich Harden, Oliver Perez, Big Game James Shields, and Dontrelle Willis.   At Dontrelle’s age Randy had 24 career wins; Dontrelle was well over twice that, nearly three times that, at the start of the season.

            As we get older, of course, the list gets shorter.  But among 27-year-olds with more career wins than Randy Johnson at the same age we still have Joe Blanton, Jeff Francis and (believe it or not) Daniel Cabrera, and also in that age group we have Dan Haren and Brett Myers, who are about 30 wins ahead of Johnson’s pace, and then we have Jake Peavy, C. C. Sabathia and Carlos Zambrano, who are more like 50 to 80 wins ahead of Johnson’s pace.   C. C. is so far ahead of Randy Johnson’s pace that he could very easily miss a year to have Tommy John surgery, have a health crisis, lose 70 pounds, and launch a second comeback, and he would still be two or three years ahead of Randy Johnson’s win pace.

            Josh Beckett is about 40 wins ahead of Randy Johnson’s pace.  Chien-Ming Wang, entering the season, was ahead of Johnson.   At age 29 we have Mark Buehrle, about 50 wins ahead of Johnson, plus Johan Santana, Jon Garland, John Lackey, Cliff Lee, Brandon Webb and Ben Sheets, all still comfortably ahead of Randy Johnson’s pace.   Somewhat more surprisingly, we also have Kyle Lohse, Jason Marquis, Gil Meche and Joel Pineiro, all (at least entering this season) ahead of Johnson’s pace.

            Don’t believe me, do you?   I wouldn’t, either.   Kyle Lohse turned 30 last October.   He had 78 wins on his birthday, 78-80.    Randy Johnson turned 30 late in the season in 1993.   At the end of that season—giving him the advantage of the extra month—he had 68 career wins.   

            As we move into the thirties, of course, the number of pitchers with more wins than Randy decreases rapidly, but it takes a long time for it to disappear.   You remember the great Oakland A’s troika from years ago, Mulder, Hudson and Zito?   All three of them still have more wins than Randy had at the same age.    If you look at the Texas Rangers, Vicente Padilla has more wins than Randy had at the same age, and so does Kevin Millwood.   Millwood is 34, couldn’t really be said to have one foot in the Hall of Fame, but he’s about 20 wins ahead of Randy at the same age.   Jeff Suppan is ahead of Randy’s pace, and Javier Vazquez, and. . .here’s another one you’re going to have to look up for yourself—Livan Hernandez.  

            As, of course, do the great pitchers in that age range, Roy Halladay and Roy Oswalt and such like.   Not sure there is any such like. 

            Who is the oldest active pitcher still ahead of Randy’s pace?   Do you consider Tom Glavine active?   Do you consider Pedro active?   If you don’t consider either of those to be an active pitcher, than the most advanced flinger still ahead of Randy’s pace would appear to be Andy Pettitte, currently with 222 wins. . ..about 35 ahead of Randy’s course. 

            Randy, of course, has a closing kick that would make Mine that Bird look like a civil war statue.   I am not suggesting that Livan Hernandez and Kevin Millwood are 300-win candidates.   They’re not.  

            My point is this:   that sometimes people would give you the impression that, to win 300 games in modern baseball, a pitcher would have to have an almost perfect career—come up early, get good early, stay healthy, stay good for a long time.   That’s just not true.    Is Kyle Lohse having a perfect career, or Jason Marquis, or Jeff Suppan?   And yet, ten years into their careers and more, they’re still ahead of Randy Johnson’s 300-win pace.    It takes a great pitcher to win 300 games, but it doesn’t take a great pitcher having a near-perfect career, like Willie Mays or Hank Aaron or Albert Pujols.  If a great pitcher had a perfect career in modern baseball, he’d win 400 games easy.  

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

3.  Only the Lonely

            Signature Significance is the term I use for small-sample events so extraordinary that it is essentially impossible for them to happen to an average team or an average player.    The concept was developed on August 21, 1984, when a rookie pitcher, Roger Clemens, struck out 15 batters and didn’t walk anybody.   It struck me at the time that this was such an extraordinary performance that it constituted virtual proof that Clemens was an outstanding pitcher, since an ordinary pitcher would literally never have a game like that.   An ordinary pitcher might get lucky and pitch a no-hitter—but an ordinary pitcher is never going to have a game with 15 strikeouts and no walks. 

            At the moment, the Kansas City Royals have lost five straight games, by the scores of 12-5, 12-5, 10-5, 7-1 and 12-6.  (They won Tuesday night, after I wrote this.)  Five straight blowout losses, combined score of 53-22—all of them home games, incidentally.  I had been really hoping that the Royals were a better team this year, and I probably told some of you they were, but five straight blowouts are hard to argue with.   Does a good team EVER lose five straight games by five or more runs?   Has that ever happened to a .600 team, ever in the history of baseball?   Realistically, would it ever happen even to a .500 team?  Or is it the kind of thing that only happens to bad teams?

            There were four teams in baseball last year that had won-lost records of 95-67 or better—the Cubs, Rays, Red Sox and Angels.   The Cubs had 14 blowout losses (5 runs or more), once had two straight blowout losses, and never had a string longer than 2.  The Rays had 18 blowout losses, and they also had one string of 2 straight, nothing more than that.  The Red Sox had 17 blowout losses, but never had two in a row.  The Angels had 20 blowout losses, and they also had one string of two in a row, no more than that.

            We can make an initial estimate, then, that a .600 team could be expected to suffer five straight blowout losses about once every 350 years, or perhaps a little less.   The logic for that is as follows.   Let us say that .600 teams are blown out about 18 times a year.   That means one game in 9, since 162/18 is 9.   The chances of a .600 team being blown out five consecutive times, then, would be about 1 in 9 to the 5^th power.   Nine to the 5^th power is 59,049.   59,049 divided by 162 is 364.5.   We could estimate, conservatively, that a .600 team should be blown out five straight times only once in 350 years.  

            There are many problems with our model. . ..my average is based on a study of only four teams.  The average of those four teams isn’t 18 blowout losses, it’s 17, which would make a significant difference.   Many of these teams didn’t play 162 games.   The number of blowout losses would vary with the number of runs scored.  

            It could be argued that the appropriate math isn’t

            18      18        18       18      18  

           ----  *  ---  *   ----  * ----  *  ----   =   1 in 59,049

            162    162     162    162      162

            It is:

17      16        15       14      13  

           ----  *  ---  *   ----  * ----  *  ----   =   1 in 141,183

            162    161     160    159      158

            Probably some of you understand why that might be the appropriate math and some of you don’t.   If you don’t, don’t worry about it; it won’t be on the Final. 

Games of this type could “cluster” for any number of reasons, such as injuries, home/road schedules, and playing series strong opposition; that also would screw up our estimate.   My experience is that things like that could cause your math to misfire, but usually don’t.    

            Anyway, since 1900 there have been 2,220 major league teams (team/seasons), of which 248 have played .600 ball or better.   Conclusions:

            1)  For a .600 team to be blown out in five straight games may have happened once since 1900, but probably has not happened twice.

            2)  For a near-.600 team to be blown out five straight times, however, has probably happened, in that there are more near-.600 teams than actual .600 teams, and the likelihood of blowouts for near-.600 is probably meaningfully larger than for actual .600 teams.

            Let’s look now at .500 teams.   The only two near-.500 teams in the majors last year were the Diamondbacks (82-80) and the Indians (81-81).  We have an advantage here, however, in that we can look at both blowout wins and blowout losses, since a .500 team could probably expect to have as many of one as of the other. 

            The Arizona Diamondbacks had 26 blowout wins, but never more than two in a row.   They had 25 blowout losses, and once had 3 in a row (June 14 to June 17—the first two losses coming, oddly enough, to the Royals.)   The Indians had 29 blowout wins, only 20 blowout losses, but they never had a streak longer than 2 in a row in either direction. 

            To minimize the appearance of laziness, let’s thrown in the last previous team to finish exactly .500.  That would be the 2005 Washington Nationals.  The Nationals played in a pitcher’s park, which caused them to have fewer blowout games.   They had only 13 blowout wins, only 18 blowout losses, and they, also, never had a streak longer than two in a row either way.

            These three teams had a total of 68 blowout wins, 63 blowout losses.   Let’s say there are an average of 23 blowout losses a year for a .500 team, estimating on the high side to get a conservative number later on.  That’s one game in 7.   Seven to the fifth power is 16,807.   We can thus estimate that a .500 team should be blown out five days in a row about once every 100 years.  

            So that would appear to be relatively rare.   There are just over 400 teams since 1900 which have finished the season within 3 games of .500 (that is, between 78-84 and 84-78, or comparable spread on a shorter schedule.)   We would expect, then, that for a .500 team to be blown out five days in a row has probably only happened about four times in baseball history.   I am, of course, relying on you guys to find the actual facts; I’m just trying to create informed speculation.   I don’t have the programming skills or the data base to do a global search.

            Now we need to look at the .400 teams, or, as we call them in the Midwest, the Royal-like teams.  In 2005 the only team in the majors to play less than .400 ball was the Royals, who finished 56-106.    The 2005 Royals had 13 blowout wins, 37 blowout losses. . .only the second figure is relevant to our current study.   The Royals never had consecutive blowout wins, but they did have a streak of four straight blowout losses (16-1, 11-0, 13-7 and 6-1, August 6 to 10.)

            In 2006 the only sub-.400 teams were the D Rays (61-101) and the Royals (62-100).    The D Rays had 12 blowout wins, 26 blowout losses, but never had a streak longer than 2 straight either way.

            The 2006 Royals had 14 blowout wins, 36 losses, and they had two streaks of 3 straight blowout losses.  

            That gives us a sample of three teams, but we have the data from the .600 teams in 2007; we can use that, since the number of blowout wins by a .600 team should be, I would think, about the same as the number of blowout losses by a .400 team.   This gives us:

            The 2008 Angels had the same number of blowout wins (20) as blowout losses—odd for a 100-win club, but it’s not a typo.   Anyway, these seven teams would suggest that the average number of blowout losses for a sub-.400 team should be about 30, or roughly (let’s say) one game in five.  Five to the fifth power is 3125, so a .400 team should be blown out five straight times about once in 3,125 games, or once every 20 years, more or less.  There are 249 team in baseball history (through 2008) which finished at or below .400.   Thus, for a bad team to be blown out five straight times has probably happened about a dozen times in baseball history.

            So. . .I would say that this probably does meet the standard of “signature significance”, that standard being that it happens only to bad teams, and that it essentially never happens to an average team.   

            Are the Royals working on a record?  Well, the math changes rapidly as teams get further below .500.   In baseball history there are not only sub-.400 teams, but sub-.300 teams, and a good many of those really terrible teams played in hitters’ parks in the 1920s and 1930s. . .the St. Louis Browns, the Philadelphia Phillies.   Some of those teams may have been blown out 50 or even 60 times in a season, thus may have had stretches of consecutive poundings as long as—I am speculating—8 or 9 games.

            But remember, the 2009 Royals have not merely lost 5 straight blowouts; they have lost 5 straight blowouts at home.   I would bet that nobody has done that since maybe the 1962 Mets. 

            Well, let’s look at the ’62 Mets.

            The Mets had only eight blowout wins, but actually had only 37 blowout losses—the same number as the 2005 Royals.  Their worst stretch was three straight blowout losses, and those happened in a three-game series in San Francisco; their worst stretch at home was two consecutive blowout losses, which happened four times.  

The 1899 Cleveland Spiders may have done even worse; they were outscored, on the season, by just short of five runs per game.  But watch out, Spiders.   The Royals are on a roll.

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

4.  Assigning Responsibility for Defensive Games

            Our next project, in the long process of explaining Win Shares and Loss Shares, is explaining how we assign responsibility for Defensive Game Shares.  

            Our process for calculating Fielding Wins and Losses for players has two stages:

            1)  We assign responsibility for Games, and

            2)  We assign credit for Wins.  

            These are separate and independent processes.   If a player is assigned responsibility for six defensive games and is assigned credit for four wins, his defensive won-lost record becomes 4-2.   It is not like the hitting system or the pitching system, in which the player is assigned responsibility for a certain number of games and is given a winning percentage based on his success, with the winning percentage being used to split the games into wins and losses.    The player is assigned games and he is assigned wins.   The percentage is a consequence of two separate calculations, rather than being the basis of the final calculation.

            It will take us many weeks to explain how credit for fielding wins is assigned.    It will, however, take only a few paragraphs to explain how responsibility for fielding games is assigned, and that is the task at hand.

            Since a team has three Game Shares for every game played and one-half of these are awarded to pitching and defense, a team has 1.5 pitching-and-defense Game Shares for every game played.  You may remember that two weeks ago, in the Monday morning blog for June 8, we assigned Game Shares to pitchers.  At that time I wrote the following:

            We start by figuring this number for each team (which is actually the fielder’s responsibility for the success of the team):

            38.8 Times (Wins + Losses)

            Minus Strikeouts

            Minus (Walks + Hit Batsmen) * 27/14

            Minus Home Runs * 9

            Plus Errors * 27/14

            Plus 3 Times Double Plays

            Let us contrast the 2005 Minnesota Twins with the 2006 Chicago Cubs.   Both teams had 162 decisions; we multiply that by 38.8, and that gives us 6285.6 in both cases.

            This culminated, several pages later, with the conclusion that the 2005 Twins had 69.54 Game Shares assigned to fielding, whereas the 2006 Cubs—because they had more strikeouts, more walks, more home runs allowed, fewer errors and fewer double plays—had only 41.85 Game Shares assigned to fielding.  (To keep the record straight, the Cubs actually had three more errors than the Twins.   This was contrary to the rest of the differences between them, and this reduced slightly the very large difference between the teams as to how much responsibility was assigned to the team’s fielders.)

            Anyway, what I’m saying is that I’ve already explained how to assign Game Shares to fielders on the team level.     What I need to explain today is how to assign that responsibility from the team to the individual player.   Let’s do the 2005 Twins.

            The players on the 2005 Twins, with their basic offensive stats, are as follows:

            You might notice that there’s a severe shortage of good hitters on that team.  With Morneau hitting just .239 and Jacque Jones just .249, the 2005 Twins finished last in the league in runs scored.   Since I’m here and I have to figure this anyway, I’ll give you the batting Win Shares and Loss Shares for this team:

            When your best hitter is Matt LeCroy, you may have some issues.   Anyway, the Twins have 69.54 Game Shares to be assigned to fielders.   The number of those to be assigned to each fielder depends on two things, and only two:

            1)  The number of outs the player makes, and

            2)  The number of innings he plays in the field.

            The rate at which defensive responsibility is assigned to a shortstop is the same as the rate at which defensive responsibility is assigned to a first basemen.   We’ll debate these choices later on; at the moment, I’m just trying to explain.   The first thing we need, then, is the number of outs made by each of these players:

            And the next is the defensive innings.   We’ll fill those in position-by-position:

It’s a really interesting team to look back on.   You can see the outstanding team that they have now being born here as the previous team fades.  Mauer and Morneau are in place but not yet the players they will become; Punto and Cuddyer are trying to find positions.  The outfield is aging (Shannon Stewart, Torii Hunter and Jacques Jones) and young players are trying to find places there.  They’re trying to sort out their young shortstops.    

Anyway, the only things that matter at the moment are the outs and the total defensive innings; the rest of that is just FYI.   To assign defensive Game Shares to individual players, first we divide the player’s outs made by the team outs made.  The team outs made was 4,406, so Joe Mauer is at 361 / 4406, which is .0826.   This number we divide by two, since this is one of two factors, equally weighted, which are used to assign defensive Game Shares.   This reduces this number from .0826 to .0413. 

Then, we divide the player’s innings in the field—999.2 (999.667) for Mauer---by the team innings in the field (1464.333).   This gives us .6827 (as Mauer played 68.27% of the team’s innings in the field.)   This we divide by eight, as there are eight players playing the field every inning (not counting pitchers, who are treated separately), and we’re looking for his share of the team’s responsibility.   This makes .0854.   This, again, we divide by two, since this is one of two factors being used to assign defensive responsibility.   That makes .0427.

Adding these two together, then (.0413 and .0427) we get .0840.   Joe Mauer will be assigned responsibility for 8.4% of his team’s Defensive Win Shares and Loss Shares.  The team total was 69.54, so Mauer is assigned responsibility for 5.82 Defensive Game Shares—69.54, times .084.   These are the Fielding Game Shares assigned to all members of the 2005 Minnesota Twins:

The regulars on this team—Mauer, Morneau, Cuddyer, Stewart, Ford, and Jacque Jones, are assigned 6 or 7 Fielding Game Shares each, rounded off.   The numbers are as high as they are because the Twins were a team that relied on their defense.   On most other teams from 2005, those numbers would be lower.

There is a wrinkle that doesn’t come up here, but which I had better explain anyway.   Occasionally we have players who are partially pitchers, and partially position players.   I gave a formula earlier to determine to what extent such a player should be treated as a pitcher, and to what extent he should be treated as a position player.   What we do here is, we discount the player’s defensive responsibility by the extent to which he is a pitcher.   If a player is 70% pitcher, we discount his defensive responsibility by 70%.   If a player is 20% pitcher, 80% hitter, we discount his defensive responsibility by 20%. 

This concludes our explanation of how Defensive Win Shares are assigned to individual players.   Now I need to explain and defend this process.   I’ll set that up as a dialogue between myself and.  .. ..Arthur Daley.  I’ll try to explain to Arthur Daley (New York Times sportswriter of the 1950s) why it is done this way.

Daley:    I don’t understand what making outs has to do with playing defense.

James:   Defensive responsibility is both created and assigned.   It is assigned from the team to the player by the team’s action in sending the player out to the field.   But before those defensive innings are assigned to the individual they must be created on the team level.   It is making outs that creates the need to play defense on the team level. 

Daley:   How’s that?

James:   For every three outs that a team makes at bat, they have to play one inning of defense.   This is the law of the game.

Daley:   So if a player makes more outs at bat or on the bases, that makes him a worse defensive player?

James:   If a player makes more outs, that gives him a larger responsibility to play defense, because it means that more defense must be played.

Daley:  Suppose that Gary Sheffield and Ruben Sierra are teammates.

James:   They must have been, somewhere.

Daley: And suppose that they alternate between playing left field and right field.

James:   They could.

Daley:  And suppose that their defensive statistics are just the same.

James:   Which they could be.

Daley:   Sheffield, because he has a .400 on base percentage, is going to make fewer outs than Sierra, whose on-base percentage is close to the Neifi Perez line.

James:   That’s right.

Daley:  But in your system, you’re going to conclude that Sheffield is a better defensive player than Sierra.

James:   He is a better defensive player.

Daley:   Let’s assume he’s not.

James:   Not better in the field?

Daley:   Let’s assume they are just the same in the field.   You’re going to conclude that Sheffield is better, because he makes fewer outs.

James:   He is better because he makes fewer outs.

Daley:  He is better at bat because he makes fewer outs.   But he’s not better in the field because he makes fewer outs.

James:   Yes, in fact, he is.   He is better in the field because he makes fewer outs at bat. 

Daley:  Why?

James:   Because making outs increases the team’s responsibility to play defense.   When you make more outs, that increases the team’s responsibility to play defense.   Therefore, if two players are the same in the field but one of them makes more outs, the one who makes fewer outs has to come out ahead when you compare the player’s defensive contribution to his defensive responsibility.

Daley:   I don’t want to think about it in those terms.  

James:   How you want to think about it, how I want to think about it, has nothing to do with it.   That’s the way it is.  Unless you can explain to me logically why it isn’t.

Daley:   Suppose that somebody else has a different fielding system.

James:   Like John Dewan, or Tom Tippett, or Tom Tango, or Dave Pinto.

Daley:  Or many others.

James:   Or many others.

Daley:  Are those systems wrong, to ignore the batter’s outs in assessing fielding performance?

James:   Well, I’m not here to say that anybody else is wrong.  I’m trying to explain why I did it the way I did it; I’m not saying other people are wrong.

Daley:   But you are, aren’t you?  If you argue that this is a necessary logical inference, aren’t you saying that anybody who fails to take this into account is doing it wrong?

James:   It’s a necessary logical step, given the way that I am approaching the subject.  I am assigning the responsibility for playing defense from the team to the individual.   In doing that, it is a necessary logical step to take into account the outs the player has made.   But other defensive systems approach the issue from different angles, and they may have it right, given their assumptions.

Daley:  But then the way they look at it may be wrong?

James:   Of course, the way that anyone looks at any issue could be wrong. 

Daley:  Do you think it’s wrong?

James:   I think it could be incomplete.   One difference between what I am doing and what other people may be doing is that this is a global system.  It is attempting to make an accounting of how everything in the game fits together. 

            I think that, if you are attempting to make an accounting of how everything in the game fits together, you need to take account of the effect of outs in increasing defensive responsibility.   But many, most or all of those other defensive systems that we talked about earlier are not attempting to make an accounting of how everything in the game fits together.    

Daley:  Let me try again.

James:   Go ahead.

Daley: Suppose that you have two teams which are identical in every respect, except that the right fielder on one team is Gary Sheffield, and on the other team it is Ruben Sierra.

James:   OK.

Daley:   Suppose that Sierra and Sheffield are the same in the outfield.

James:   OK.

Daley:  Will the team with Sierra in the outfield allow more runs?

James:   No.   They will win fewer games, but they will not allow more runs.

Daley:  They will win fewer games because they score fewer runs, but they will not allow more runs?

James:   I suppose not.

Daley:  Then why are the players not even, as defensive players?

James:   Because the teams you have postulated cannot exist.   You’ve created an imbalanced statistical universe.

Daley:  How so?

James:   If one player makes more outs than the other, then one team must make more outs than the other.  If one team makes more outs than the other, then they must play more innings of defense.    You’re creating a situation in which one team makes 4400 outs at bat and the other makes 4300 outs at bat, everything else is the same, but they play the same number of innings  on defense.   That can’t happen.  Your theoretical universe cannot exist.

Daley:  But we account for that difference on offense.

James:   You can account for that difference on offense if you want to.   But it isn’t on offense.   It’s on defense.  

            What I am saying is that, while we are in the habit of thinking of offense and defense in baseball as un-connected, they are in fact not un-connected.   There is a very important connection between them, which is the rule that for every out you make on offense, you must record an out on defense.   9^th innings for home teams excepted, of course.

Daley:  I think we should make a rule requiring the home team to bat in the ninth inning if they are ahead, just so the stats balance.

James:   Right.   It’s like those games they play in September after both teams are eliminated.   If they don’t matter, why do they play those games?   And if it’s for the integrity of the schedule, why don’t they make the home team hit in the bottom of the ninth?

Daley:  Anyway.

James:   Seriously.

Daley:  If you look at this as a practical issue rather than a theoretical one, does this policy help or hinder?

James:   It both helps and hinders, but it helps a great deal more than it hinders.

Daley:  Where does it hinder?

James:   Ted Williams vs. Vic Power.   Ted Williams rates as a better defensive player than he was commonly perceived to be through much of his career, and no doubt this is because of his very good on base percentages, which reduce his responsibility to play defense.  Vic Power, on the other hand.  ..while my system does a very good job of getting higher defensive values for the Gold Glove first basemen than for the Jason Giambis of the world, it doesn’t do well by Vic Power.  His low on base percentages don’t cause this, but they don’t help, either. 

Daley: And where does it help?

James:   Over the course of a career, this practice is a tremendous help in pushing players toward the defensive winning percentages that they deserve.   Over the course of a career a good defensive player who is a weak hitter will virtually always have a high ratio of defensive innings to outs made, because he will virtually always be deployed by a manager so as to maximize his defensive innings and minimize his at bats, whereas a good hitter who is not a good fielder will virtually always be deployed by his manager, at some points in his career, so as to maximize his at bats and minimize his innings in the field.   The result of this is that defense-first players, in our system, will virtually always wind up with high defensive winning percentages, while poor defensive players will virtually always wind up with low defensive winning percentages, for their careers.   A highly desirable outcome.  

Daley:   I’m not sure I am following you.  Why exactly does this element of the system create a bias toward the better fielders?

James:  A player is given defensive accountability both for his innings played, and for his outs made, but he only has the opportunity to earn defensive credits—defensive win shares—while he is in the field.   He has to offset the outs he makes at bat while he is in the field.   If a player is used as a defensive sub, or if he is pinch hit for, he tends to have a high ratio of innings in the field to plate appearances, and thus a high ratio of defensive innings to outs made.   This gives him a larger opportunity to earn defensive credits, compared to his defensive responsibility, so he tends to do well.   Conversely, if the player is used as a pinch hitter, or if he is taken out of the game for defense, then he tends to have a low ratio of defensive innings to plate appearances, and thus a low ratio of defensive innings to outs made.   Thus, he has less opportunity to contribute in the field, relative to the defensive responsibility he is assigned.

Daley:   So it is biased against that player.

James:  If assigning high defensive winning percentages to good defensive players and low defensive winning percentages to poor defensive players is a bias, then yes, it’s biased.  But I don’t think you can go to hell for that.  

Daley:  But there must be some problems with the system, right?

James:   Sure.

Daley:  We’ll get to those in a minute.  First, I wanted to ask about a couple of other features of your stupid system.

James:   Go ahead.

Daley:  You assign the same responsibility to play defense to a shortstop that you do to a left fielder.  

James:   Right.

Daley:  Do you really believe that a left fielder has the same responsibility to play defense that shortstop does?

James:   From a certain perspective, no, but from a certain perspective, yes.  

Daley:  A weasel answer if I’ve ever heard one.

James:   It depends on exactly what you mean by the word “responsibility”.  We’re trying to represent the impact of the player on the team.   From the perspective of the team, everybody has to take responsibility for getting the other team out of the dugout and back into the field.  I don’t know that Phil Rizzuto has more responsibility for that than Johnny Mize, just because Rizzuto is playing shortstop and Mize is playing first.   Rizzuto has more skill at playing defense than Mize, and he is more “responsible” for getting the other side out, in a certain sense, because he accomplishes more toward that goal.   But he is not more “responsible” in the sense that it’s one player’s job more than the others’.

Daley:  But the average shortstop is contributing more to the defense than the average left fielder, surely?

James:   Certainly.   And our system reflects that.

Daley:   But if the average shortstop is contributing more to the defense than the average left fielder, wouldn’t it be easier to just assign him a larger responsibility for playing defense, and thus compare the player to the average at his position?   The average left fielder is 1x; the average shortstop is 2x.

James:   If you can figure out how to make it work that way, power to you.   I can’t.  That is what most global analytical systems do, but I think it creates both logical and practical problems.   Logically, if you have a first baseman who creates 70 runs and makes 300 outs and a teammate shortstop who creates 70 runs and makes 300 outs, they are the same as offensive players.   If we say that they are even as hitters but that the shortstop is more valuable because of his defense, that, to me, is more accurate than saying that the shortstop is a better hitter because he plays shortstop.   He’s not a better hitter because he plays shortstop.   He plays shortstop because he is a better fielder.

            That said, I am far from convinced that I have exactly the right structure in place to address those issues.   I tried it every different way that I could think of, and this seemed to create fewer analytical absurdities than anything else.   I experimented with assigning a larger defensive responsibility to the players at the right end of the defensive spectrum (shortstop and center field)  than to those on the left end (first base and left field).    It just doesn’t really work, for me.   If you assign the shortstop 10 game shares and the first baseman 3, then the average shortstop is a .500 fielder and the average first baseman is a .500 fielder, which means that, when you consider hitting as well, the average first baseman has a higher winning percentage than the average shortstop, which is irrational, so then you have to go back to pretending that the shortstop who creates 70 runs is somehow a better hitter than the first baseman who creates 70 runs.   But if a shortstop who creates 70 runs is equal, as a hitter, to the first baseman who creates 90 runs, then why does the first baseman hit third and the shortstop hit 7^th?  

            My system says instead that “first basemen are better hitters than shortstops; shortstops are better fielders than first basemen.”   I think that statement more accurately describes the real universe than does the other alternative. 

Daley: Let’s talk about parts of your system that don’t work.

James:   OK.

Daley:   What parts of your system don’t work?

James:   Well, one thing that doesn’t work as well as it should is the team performance bias in the individual fielding records.

Daley:  Team performance bias. . .?

James:   Take a 90-win team that allows 700 opposition runs and a 60-win team that allows 850 opposition runs.   It is overwhelmingly likely that the 90-win team is a better defensive team than the 60-win team.   In fact, the Royals. . .the reason they have been getting blown out the last week is their fielding.   Their defense over the past week has just been horrific.   

            The problem is, when you look at 60-win teams and 90-win teams in the fielding stats, there isn’t much difference.   There’s a small difference in fielding percentages, but that’s just a few plays.   No matter how good you are or how bad you are, as a team, you’re going to record 27 outs a game.   If you’re a bad team, you’ll have just as many double plays, on average, and if you’re a bad team you’ll actually have more outfield assists, and more assists at the positions where assists tend to carry more weight on defense.

            To correct for this, we start with the assumption that a 90-win team allowing 700 opposition runs is better in the field than a 60-win team allowing 850 runs, as they almost always would be.   Then, starting with the assumption that there is more defense being played on the Dodgers than on the Royals, we assign credit for that defensive accomplishment from the team to the individual players.

            This has the effect, over a career, of biasing the system in the direction of the truth.   It causes players who play on good teams to rank, over the course of a career, as better defensive players than players who play on bad teams.   It causes them to rank that way because they are that way.

            The problem is, the bias works well over the course of a career, but it causes off-the-wall variations year to year.   If you take a team that has an off season, you will very often find that the second baseman has defensive records over a period of years that read 5-2, 4-1, 1-6, 4-2, 4-3.   Everybody on the team gets hammered for their defense in the off season, and there clearly is some problem with the system, because the player shouldn’t be punished for the failures of his team, beyond the extent to which his personal failures have contributed to the failure of the team.   I know there is some problem related to this element of the system, but I don’t know what it is and I don’t know how to fix it.

Daley:  Thank you for your time.   This is the first interview I have done since I died in the mid-1950s.

James:    I hope it was worth waiting for. 

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

5.  McCovey and Bill Terry

            When you think about it, Willie McCovey and Bill Terry have much in common.   Both players:

            1) Played first base,

            2) Played their best years for the Giants,

            3)  Batted left,

            4)  Threw left, and

            5)  Are in the Hall of Fame.

            Both players are from the South.   McCovey was born in Mobile, Alabama; Bill Terry 330 miles away in Atlanta.  Both players’ names are variations of “William”—Willie and Bill.   Both players were, in a manner of speaking, rookie sensations; I’ll explain about Terry in a minute.   Both players, however, found themselves on a Giant team that already had a Hall of Fame first baseman in mid-career.   The 1920s Giants had George Kelly; the 1950s Giants had Orlando Cepeda.   Both players thus struggled for two or three years to get regular playing time, despite their obvious ability with the bat.   Both players, however, were able to get in the lineup in time to have a Hall of Fame career.   At the age of 31 both players had their best seasons, and, we will argue, both players won the NL MVP award; more on that later.  Both players were elected to the Hall of Fame while still young enough to enjoy it—McCovey at age 47, Terry at age 55.  

            When I was young, Bill Terry was a mythic figure, a career .341 hitter who managed the Giants after John McGraw.     No one, up until 1968 and probably after, would have dared to put McCovey in the same group with Bill Terry.  McCovey’s career average was just .270, and he played much of his career in the shadow of Willie Mays. 

Of course, times and people and perceptions change, and, as no one anymore remembers seeing Terry play, his image has faded and his reputation dimmed.  In the last Historical Abstract I rated McCovey the 9^th best first baseman of all time, and Terry in the 20s.

            Let’s look at it again.     Willie McCovey came out of the minors at the end of July, 1959, and was a rookie sensation of the first order:

            McCovey played at an MVP level the last two months of 1959, and not only won the NL Rookie of the Year Award playing in 52 games, but won it in a unanimous vote.   There was, however, a fly in the ointment.   With McCovey at first base Cepeda had to play the outfield.   Cepeda was a really terrible outfielder, and the Giants went 10-13 in September, 1959, finishing the season five games behind a Los Angeles Dodger team that frankly the Giants should have whupped up and sent home crying.  

            At this point in the story we encounter something that I don’t quite understand.   I thought I understood it, but, doing some research, I see that I didn’t.   Over the first two months of the next season (1960), McCovey played OK; not great, but pretty good.  McCovey had played 40 games through May 31, hitting .267 with 9 homers, 32 RBI. . .not bad numbers.   The Giants as of May 31 were 26-16; as of June 8 they were 30-19.   They had a small slump, dropping to 33-25 as of July 17—in second place, four games out of first.

            On that date the manager, Bill Rigney, either quit or (more probably) was fired.   What apparently happened—I’ll act on this assumption until somebody corrects me—was that the organization panicked when they had a little slump, and fired Rigney.  McCovey had been part of the slump; he was down to .247, .518 slugging.  The new manager, Tom Sheehan, benched McCovey, and pulled Cepeda out of the outfield to play first base.   It sounds stupid, but I think that’s what happened.   McCovey doesn’t appear to have had a significant injury, in that he continued to play (as a substitute) for the rest of the season, with a couple of weeks out in late July. 

            McCovey finished 1960 at just .238, although, with his power and walks, his OPS was still 130 points above the league norm:

            McCovey hitting .240 was still a good player, and benching him decidedly did not help the team.   The Giants, in contention when Rigney left, were down to 69-71 by early September, although they rallied to finish over .500.  

            Alvin Dark replaced Sheehan as the manager in 1961, but Dark was unwilling to put Cepeda back into the outfield, so McCovey stayed on the bench.    He is one of very few Hall of Famers to begin his career with several years as a part-time player:

            A part-time player—but one of the best hitters in the league.   In 1963 Dark found a sort of solution:  move McCovey to the outfield.   McCovey tied with Hank Aaron for the league lead in home runs:’

            McCovey was probably a worse outfielder than Cepeda.   His 1963 record breaks down to 22-1 as a hitter, but 2-5 in the field.   And, in 1964, McCovey had serious trouble with his feet, and didn’t hit:

            McCovey’s career to this point had been one of constant struggle and frustration—and yet his career won-lost record to this point is 76-32.    In 1965 Cepeda was hurt and missed almost the entire season, which enabled McCovey to play first base and hit 39 homers:

            Let us go back now and catch Bill Terry up to age 27.  Terry entered pro baseball very late, and reached the majors at age 24.  In 1924, his rookie season, he had an undistinguished season, but got hot in the World Series, and played a critical role in one of the most fascinating managerial encounters of all time.  In the first game of the series Terry went 3-for-5 with a home run, and the Giants won 4-3.   In the second game of the series a left-hander started for Washington, so Terry didn’t play, and the Giants lost 4-3.   In the third game Terry went 2-for-4, and the Giants won 6-4.  In the fourth game of the series a left-hander started, so Terry didn’t play (except for a pinch hitting appearance), and the Giants lost 7-4.  In the fifth game of the series, Terry had a triple and two walks, and the Giants won 6-2.  In the sixth game of the series a left-hander started, so Terry didn’t play—and the Giants lost.  

            By this point everybody in the baseball world had noticed something.   It seemed to be better for the Senators when Bill Terry didn’t play.   Bill Terry, a rookie who had hit .239 during the season, had become the focal point of the World Series.  

            Bucky Harris, the Senators’ manager, had a problem.   What he wanted to do in the 7^th game of the World Series was to start George Mogridge—a lefty who had beaten the Giants earlier in the series—but then switch in the middle of the game to his right-handers, Firpo Marberry and Walter Johnson.   Marberry was a hard-throwing rookie who was really good for two or three innings, then often ran out of gas.   Walter Johnson was 36 years old and had only one day of rest, but. ..well, he was Walter Johnson, and he was 23-7 that year.   Harris figured he was good for two or three innings to close out the series.

            Harris knew, however, that if he started Mogridge Bill Terry would open the game on the bench, which meant that he would enter the game as soon the right-handed pitchers entered the game.   He would have to face Terry two or three times, and Terry—hitting .500 for the series--had been dominating the Senators’ right-handed pitchers, including Walter Johnson.   Harris thus decided to start a stalking horse.   He put onto the mound as his starting pitcher Curly Ogden, a 23-year-old who was a fifth starter type, but who was right-handed, thus getting Terry into the starting lineup.   He let Ogden face two batters, and then pulled him out of the game, bringing in the lefty, Mogridge, who was his “real” pitcher for the day.

            Harris now had a lefty-on-lefty matchup against Bill Terry.   That was his goal.  This strategy, which has only been used a few times in baseball history, was done explicitly to contain Bill Terry; there wasn’t any other reason behind it.    This gave McGraw a dilemma:  leave Terry in the game to face the left-handed Mogridge, or take him out of the game and lose him for the rest of the game.   McGraw opted to let Terry face Mogridge—twice.    Terry went 0-for-2.   His third time up McGraw pinch hit for him.   Harris immediately switched to his right-handers, the right-handers shut New York out for the rest of the game, and the Senators won the game (4-3 in 12 innings) and series (four games to three).  

            Despite his success in the 1924 World Series, and despite hitting .319 in 1925, Terry struggled for two years after that, just as McCovey did after 1959, and for the same reason.   If Terry played first base George Kelly—a very good first baseman, and an established star—had to play somewhere else, and George Kelly was nowhere near as good somewhere else as he was at first base.    Terry tried to play the outfield, but he was terrible in the outfield.  (There is an anecdote somewhere I wish I could remember clearly.  It had something to do with Terry taking fly balls in the outfield, trying to get some playing time, and McGraw ordering him out of the outfield before he killed somebody out there.  Anybody remember that anecdote?)   Anyway, this is Bill Terry up to age 27:

             Through age 27, then, McCovey had struggled mightily but had a career won-lost log of 98-39, whereas Terry had struggled mightily, and had a career won-lost record of 30-22.    From ages 28 to 30 both players were in the lineup, and both players were consistently outstanding:

            McCovey led the National League in 1968 in home runs, RBI, slugging percentage and OPS, at .923.   Bill Terry in 1929, hitting .372, was good but not as good.

His .941 OPS was 12^th among 46 players who qualified for the National League batting title that season.  

            At age 31 both players were, in a manner of speaking, MVPs:

            McCovey was the NL MVP in 1930, and had one of those rare over-the-top seasons, a season so good that the credit we give McCovey for his team’s wins is larger than the area of responsibility assigned to him, resulting in a winning percentage greater than 1.000.   Bill Terry, on the other hand. . .all he did was hit .400.   He was the last National Leaguer to hit .400, and the last player to hit .400, other than Ted Williams in 1941. 

            My memory—and again, you will have to forgive my half-baked memories of these things—but my memory is that there were two National League MVPs in 1930, given by different sources.   One source gave the NL MVP Award to Hack Wilson, who drove in 190 runs at the time, and picked up another one years later.    The other source gave the NL MVP Award to Bill Terry.   That’s what I remember.  

            In any case, Terry’s career won-lost record we have now figured as 127-56, which is extremely good.   McCovey comes in at 200-46, which is ridiculously good.   And both players had great seasons again at age 32, although their numbers were down:

            In 1971 McCovey, now 33 years old, began having chronic problems with his feet.   I misstated that a little.   McCovey had always had problems with his very long, very narrow feet.  By his early 30s these problems were getting worse, and also he began having other problems with his lower extremities, his ankles and his knees.    Although McCovey remained a devastating hitter with outstanding production rates, it would always be a struggle for him to stay in the lineup.    Terry, on the other hand, remained a regular—and a highly productive regular—for four more seasons:

            Terry was getting 200 hits every year except 1933, and the National League had entered a pitcher’s era.   Not a lot of people understand this; people think “1930s—big hitting era.”   The National League ERA in 1933 was 3.33.   The American League continued to pile up big hitting numbers throughout the 1930s, but the National League ERAs after 1933 were 50 points to a full run lower than the American League.  

            Anyway, by 1935 Terry had significantly closed the gap on McCovey.   By age 36 Terry’s career won-lost log was 247-93; McCovey was 280-67.   McCovey was still 29½ games ahead, but Terry had caught up by about a dozen games.  But Terry had only one more season to play:

            Terry was the Giants’ manager by now, and, no longer able to dominate as a hitter, he decided to call it quits as a player.   He managed the Giants to the NL Championship in 1936 and 1937.

            My opinion, in all honesty, is that Terry was a poor manager.   He was harsh and sarcastic, and he was not supportive of young players.    He was able to win in ’36 and ’37 because the other good team in the league was being managed by Frankie Frisch, who was a terrible manager.   Nobody else had the talent.

            McCovey left the Giants in 1974.   He was with the Padres for a few years, played 11 games for Oakland one year, and then returned to the Giants in 1977:

            I visited San Francisco for the first time in June of 1977, on a bus trip with my girlfriend, who is now my wife.    The Giants were running advertisements promoting Willie.   “Come on out and see . . ..(footage of Willie McCovey, hitting a home run, artificially loud crack of the bat). . ..Big Mac.   Just the way it used to be!”   We didn’t have enough money to go to a game.  He wasn’t just the way he used to be, of course, but he had one last good year, and there was something heart-warming about it, and something very sad.   He left the Giants where found them, struggling to stay out of the second division.

            On the bottom line, Bill Terry was 254-99 in his career, a .720 percentage, while McCovey was 335-123, a .731 percentage.   Terry was a much better fielder than McCovey:

            And Terry was also a truly great hitter, with an .800 offensive winning percentage.   But McCovey was even greater:

            McCovey was clearly greater, but Terry probably was better than I gave him credit for being in the last Historical Abstract.  I rated him the #26 first baseman of all time, behind Norm Cash, Mickey Vernon and Fred McGriff (who was always referred to as the new Willie McCovey.)  I think probably he (Terry) deserved better than that.   Terry’s career started very late and ended a little early, but in between there he was very much a championship quality player.