In posts regarding recent articles we have had comments/queries about Sandy Koufax and Wes Ferrell, and I thought that I would respond to them here.   

              Regarding Koufax, Koufax was greater than the 21^st century analysts understand.   21^st century analysts are sometimes hung up on WAR—Wins Above Replacement—but you would get a different picture if you could look at PCAR—Pennant Contributions above Replacement.  

              First of all, we have to assert there that Pennant Contributions can be different from Win Contributions.   I doubt that any of you would question this, but suppose that you have two players of equal value, both on the same team, and suppose that that team goes 74-88, 95-67, 84-78 and 94-68 over a four-year stretch.  Suppose that one player plays well in years 1 and 3, when the team has no chance of winning the pennant anyway, but plays poorly in years 2 and 4, when the team has a chance to win.   The other player. . .you get the point.   The win contributions of the two players are even, their opportunities are the same, but their Pennant Contributions are different.

              Second, I would argue that, to the extent that Pennant Contributions may be discerned, and to the extent that Pennant Contributions may be seen relative to opportunities, that Pennant Contributions are actually MORE important than Win Contributions.  Win Contributions are important because they are Pennant Contributions.   Singles are important because they are parts of runs.   Runs are important because they are parts of wins.   Wins are important because they are parts of pennants.   We don’t want to discriminate against Ernie Banks because he didn’t have the opportunity to win his team the pennant, but winning the pennant, in the 20^th century, was the main purpose of the season.  

              In the 21^st century it is different; now, if you don’t win the pennant you can still win the Wild Card, it’s not that big a difference.   If you do well in the post-season nobody will remember whether you won the division by a game or lost it by a game.    In the 20^th century, if you finished one game short, it was a lost season—if not a disastrous season.  The 1964 Cincinnati Reds finished one game short; very few people remember that now.    The 1964 Phillies also finished one game short, and that came to be regarded as an epic disaster.     In the 20^th century, you won the pennant or you didn’t.

              That BEGAN to change in 1969, completely changed toward the end of the 20^th century.   Anyway, in Koufax’ time winning the pennant was the central challenge of the season.

              Koufax’ contributions toward winning the pennant are, I would argue, greater than any other pitcher of the 20^th century.   There are three seasons—1963, 1965, and 1966—in which if Koufax had been JUST great, the Dodgers would not have won the pennant.   In 1966 he was 27-9.   If he had been 25-11 that would still have been a great season, but the Dodgers would not have won the pennant.   In 1965 he was 26-8.   If he had been 24-10 that would still have been a great season, but the Dodgers would not have won the pennant.  

              1963 is not as easy to explain, since the Dodgers won by six games—but the Cardinals were one game behind with ten to play.    It just opened up in the last week.   Koufax was 25-5.  If he had gone 19-11, assuming that one of the losses would have been to the Cardinals, the Dodgers would not have won the pennant.    Realistically, if he had gone 21-9, the Dodgers might well have wound up on the short end.   Koufax was 4-0 with a 0.96 ERA against the Cardinals.   He started five times against the Cardinals.   The only game he didn’t beat them he pitched 12 innings, left with the score 1-1, and the Dodgers won the game.  

              There are three seasons in there in which, if Koufax had been JUST the best pitcher in the league, but not the best pitcher in the league by a wide margin, the Dodgers would not have won the pennant.   There is no other pitcher in the 20^th century who did that.   I remember I studied this ten, fifteen years ago. . .not sharp on the details, but I think the only other pitcher in the 20^th century who had two of those monster-impact seasons for a team that would not have won without that kind of season was Carl Hubbell, in ’33 and ’36.   I believe that Kid Nichols had some high-impact seasons in the 1890s, and Randy Johnson did at the turn of the century.   (Among position players, as I recall, the highest-impact seasons were by Babe Ruth and Roberto Alomar.)  

              Also, this exceptional pennant impact of Koufax is, to some extent, a natural consequence of the shape and pattern of his career.   To some extent it is a matter of being in the right place at the right time, of course, but to some extent not.  

              Suppose that you have two pitchers, both of them .625 pitchers.     One pitcher, however, is a .625 pitcher every year, or, more realistically, has a bunch of years in the .600-.650 range.   The other pitcher is a .500 pitcher half of his career, a .750 pitcher the other half.   Which pitcher is likely to have more impact on pennant races?

              Suppose that you have two pitchers who have 300 starts/250 decisions each.    One pitcher, however, makes 30, 33 starts a season, 20-25 decisions a year.   The other pitcher makes the same number of starts, but 40-42 starts a season, 30-35 decisions a season.   Which pitcher is likely to have more impact on pennant races?

              There is an old saying, which was popular as anti-sabermetric bigotry back in the day when we were fighting those battles.   The saying is that a statistician is a person who will say that if you have one foot in a fire and the other in a block of ice, on the average you’re comfortable.   This is a case in point.   Two pitchers may look the same in their aggregate totals, but they may be very different in their actual impact.   Koufax had a fire-and-ice career.

              I tried to demonstrate this point with a simulation, although my simulation tends to show that my point is not all that significant, but I’ll share it with you anyway.   I set up a simulation in which I created teams with realistic expected winning percentages.   A team could have an expected winning percentage anywhere between .300 and .700, but most teams clustered around .500.   Then I generated 162 random numbers for each team.   If the random number was less than the team’s expected winning percentage, that would be a win; if the random number was higher, a loss.  

              Also, a team might go 94-68 and win the pennant, or they might go 100-62 and not win the pennant; that also varies.   A team had a 10% chance of winning the pennant if they won 94 games, 20% if they won 95, 30% if they won 96, 40% if they won 97, 50% if they 98, jumps up to 70% if they won 99, 80% if they won 100, 90% if they won 101, 100% if they won 102.   

              With those conditions, the team won the pennant 1,012 times in 10,000 trials.   I was aiming for 10%; that was as close as I could come.   Koufax’ great years were in a 10-team league, so I was simulating that.  

              Into that base, I inserted four pitchers with .625 winning percentages.  

              Pitcher A was responsible for 24 decisions a season, and had a .625 expected winning percentage every season.

              Pitcher B was responsible for 24 decisions a season, but had a .500 expected winning percentage in half of the seasons, .750 in the other half.

              Pitcher C was responsible for 33 decisions a season, and had a .625 expected winning percentage every season.

              Pitcher K, the Koufax-model pitcher, was responsible for 33 decisions a season, had a .500 winning percentage in half of the seasons, but a .750 winning percentage the other half.  

              Pitcher A, the least like Koufax, increased the number of pennants won by his team from 1,012 in 10,000 seasons to 1,240, an increase of 228.    Pitcher K, Koufax, increased the number of pennants won by his team from 1,012 to 1,355, an increase of 343.  

              But when you look at it, you realize that that’s not actually all that big a deal.   The issue here is the PROPORTIONAL impact of his starts, per start, not the gross impact of 33 games as opposed to 24 games.   Pitcher A is responsible for 24 games per season and increases his team’s pennants by 228 in 10,000 years, or 9.4 pennants/10,000 years for each start.   Pitcher K is responsible for 33 games per season and increases his team’s pennants by 343, or 10.4.    The Koufax-model pitcher IS having more impact on the pennant race, per start, than a pitcher with the same career stats, but only 10% more.  

              OK, Wes Ferrell.

Ned Garver

              Changeup.   Regarding Ned Garver, who some people also asked about, I have him with deserved vs. actual won-lost records as follows:

              Garver had career luck score of negative 42.5, so he was obviously very unlucky; I think he just didn’t make the cut when I listed the unluckiest ever.  While we’re talking about Garver and Garvin we might throw in Jerry Garvin, who was also very unlucky in a short career.  

Wes Ferrell

            In my series of articles a couple of weeks ago, David Kaiser posted repeatedly questioning why Wes Ferrell was not on the list.   From two different posts:

The huge omission in Bill's list is my choice for the unluckiest pitcher of all time, Wes Ferrell. He had 5 seasons of 4 WAA or more, which ties with Newhouser for the best of his generation (Feller had 4.) But he pitched for lousy teams.

There is however a big omission from this list: Wes Ferrell. He is an exception to one of the rules Bill laid down at the beginning of this article--he did pitch his whole career with bad teams. He is tied with Hal Newhouser, with 5 seasons of 4 WAA or more, within the GI generation (born 1903-24.) Feller had only 4 such seasons, as did Hubbell. He is very overqualified for the Hall of Fame.

              Ferrell had a career won-lost record of 193-128, which is a borderline Hall of Fame won-lost record.  I have him as having a deserved Won-Lost record of 169-134, so not only do I not agree that he was terribly unlucky, it would be my judgment that he was actually lucky.    He wasn’t really as good as his won-lost record.

              Let us assume, for the sake of argument, that my analysis of Ferrell is flawed in that it fails to account for his quite exceptional excellence as a hitter.   I would not only be willing to assume that; I would assert that this was probably 80% true.   Much or most of the discrepancy between Ferrell’s actual and deserved won-lost records WAS, in fact, accounted for by his excellence as a hitter. 

              But as to saying that Ferrell was an UN-lucky pitcher. . . . I don’t get it.  I don’t understand what the hell David is talking about, frankly.    Ferrell did NOT pitch for lousy teams, nor did he pitch his whole career with bad teams, nor was he unlucky in any respect.  

              Ferrell pitched for only three bad teams in his career—the 1927 Cleveland Indians, the 1928 Indians, and the 1941 Braves.   He pitched only one inning in 1927, only 16 innings in 1928, and only 14 innings in 1941, so those teams account for only 1% of his career.   And they weren’t all THAT bad; the Indians were 66-87 and 62-92 and the Braves were 62-92.   Those are records that the Kansas City A’s would have considered "almost in contention."

              Other than those three, these are the won-lost records of the teams that Ferrell pitched for:

              I really don’t understand how anyone could interpret that as pitching for lousy teams or pitching his whole career for bad teams.    The career winning percentage of Ferrell’s teams, weighted by the number of innings that he pitched for each team, is .514.    If you take out Ferrell’s decisions, Ferrell’s wins and losses, that figure drops to .489, but that still doesn’t get us anywhere near "lousy" or "all bad".  

              Perhaps Mr. Kaiser would like to make the argument that these teams stayed near .500 because of good pitching but had poor offenses, but that would not be true, either.   The offenses of Ferrell’s teams, park adjusted, weighted by the number of innings that he pitched in each season, were 4.7% below the league average.   They were a little below average.   Defense?   1930s fielding by team is difficult to evaluate, but two of Ferrell’s teams led the league in double plays, one led in errors.   I’m not really seeing the problem.  

              I think that Mr. Kaiser and I have similar views of several issues.   In particular, he has advocated (I think) the opinion that too much attention is paid to career standards in evaluating a pitcher as a Hall of Fame candidate, and not enough attention to how good the pitcher was in his core seasons.   I would agree with that.   Also, I think Wes Ferrell DOES have at least a reasonable Hall of Fame argument, based on that argument.   But as to the argument that Ferrell was unlucky to have a career won-lost record of 193-128. . . I don’t believe it, I don’t 70% believe it, I don’t 1% believe it.   I think that argument is 100% false.