1.  The Heath Bell Question

            Heath Bell is leading the National League in Saves while pitching for a team that doesn’t win an awful lot of games, which is slightly counter-intuitive, and thus puzzles the simple-minded person living inside us.    The more sophisticated parts of our brain understand that Save Opportunities are not directly proportional to Team Wins.   When a team wins a lot of games, some of them are blowouts.  When a team wins fewer games, more of those tend to be close games.   Thus, it’s not stunningly improbable that a pitcher can lead the league in Saves while pitching for a team that’s a little under .500.  

            It is not stunningly improbable, OK, but what is the relationship?   If you pitch for a 90-win team you don’t get 25% more Saves than you would pitching for a 72-win team, but do you get 10% more?  15% more?  12.5% more?   What is the relationship?

            I studied this in the 1980s, but those studies would be obsolete even if Save totals hadn’t entirely changed since the 1980s.    In the 1980s you could lead the league in Saves with 30.   That’s harder to do now.  

            I decided to study the issue in this way.   First, I took a spreadsheet that has each team’s Wins and Saves.   Second, I sorted teams by decades—1930s, 1940s, 1950, etc.   Third, I eliminated the teams from the strike years and the 1960-1961 teams that played only 154 games while everybody else in the decade was playing 162.  

            Then I sorted each group in declining order of team wins, and figured a moving average of saves related to wins.   I’ll finish explaining that in a moment.   This is the data for the 1930s:

            Each line in this group (or the ones below) represents 50 teams.   When you arrange the teams in declining order of wins and form 50-team averages on each line, at some point the teams will have an average very near 93.00 wins.   The “93 wins” line above represents the 50 teams from the 1930s that averaged mostly nearly 93.00 Wins.   These teams generally average 153 decisions, rather than 154, because of rainouts. 

            In the 1930s—the starting point of our study—teams averaged “Saves” in a little less than 20% of their Wins.   Of course, these Saves were not credited at the time; these are based on after-the-fact studies, and the standard that was used to figure these, in the late 1960s, was simply “finishing a game for another pitcher who gets a Win.”    If you got the last out of a 17-0 game, in the ICI counts, that was a Save.   Some of the Saves counted here were, in fact, 17-0 games, or 18-0, or something like that; that was the rule that was used. 

            Anyway, as you can see, at this time there was a relatively straight-line relationship between Wins and Saves.   Comparing .600 teams to .500 teams; the .600 teams had 20% more Wins; they had 20% more Saves.   There was actually some “bending” of the line, in the range between 60 Wins and 85 Wins.   Sixty-win teams averaged Saves in 19% of Wins.    85-win teams averaged Saves in a little less than 17% of Wins.   Let’s go on to the 1940s.  From now on I’ll chop these charts down a little:

            This chart ends at 61 Wins because we run out of teams.   There aren’t enough bad teams in the 1940s to make a group of 50 teams averaging 60 Wins, so we drop the count at 61.

            Anyway, the Save percentages increased very slightly in the 1940s, from 17% (for a .500 team) to 19%.    The curve representing Saves versus Wins is just the same.  It starts off highest at the bottom—21%--then drops steadily to 85-86 wins, then goes back up slightly above 86 Wins.   Again, as above, .600 teams in the 1940s had 20% more wins than .500 teams—and about 20% more saves.   Let’s look at the 1950s:

            Here there is a real jump in Save percentages, from about 19% of Wins to about 28%. (77-77 data is slightly out of line.)  The pattern we have seen before is still visible here, but it’s even flatter.   The percentages of Wins which were Saved was essentially the same for first-place teams as for last-place teams—but remember, the awarding of Saves is still indiscriminate, paying no attention to the score.   It is the margin of victory which causes the percentages to be different for good and bad teams, so taking that out of it, all you really have is a small difference in the percentage of non-completed games.   The same will be true for the 1960s, and in the 1960s we will be dealing with teams playing 162-game schedules:

            Here we can observe:   Bad teams in the 1960s had more Saves than good teams in the 1950s.   The best teams in the 1950s averaged about 26 saves.   The worst teams in the 1960s averaged about 27.   The percentage of team Wins which were “Saved” jumped in the 1960s from about 28% to about 38%.   The percentage is highest for last-place teams (43% at 68-93), declines until we get to second- or third-place teams (37% at 88-74), and then goes up slightly at the top end of the scale.

            In the 1970s, finally, we have actual Saves.   “Actual” saves:

            1)  Were awarded contemporaneously, rather than after the fact, and

            2)  Depended at least a little bit on the score.

            The “official” Save rule, first adopted in 1969, was changed several times during the 1970s, so that what we’re really looking at here is a mishmash of different Save categories.  But this is the data for the 1970s, such as it is:

            The overall Save percentage, which had jumped from 27% in the 1950s to 37% in the 1960s, didn’t increase much in the 1970s (to 38%), for two reasons.   First, as Saves were being officially credited, the counts no longer include Saves credited in lopsided wins.   Second, after the Designated Hitter Rule was introduced (1973) the American League for a few years had a lot of complete games.     The American League, which had 382 Complete Games in 1970, had 650 in 1974.    As the American League in 1974 had 969 Wins, one can see that a very large percentage of Wins were complete games (although a few complete games also occurred in Losses.)  

            Here again we have the familiar percentage:  Save percentages are highest for last-place teams (40%+), drop steadily until we get to second- or third-place teams (35.8% at 85-77), and increase slightly at the top end of the scale. 

            The decrease in the percentages of Wins which were Saved is more significant here, but we’re still looking at a reasonably flat line.    .500 teams have 25% more wins than .400 teams.   In the 1970s, they had 16% more saves.   .600 teams have 50% more wins than .400 teams.   They had 36% more saves.   MOST of the increased Wins yielded increased Saves.  But in studying the 1970s, we are still a long way from studying modern bullpens.   One can’t say when “modern” bullpens began—but it’s certainly not before 1995.   Let’s look now at the 1980s:

            In the 1980s the percentage of Wins that were Saved increased from 38% to 47%.   The line, however, gets even flatter.  For last-place teams, 48% of Wins were Saved.   For first-place teams, 46% of Wins were Saved.   There’s a difference, but it’s a small difference.   In the 1990s the “Save Percentages” moved over 50%:

            But there remains a fairly straight-line relationship between Team Wins and Team Saves.  The top 50 teams in this study averaged 43% more Wins than the bottom 50—and 37% more Saves.   

            And finally, our own generation:

            The Save percentages for our current decade have actually gone down, presumably because the very high numbers of runs scored in the years 2000-2005 increased the number of games decided by more than 3 runs.

            Pitchers on teams winning about 75 games, in our own time, average about 38 Saves.    Pitchers on teams winning 95 games average about 46 Saves.   It’s not a huge difference, and it is not shocking that Heath Bell can lead the league in Saves while pitching for a weaker team, particularly since he pitches in a pitcher’s park (meaning fewer games decided by more than 3 runs.)    It’s not shocking, but the relationship of Teams Wins to Team Saves is much more of a flat-line relationship than I would have guessed that it would be.      

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

2.  RBI vs. RBI Opportunity Data

            A year ago on this site, in articles published on May 23 and May 28, 2008, I proposed the concept of “RBI Opportunities”.   We now have the actual data proposed by those articles on line in the statistics section, and I thought I would take a minute here to take a look at some of it. 

            What are “RBI opportunities”?   RBI Opportunities are the total of actual RBI, plus missed RBI opportunities. 

            OK then, what are missed RBI opportunities?

            Batters are charged with missed RBI as follows:

            1.00 for a runner left on third base with less than 2 out,

            0.70 for a runner left on second base or left on third base with 2 out,

            0.40 for a runner left on first base,

            0.10 for an out made with the bases empty, and

            1.00 for grounding into a double play. 

            No missed opportunity is charged to a player who does not make an out, and no opportunity is charged to a player who records a successful sacrifice bunt.  

            We have actual data now, and I have to say:  It is really good data.   This thing works.   Most of the people who would traditionally bat in the middle of the order have good RBI percentages; most of the people who wouldn’t, don’t.    We have actually developed here a meaningful way of distinguishing RBI ability from RBI position.   We have a way of measuring—not perfectly I suppose, but accurately—to what extent a player drives in 100 runs because he’s a good RBI man, and to what extent he drives in 100 runs because he has a lot of chances to drive in runs.   This opens up to us questions that were more difficult to answer before, questions about lineup construction, for example.  Dustin Pedroia last year hit .326 with 54 doubles and 17 homers.   If given the chance, would he be just as good an RBI man as Jason Bay or Kevin Youkilis?

            No, he wouldn’t.   Pedroia’s RBI percentage last year was .331, which is OK but middle-of-the-pack, far below Youkilis’ .449, Jason Bay’s .391, or even David Ortiz, having a poor season in 2008, still at .415.    These were the ten most effective RBI producers in the major leagues in 2008:

            Pujols’ RBI percentage has always been over .400, and in 2006 reached a peak of .505.   Manny Ramirez in 2002 finished at .502.  Although Joe Mauer is at .600-something so far this year, Pujols and Manny are the only players I have found, other than Barry Bonds, to finish a season over .500.   Bonds has ridiculous numbers in this area as he does in every other—56.6% in 2002, 56.5% in 2004. 

            Of course, the revelation that Albert Pujols and Manny Ramirez are the most effective run producers in the majors is hardly stop-the-presses material, but that’s not the point.    Would you have guessed that Nate McLouth was producing RBI more efficiently than Justin Morneau (although Morneau was still extremely good), that Alfonso Soriano was a touch ahead of Adrian Gonzalez, that Dan Uggla was ahead of A-Rod, or that David Murphy was ahead of Jermaine Dye?

            These charts only go down to Jermaine Dye, but Dye was still easily in the top half of run producers.   The worst RBI man in the majors in 2008, among players batting 400 or more times, was Pujols’ teammate, Cardinal shortstop Cesar Izturis:

            Izturis has no power, of course, but he also hit .275 with the bases empty, .216 with runners in scoring position.   Chone Figgins has had years in the past when he was a fairly decent RBI man, with a .360 RBI percentage in 2007.   Last year, he just didn’t produce in those situations.  Of course, everything is relative.  Izturis and Figgins look like Joe Carter and Tony Perez compared to Pittsburgh’s Brian Bixler, who had 30 RBI opportunities (29.6) but produced only 2 RBI, or the Giants’ Ivan Ochoa, who had 40 chances (39.7) and produced only 3 RBI.  But those guys had just over a hundred at bats; I used a minimum for this list of 400. 

            Nelson Cruz was on the other side of that.  Given just 115 at bats in 2008, he had 56.4 RBI Opportunities, but 26 RBI—a Pujols’-like production rate.   Chris Dickerson and Mike Napoli had very high rates.   

            Given a systematic method to measure these things, given that we believe in that method—and I do believe—we are then in position to ask any number of questions.   Who led the majors in RBI opportunities in 2008?   The man who also led in RBI:

            Ryan Howard led the majors in RBI Opportunities, but he got that position on merit; his .446 production percentage was one of the best in the majors.   One can’t say the same about Garrett Atkins or James Loney.  Those guys had huge numbers of opportunities to produce runs, and by and large, didn’t.    Loney drove in 90 runs, Atkins 99.   Those numbers look good.   They’re not good.   Both men were below average as RBI producers, given the number of chances they had.   These were the major league hitters who wasted the most RBI chances in 2008:

            What’s the normal standard for RBI to RBI Opportunities?  It’s easy to remember:  it’s one out of three.  It was actually .332.    The 301 players for whom I have looked up data for 2008 had 52,194.5 RBI Opportunities, and 17,343 RBI.   One out of three.    Above .400 is excellent; below .250 is poor.

            The typical player has about 41 RBI opportunities per 100 at bats.   The player with the highest ratio last year was Jason Giambi (263.4 opportunities in 458 at bats, .575) while the player with the fewest chances to drive in a run was Ichiro (169.3 opportunities in 686 at bats, or .247.)    Ichiro was both a player with few opportunities, and a poor RBI man given his opportunities. 

            In general, baseball does a good job of delivering the most RBI opportunities to the most productive hitters; this is one of the things we learn from having this data.   This, however, is not universally true.   There were some players who had very good RBI production rates, but relatively few opportunities.   Johnny Damon had an RBI success rate of .397, and David DeJesus a rate of .401—yet they had relatively few opportunities to drive in runs.   The same for Nate McLouth, Curtis Granderson, Denard Span, Hanley Ramirez, Grady Sizemore and Ian Kinsler.   What do you notice about those men?

            They all run well.   Sometimes, when a player runs well, that is used as a reason to put him in a spot where he won’t drive in runs.    It is possible, for all I know, that this is the correct and logical way to construct a lineup.   To know for sure, we would have to measure the number of RBI opportunities created by batting these players leadoff, and then contrast that with the RBI lost.    Also, some of these men may show up as productive RBI men because, in 2008, they hit well with runners in scoring position.   Of course, there is no such thing as an ability to hit with runners in scoring position (other than the general ability to hit), so that also might be an illusory benefit that would disappear as soon as you pursued it.

            On the other side of that chart—players who had a lot of RBI opportunities and didn’t do much with them:   Varitek, Francoeur, Loney, Atkins, also Gary Sheffield, Jack Hannahan, Paul Konerko, Ben Francisco and—go figure—Marlon Byrd.  I don’t know why Marlon Byrd had a larger-than-average number of RBI opportunities, but he did. 

            In general, RBI effectiveness is closely correlated to slugging percentage—but this, also, is not always true.    Jeff Mathis in 2008 batted .195 with a .318 slugging percentage—but drove in 42 runs in half-time play, with an RBI success rate of .346.   Brian Buscher, Jason Michaels and Jesus Flores all were much more effective RBI men than they should have been, based on their hitting stats, as were Ryan Garko and Yadier Molina.   Not to mention Joey Gathright (did you have to mention Joey Gathright?)   Gathright drove in only 22 runs, but, given his .272 slugging percentage, that’s a lot more than you would expect him to drive in.   On the other end. …well, the guys we’ve already beat up on.   Varitek, Izturis, Chone Figgins, those guys.   Also A-Rod, Rickie Weeks, Miguel Tejada and Jhonny Peralta.

            The 2009 data is still too raw to mean a lot.   Pujols is over 50%, Jason Bay just under 50%, Mauer is at 60%.   I’m very pleased with the data, and I expect to write more about it as the season runs.   I think it provides a pathway into questions about value and about lineup construction that will enable us to discuss with facts things that we sort of knew anyway by intuition. 

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

3.  Assigning Win Shares and Loss Shares to Pitchers

            Last week I explained how to assign Game Shares to pitchers, Game Shares being the sum of Win Shares and Loss Shares.   This week I’ll explain how to divide those into Win Shares and Loss Shares.

            It’s a fairly straightforward process, but before I get to that, I need to say something else.   Two weeks ago, when I explained how to assign Win Shares and Loss Shares to pitchers based on their hitting, a discussion developed in the reader comments section about the defense of Designated Hitters.  That discussion was heading down a blind alley, so I tried to explain, in last week’s blog, that the discussion was proceeding on false assumptions.   This simply caused a few people to pick up that comment and roar off down a different blind alley, based on a different set of false assumptions.  

            OK; I give up.   Obviously you have the right to say anything you want to say about the Win Shares/Loss Shares system, true or false, and also, the things you have said have been very helpful to me.   I have already fixed a few problems with the system based on reader comments, and I know how to fix a couple more, based on your feedback.   I appreciate this very much.  

            But your comments are much more useful, to me, when you stick to debating the parts of the system that I have explained and therefore you can understand, rather than departing into debates on parts of the system that have never been explained to the public, and which therefore you can’t actually know anything about.  I’m trying to go through an orderly process of explaining this system to the public, getting feedback on it, and making it work better.  Speculation about parts of the system that have not been explained is not useful, and in the future I’m just going to ignore it.  

            Alright, the pitcher Winning Percentage is based on two figures:  the Run Context Rate, and the pitcher’s Runs Allowed Rate.    For porpoises of illustration I’ll use the two candidates for the American League Cy Young Award in 1980, Steve Stone (who won the award) and Mike Norris (who finished a close second in the voting, 100 points for Stone, 91 for Norris.)   What the hey; let’s include Goose Gossage, who finished third in the voting with 37.5 points.  

            Steve Stone in 1980 pitched 251 innings, struck out 149 batters, walked 101, had a 3.23 ERA—good but unremarkable numbers—but finished with a  won-lost log of 25-7.    In that era, the Cy Young voting depended very heavily on won-lost records.   It still does, I suppose, but not as much.    Mike Norris pitched 284 innings, had more strikeouts (180), fewer walks (83) and a better ERA (2.53), but finished “only” 22-9.   Stone’s advantage in the won-lost record outweighed, in the minds of the voters, Norris’ advantages in innings pitched, strikeouts, walks and ERA.    Goose Gossage, closing games for the Yankees, pitched 64 games, 99 innings, had a 6-2 record, 33 saves, 103 strikeouts and 37 walks.  

            The first thing we have to do is to decide how many Game Shares to hold each pitcher responsible for.   This part of the system was explained last week, and I won’t repeat it here; it works out to 19.53 Game Shares for Goose, 28.36 for Stone, and 32.04 for Norris. 

            The next thing we have to find is the “run context rate” for each pitcher.   All three pitchers worked in pitcher’s parks, with an Applied Park Factor of .978 for Gossage (Yankee Stadium), .991 for Stone (Baltimore’s Memorial Stadium) and .930 for Norris (the Oakland Coliseum). 

            To find the context rate we take

            The league runs allowed

            Plus the league earned runs allowed

            Divided by

            18 times the league innings pitched,

            Times the applied Park Factor.

            What we’re doing here is, we’re counting earned runs twice and un-earned runs once and dividing by 18, thus reaching a figure that is half-way between the league ERA and the league runs allowed per nine innings rate, including un-earned runs. 

            I don’t believe it is reasonable to argue that a pitcher bears no responsibility for un-earned runs.   Let’s say there is a walk, and error, and a single, leading to a run scored.   It’s an un-earned run—but is it really true that the pitcher bears no responsibility for this run being scored?   Of course it is not.   The run wouldn’t have scored without the error, true—but it also would not have scored without the walk and the single.  

            In his third game in the majors (April 19, 1984), Dwight Gooden had an inning in which he gave up three hits and two walks and hit a batter with a pitch, and four runs scored.   However, because of one error, all of the runs were un-earned.    Gooden put six runners on base; the error put one runner on base.    Who is most responsible for those four runs?  

            Yet in the official stats, because of the one error, Gooden entirely escapes responsibility for the four runs.  I don’t see that that is reasonable.   I hold the pitcher to be one-half responsible for un-earned runs.   

            We figure the league runs allowed rate, holding the pitcher one-half responsible for un-earned runs, and then we park-adjust it.   This creates a context rate of 4.17 for Gossage, 4.23 for Stone, and 3.97 for Norris.  

            The next thing we need is the pitcher’s runs allowed rate.   Actually, we figure two runs allowed rates (RAR-1 and RAR-2), and then combine them.   RAR-1 is just the pitcher’s runs allowed rate, with un-earned runs counted at half.   Gossage allowed four un-earned runs, which moves his runs allowed rate from 2.27—his ERA—to 2.45.   Stone allowed 13 un-earned tallies, which moves him from 3.23 to 3.46.  Norris allowed eight un-earned runs, which moves him from 2.53 to 2.66.

            The other runs allowed rate (RAR-2) is based on the pitcher’s wins, losses and saves; it’s the won-lost record, essentially, stated as a runs allowed rate. 

            I can hear a few of you groaning out there, so let’s take a minute to deal with that.  Your groans are based on the fact that there is no clear predictive significance to won-lost records.   Pitchers “win” games and “lose” them because they allow runs, and because their teams score runs.    The won-lost record is not an independent measure of a pitcher’s ability.

            I was the first person to make that argument, I believe, and I still believe it in the main—but let’s be reasonable.   Take the extreme case.   Let’s take Elroy Face, 1959, and Luis Ayala, 2004.   Their innings pitched are almost the same—93 for Face, 90 for Ayala.  Their ERAs are almost the same (2.71 for Face, 2.69 for Ayala.)  Their park-adjusted run context rate is almost the same (4.26 for Face, 4.24 for Ayala.)   Even their strikeouts and walks are about the same (69-25 for Face, 63-15 for Ayala).  If you evaluate them without regard to their won-lost records, they’re going to come out the same—yet Face had a historic season, finishing 18-1, while Ayala finished 6-12.   Face had the highest winning percentage of all time, and finished in the top ten in the National League MVP voting.   Ayala, I am fairly confident without looking, was not mentioned in the 2004 NL MVP voting.  

            It is not reasonable to treat these two seasons exactly the same.   When one pitcher has a better won-lost record than his teammate despite the same ERA or a higher ERA, there could be many reasons for it.   It could be—generally it is—simply that his team has scored more runs for him.   But it could also be that the pitcher with the better record has pitched better in close games, has pitched better at times when he had a chance to win the game.   It could be that the team plays in an extreme hitter’s park, for example, and that the pitcher with the better won-lost record/higher ERA has pitched more innings at home.    It could be that the team plays in a pitcher’s park, and that he has pitched more innings on the road.   Such things do happen.

            More runs are scored when the temperature is higher, as I’m sure you all know.  

It could be that, by the luck of the draw, the pitcher with the better record has happened to work on nine days when the temperature was 95+, no games in the sixties, whereas the pitcher with the better ERA has worked in seven games with the temperature in the sixties, three games in the 90s.   The pitcher who has had to work in the heat will have a higher ERA, but he will also benefit from more runs scored—so he IS, in a sense, “responsible” for the extra runs scored.    He’s responsible for them because he has to deal with the conditions that created them.  Such things are very possible, and they happen, and they have an impact on won-lost records and ERA.

            It could be that they’re both Cubs, and one pitcher pitches seven games with the wind blowing out at Wrigley, the other pitches two.  The predictive value of such effects is extremely difficult to isolate due to the many different factors that go into pitcher’s records and the unstable nature of won-lost records.   It’s like trying to take a person’s height and weight while he’s on a carnival tilt-a-whirl.   The fact that such influences on a pitcher’s record are difficult to isolate does not mean that they do not exist.  

            In general, we make a better estimate of a player’s performance when we look at more factors of his record.   I’m not going to get carried away with the importance of won-lost records; I’m not going to move Steve Stone ahead of Mike Norris based on his 25-7 won-lost record.   But I’m not going to entirely ignore it, either.   I’ll give it the weight that I think that it deserves.

            OK, here’s how we figure RAR-2, which is the runs allowed rate, based on wins, losses and saves.

            Take the pitcher’s wins,

            Add one,

            Divide the pitcher’s Saves by 15,

            Add that,

            Divide by his losses, plus one,

            Take the square root of that,

            And divide the run context rate by the result. 

            We add one to the wins and losses so that a pitcher who is 4-0 doesn’t come out with a runs allowed rate of zero, and a pitcher who is 0-4 doesn’t come out with a runs allowed rate of infinity.

            We count each 15 saves as a win because closers very often have won-lost records like 3-4, 4-5, 2-4, etc.   If you ignore the fact that they’re closers, it tends to drag their effective runs allowed rates unrealistically high.  

            For Goose Gossage, 1980, we take his wins (6),

            Add one (7)

            Divide his saves by 15 (33/15 = 2.2),

            Add that (9.2),

            Divide by losses plus one (9.2 divided by 3 = 3.07),

            Take the square root of that (square root of 3.07 is 1.75),

            And divide the run context rate by the result (4.17 divided by 1.75 = 2.38).

            Essentially, we’re using the Pythagorean formula to triangulate the pitcher’s position based on the run context rate, his wins, and his losses.  Gossage’s “Run Context Rate”, based on his wins, losses and saves, is 2.38—almost the same as his RAR-1, which was 2.45.   Norris’ rate also is almost the same—RAR-1 was 2.66; RAR-2 is 2.62.  They’re usually about the same; most pitchers wind up with about the won-lost record they deserve, based on their ERA. Stone, however, because he had a better won-lost record than ERA, has an RAR-2 (a runs allowed rate, based on his won-lost record) of 2.35:

            Take his wins (25)

            Add one (26),

            Divide his saves by 15 (no saves. . .no effect),

            Add that (still 26)

            Divide by losses plus one (26 divided by 8 = 3.25),

            Take the square root of that (1.803),

            Divide the run context rate by the result (4.23 / 1.803 = 2.35). 

            We combine these, then, by multiplying the RAR-1 by .80, and the RAR-2 by .20, basing the pitcher’s effectiveness rate 80% on his runs allowed and ERA, and 20% based on his wins, losses and saves.   For the pitchers we have mentioned here, this is:

            To state this as a winning percentage, we then compare the Runs Allowed Rate and the context rate, using a similar formula to that that used for hitters, but reversing the poles.   Referring to the Runs Allowed Rate as “A” and the Run Context Rate as “B”, the pitcher’s winning percentage is:

            (1.6 * B – A) / 1.2 * B

            If B=A (that is, if the pitcher’s runs allowed rate is the same as the context rate) this works out to 0.6 divided by 1.2, or .500.   If the pitcher’s runs allowed rate is higher than the context rate, this works out to a figure lower than .500.   For these pitchers, this works out to

            We then apply the winning percentage to the Game Shares, which were given earlier in the article:

            Gossage and  Elroy Face are almost equal in terms of innings pitched, runs allowed rate, and winning percentage.   Gossage winds up at 17-3 and Face at 11-2, however, because

            a)  Gossage’s Saves are an indication that he was pitching leveraged innings, and

            b)  Gossage had many more strikeouts and walks, meaning that he was exercising more control over the innings that he pitched.

            Some of you who are really sharp will point out that Face’s 19 decisions in 93 innings pitched are also an indication that he was pitching leveraged innings.   True enough.    Perhaps we’ll fix that in the next generation.  We worry more about the usual event than the rare event.   Face didn’t have 19 decisions every year; that was a fluke.  He normally went 5-2, 6-6, 7-5.   Gossage always had Saves.  

            The 1980 Cy Young vote went 1. Stone, 2. Norris, 3. Gossage.  By our math it should have gone Norris-Stone-Gossage or Norris-Gossage-Stone, depending on what replacement level you use (.30 replacement level, Stone edges Gossage; .40 replacement level, Gossage edges Stone.)    Mike Norris was obviously the best pitcher in the American League in 1980.

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

4.  Minnie Minoso Vs. Andre Dawson

            And, for our player head-to-head comparison this week, let’s do Minnie Minoso vs. Andre Dawson.   Both players were outfielders.   Neither player is in the Hall of Fame, but either player could be.   Both players have strong ties to Chicago, having played at least some of their best years there, and both players were very popular players, for different reasons.   Dawson was popular because he was a class act, dignified and respectful of the game; Minoso was popular because he was colorful and exciting.   

In one respect they are opposites.   Andre Dawson came to the majors at the age of 21, had some fine years early in his career, and appeared, through much of his career, to be on a Hall of Fame glide path.   At ages 22 and 23, 1977-1978, he was probably regarded as the best young outfielder in baseball:

            Minnie Minoso has different reported ages, and I don’t claim to know when he was born.   I think the most commonly listed birth date for him is November, 1922, but there may be better evidence for November, 1924.   I’m going to treat his birth date as November, 1924, which means that he debuted at age 24 in late 1949, but missed the majors in 1950.   Dawson, on the other hand, hit .308 with 34 stolen bases in 1980.

            Dawson in 1980 was equal in value to Mike Norris the same year; both players come in at 25-7.   They were among the biggest stars in the game.  Through the age of 25, Dawson had a career won-lost record of 86-51.   Minoso at 25—perhaps 27—was 0 and 1.   He was far, far behind. 

            Minoso was born in Cuba with very dark skin, and was clearly and absolutely covered by the color line.   In the past, I have argued that he lost his best years due to segregation.    There are people who argue that I am wrong about this, and I will give you their argument as best I understand it.  

            1)  Minoso was not born in 1922, as listed, but 1924 (as assumed here).  Thus, he was in the majors when he was 24 years old, not an unusual age to begin a career.

            2)  Minoso’s late start, to the extent that he did start late, was caused not by the color line—which was broken in 1947—but by Minoso’s failure to hit in 1949. 

            I don’t see it that way.   First, even if we assume that Minoso was born in 1924, his career start is still very late.   Duke Snider was born in 1926; he was in the majors in 1947, and a regular in 1949.    Compared to guys like Snider, Dawson and Yogi Berra (born 1925, a regular in 1948), Minoso was either four years or six years late in getting his career started, depending on whether he was born in 1922 or 1924.   In any case, he was hundreds of games behind schedule.

            And second, of course his race had everything to do with that.   The color line did not shatter suddenly and disappear in 1947.   It eroded slowly over a period of many years.   Minoso got a brief shot with the Indians in 1949, when Bill Veeck owned the team.   When he reported to spring training in 1950, Bill Veeck no longer owned the team.   

            One of my favorite movies is The Verdict, which stars Paul Newman as a down-and-out, alcoholic lawyer doing battle on behalf of a woman put into a coma by a doctor’s mistake.    At one point Newman hires an expert to testify on his behalf, a doctor, not knowing that the doctor is an elderly black man.   The expert is patronized by the court, and appears to have little impact on the jury.

            When I watched the movie many years after it was made with my daughter, the movie held up extremely well, but my daughter couldn’t relate to the stuff about the black doctor.   “So he’s black, so what?” she asked.   “This was after the Civil Rights Act, wasn’t it?   What difference does it make that he’s black?”   I had to try to explain—racism didn’t dissipate like the morning mist as soon as the Civil Rights Act was passed. 

            The same here; there wasn’t a hard and fast color line after Jackie Robinson, but there was still a color line.    If Minnie Minoso was as good as a white player, the white player was going to play.   Minoso only got to play when he had clearly established that he was better than the white guys.  Minoso’s career started four years or six years late because he was black.  

            Anyway, he finally got started in 1951, when he was (let’s say) 26, went 6-for-14 for Cleveland early in the season, and was traded to the White Sox.  Minoso hit .326 in what was really his rookie season, scored 112 runs and stole 31 bases:

            Dawson, on the other hand, lost probably his best major league season to the 1981 strike.   Dawson hit .302 that year, with 24 homers and 26 stolen bases, a career-high .918 OPS.    We have his record at 20-2.   It probably should have been 30-3—his best year.  

            Minoso got to play in 1951 because the White Sox needed a third baseman, and Minoso could play a little bit of third base.   He wasn’t really a third baseman, though.  He hit .320-something with walks, speed and power, so obviously he could play, but he went to the outfield in 1952.   For two years after that, ages 27 and 28, Minoso and Dawson were past the things that had held them back, and both had fine seasons:

            Dawson at this point of his career was 154-73, apparently en route to the Hall of Fame.   Minoso was 68-26—a better percentage, .720 to .679, but far behind in bulk.

            In 1984, however, Dawson began to have trouble with his knees.   He was playing in Montreal, a turf park, and the surface at Olympic Stadium was like concrete.   Dawson’s knees betrayed him, and his performance began to decline:

            Minoso, outplaying Dawson every year, was catching up.   

            Minoso was a magnificent player.   In 1954, when he hit .320 with 116 RBI, he also hit 18 triples, stole 18 bases, and scored 119 runs.   He drew 77 walks, and struck out only 46 times.    He was a Gold Glove outfielder.   The Gold Gloves weren’t introduced until 1957, when Minoso was well past 30, but Minoso won three Gold Gloves (1957-1959-1960).   What he did well was “everything”.   He hit for high averages, walked, didn’t strike out, hit for power, ran extremely well, and played outstanding defense in the outfield.   He was good every year.  Beginning in 1951 he rang up ten straight outstanding seasons, 1955 being the weakest of them.  

            He also did it with style.  He was flashy, and he was aggressive on the base paths.  He played hard.    He was a great player.  Dawson had more power and equal speed, as a young player, but Dawson didn’t command the strike zone the way Minoso did.  

            In the winter of 1986-1987 baseball went through a difficult labor situation, as a part of which major league teams agreed not to sign one another’s free agents.   It was called “collusion”.   Collusion is illegal; it is prohibited by labor law.   Eventually a court ruled against them, and Major League Baseball was ordered to pay some very large amount of money to the players in compensation for their illegal actions. 

            In the middle of this collusion, the only free agent who got signed was Andre Dawson, and the way he got signed was this.   Dawson went to the Cubs and said (I am paraphrasing) “I want to play for the Cubs.   I want to play for the Cubs because I like Wrigley Field, I like day baseball, and I really need to get off the artificial turf, which is killing my knees.   Just pay me whatever you want to pay me; I’ll play for you.”    He actually took a standard player contract, with no numbers in it, signed it, and handed it to the Cubs.   You fill in the dollar amount; I’ll play for whatever you choose to pay me. 

            It was a dramatic gesture, and the teams were trying to pretend that there was no collusion going on, so they couldn’t exactly refuse to sign Andre Dawson, since they could not pretend that his contract demands were unreasonable.   Dawson signed with the Cubs, where, in 1987, he won the MVP Award:

            Dawson won the MVP Award because, following his dramatic gesture of the previous winter, he led the league in Home Runs and RBI.   But he walked 32 times on the season, giving him a .328 on base percentage.    His on-base percentage was the league average, playing for a last-place team in a hitter’s park, and he couldn’t really run anymore.   I saw him as a pretty good player, no more than that.   Given a 20-person MVP ballot, I would not have listed Andre Dawson on the ballot.  In my mind, it was one of the worst MVP selections of all time.

            I didn’t mean this to be disrespectful of Dawson; Dawson was a fine man and a fine player.   But I still remember Harry Carey, Cubs broadcaster; I thought the world of Harry Carey, and he was always nice to me.  I still remember watching the Cubs one afternoon early in 1988, and hearing Harry say “I still can’t figure out why Bill James doesn’t think that Andre Dawson was deserving of the NL MVP Award.   I mean, look at this man. . .” and he ran threw Dawson’s virtues:  his league-leading 49 homers, his grace on and off the field, his dramatic action in coming to the Cubs at a price to be determined by the team.   I don’t deny any of that.   You score runs by getting people on base, and the Cubs finished last.

            At age 33 Dawson and Minoso had nearly identical seasons, apart from the fact that, of course, Minoso walked and Dawson didn’t:

            Dawson was now 240-140 in his career, a .632 percentage; Minoso was 185-66, .738.   Both players had two very good years left, ages 34 and 35:

            OK, Dawson wasn’t that good in 1989.   I was just being nice.   At age 36 Dawson and Minoso both went into decline, but both stayed healthy and stayed in the lineup:

            At age 37 Dawson had one more good year.  Minoso, traded to the Cardinals, got off to a slow start, and then got hurt in May, 1962:

            After that they were both just playing out the string, including a couple of famous late-in-life publicity stunt re-appearances by Minnie Minoso, then coaching for the White Sox:

            On the bottom line, I have Dawson at 319-226, a .585 percentage, and Minoso at 252-119, or .679.   Both players, by my account, are at least minimally qualified to be considered Hall of Fame candidates, Dawson because he had 300 Win Shares, and Minoso because he was 100 wins over .500.   

            As an aside, I seem to have lost somehow the “Right Field adjustments” that I put in a few weeks ago, concluding that right fielders were inadequately compensated in the defensive analysis.    Breaking down Minoso and Dawson into offense and defense:

            Dawson in the field was 36-20 through 1983, when his knees were good, but 24-44 the rest of his career, after his knees went.   That’s fairly typical.   Minoso in the field was 34-22 through 1957, the year of the first Gold Glove Awards, but 14-21 after that year.   Everybody’s defense goes to hell in their thirties.

            Anyway, I lost those right field adjustments somehow, and I’ll have to put those back in.  That will help Dawson in the field, probably enough to get him back to .500 overall, 62-62.   That will also get him very close to being over both of the Hall of Fame standards, 300 Wins and +100 Win Shares, which would mean that he was a clear-cut Hall of Famer by my standards.

            I think they’re both Hall of Famers, to be honest; Dawson was a very good player for a long time, and Minoso was a significantly better player than Dawson once he got a chance to play. 

            I stay out of Hall of Fame debates when I can, but I advocate for Minoso because he was a childhood favorite, and because I believe that, given a chance to play at age 22 like Dawson or most other players of that caliber, he would have cleared the Hall of Fame bar with ease.