29.  Decision Theory

Hey Bill, re: the question below about 20-game winners past and present. In RP, there is a thread (Twenty Games Winners, Present vs. 70s) projecting out Ws had the 19 and 18 game winners of the past five years with full season starts got the GS that the league leaders in the '70s got. My question is, see, there is one, has anyone ever studied what the W/L/ND impact is of going 7/8/9, respectively? Obviously, sometimes Mariano saves your win, but sometimes you leave in the 7th tied and your team picks up a win or rallies and gets you out of the loss. Could simply backing out reliever's decisions get you this? This is on one cup of coffee, so forgive any logic disconnects.

                Well, I think I’ve run that data before, but it’s been a couple of years and it’s not hard, so. . …let’s just start with the percentage of time a pitcher gets a DECISION (win or loss) if he pitches 5 innings, 5.1 innings, 5.2 innings, 6 innings, etc.:

                One pertinent question is whether these numbers have changed. . ..whether a pitcher who pitches 7.1 innings in a game is more likely or less likely to get a win now than he was at some point in the past.    Pitchers who get no one out in the game still get the decision 56% of the time, which, when you think about it, pretty sharply limits how much change can occur in the ratio of decisions to games.    In any case, I studied the data in three groups:

                A)  Both Leagues from 1952 to 1972,

                B)  National League, 1973 to the present,

                C)  American League, 1973 to the present (DH era). 

                Let’s get rid of the data for pitchers who pitch more than 9 innings in a game; we’re not going to learn anything more from wrestling with that.   For the rest of the data, we’ll compare the "A", "B" and "C" groups:

                The Designated Hitter Rule has substantially increased—increased—the decision percentage for starting pitchers working more than six innings but less than 9.  Why?

                It turns out that it is an apples-to-oranges comparison, much more so than I had anticipated.   I had thought that pitchers who pitched 7 1/3 innings in 2013 would be pretty much the same as pitchers who pitched 7 1/3 innings in 1960, but not so at all.    Pitchers who pitched exactly seven innings in Group A (1952-1972), had a very good ERA (3.21) but a .328 winning percentage.    They were taken out after 7 innings for a pinch hitter, which means (generally) that their teams were behind.

                With the DH Rule (Group C), pitchers who were taken out after exactly 7 innings had a comparable ERA (2.81), but a dramatically better winning percentage (.681), because they were not being removed for pinch hitters.   The National League since 1973 (Group B) is more like the American League than it is like the National League 1952-1972; pitchers working exactly 7.0 innings have a 2.51 ERA and a .610 Winning Percentage.    A .610 winning percentage is still low for a 2.51 ERA, so some of those pitchers are still being removed because the team is behind a run and needs a pinch hitter, but most pitchers removed after 7 innings now—even those removed for a pinch hitter—are no longer being removed because the team is behind and needs runs.   They’re being removed because they’ve thrown 100 pitches.

                The data for pitchers removed after 6 2/3 innings is less striking than the data after 7.0, because the pinch hitter issue is not part of the equation, but that data, too, changes over time.    In 1952-1972, NL pitchers removed after 6 2/3 innings had a .495 winning percentage.   Since 1973 it is .594—essentially 100 points better.    Again, it’s apples and oranges; a pitcher removed after 6 2/3 in 1960 was being taken out because he was getting hit hard.   A pitcher removed after 6 2/3 in 2000 was being taken out because he had thrown 100 pitchers.

                So. . .the groups are just not the same; the decision percentages are pretty much the same, but that simple fact is disguising internal changes in the data which are extremely profound.   I’m out of conclusions here, but I’ll run the rest of the data. . .Winning Percentage and ERA for pitchers pitching X innings in each of the three groups.   All of the groups outlined below are based on hundreds of pitcher/games, so there shouldn’t be pure flukes in the data:

30.  Herb Score Vs. Don Newcombe, 1956

                The first Cy Young Award given went to Don Newcombe in 1956, also the National League MVP in that season.   There was one ballot per team, 16 voters total, and only one slot on each ballot, rather than the extended ballot we use now.  Four pitchers were mentioned in the voting—Don Newcombe, with 10 votes, his teammate Sal Maglie, with 4 votes, Whitey Ford, with 1 vote, and Warren Spahn, with 1 vote.

                I mentioned earlier that, since there are multiple ways to interpret the data from this study as to who deserved the Cy Young Award, I would treat the award as "correct" as long as there is any reasonable interpretation which would support it.   If the award winner leads the league in any of our categories, generally speaking, it’s a reasonable selection.    In this case, however, there is a clean sweep.   Herb Score leads in every category that can be derived from this data.    Score leads Newcombe 238 to 223 in the raw total, despite Newcombe having three more starts, which essentially gives Newcombe a 15-point head start.  He leads him 7.21 to 6.19 in the average per start, leads him 73 to 43 in margin over 5.00, leads him 139 to 115 in margin over replacement (3.00), and destroys him the Good Game Record, Score clocking in at 25-6, Newcombe at 20-12.  Newcombe is not in the top five in most of these categories.

                Newcombe won the Cy Young Award, of course, because he "Won" 27 games, "Lost" 7; not entirely because of that, perhaps.    He was also in New York, whereas Score was in Cleveland, and he also played for a pennant-winning team, whereas Score’s team finished second.  In any case, we are not here to debate why Newcombe won, but whether the vote might be justified by some other, equally logical approach to the problem.

                It seems to me that there are two arguments that could be offered on behalf of Newcombe’s selection.   Herb Score was a wild, hard-throwing kid who led the majors in strikeouts that year by 71 (263 to 192), but who also walked 129 hitters, second-most in the majors. Score receives an unfair advantage in the system (some would argue) because the Game Score system rewards pitchers for strikeouts, as opposed to ordinary outs.

                Second, Don Newcombe was an outstanding hitter, quite possibly the greatest hitting pitcher in the history of baseball.  To focus on his run support, without remembering Newcombe’s own contributions to his team’s offense, denies him credit for his real skills. 

                But both of those arguments, it seems to me, fail quickly.   While Newcombe was in fact a tremendous hitter, he didn’t really do much with the bat in 1956.   He was used as a pinch hitter about a dozen times, but he hit .234 on the season, 2 homers, 12 walks, a .654 OPS.   Herb Score was not a great hitter, but in 1956 he hit .184, 3 doubles, 1 homer, 9 walks, a .513 OPS.   The difference between them, as hitters, is three or four runs.    It doesn’t do very much to offset the differences between them as pitchers.

                Second, it is not true that the Game Score system gives a strikeout pitcher an unfair advantage; not only is this untrue, but the people who insist on making that argument, after all of these years, should be trampled by elephants, poisoned, water-boarded, covered with sugar-water and tied to a colony of wasps, and buried in rocks; that would be a fair and just punishment for them, just my opinion.   The Game Score system gives strikeout pitchers not an unfair advantage, but an entirely fair and rational recognition of their contribution.   When assessing a player’s value, it is not arbitrary or irrational to give the player credit for those things he does which have value.  The Game Score doesn’t over-value strikeouts; it actually under-values them.

                But let us set that argument aside, and suppose for the sake of consensus that the argument was correct, that Score should not have received credit for his strikeouts.    Score struck out 124 hitters more than Newcombe did.  It takes essentially 10 points in the Game Score system to move a pitcher one level in a game.   The 124 strikeouts would only move Score, over the season, by 12 levels. . .let’s call it 13.

                But Score’s advantage over Newcombe was more than 30 levels.   Score’s edge in strikeouts would account for less than half of the difference.

                But the Game Score system has a point not only for a strikeout, but also a negative point for a walk.   This "adjustment" assumes that we are taking away the points Score would get from his strikeouts—but leaving in the system the points that he loses for walks.   Score had 263 strikeouts, 129 walks; Newcombe had 139 strikeouts, 46 walks.   By this approach we are treating Score as if he had 139 strikeouts, 129 walks, while continuing to assume that Newcombe had the actual strikeout to walk data that he did have, 139 to 46.   

                That is obviously not fair or rational—but even so, Score remains comfortably ahead of Newcombe.    Let’s look at the real facts.   The American League ERA in 1956 was 4.16, and Score’s ERA was 2.53.  His ERA was 1.63 runs better than league.   The National League ERA was 3.77, and Newcombe’s ERA was 3.06.  He was 0.71 runs better than league.   Score was about 43 runs better than an average pitcher; Newcombe, about 21 runs.  The actual reason that Score scores as better than Newcombe is that he allowed many fewer runs.   Yes, Ebbets Field was a hitter’s park, and Memorial Stadium in Cleveland was a pitcher’s park, but Score is so far ahead that making the adjustment for that frankly does almost nothing to get Newcombe back in the game.

                Other than those two arguments, both of which seem to me to be clearly false, I don’t see how one could defend the selection of Newcombe over Score.   Newcombe was credited with 27 wins because the Dodgers scored 210 runs for him in 36 starts (5.83 per start), whereas Score was credited with "only" 20 wins because the Indians scored only 153 runs for him (4.67 per start); that is the bottom line, and that is the real truth.

31. Has There Ever Been A Cy Young Pitcher

Who Was In Reality A Below-Average Pitcher?

No.   But Pete Vuckovich was close.

32.  Has There Ever Been a 20-Game Winner

Who Was Really Below Average?

                Stan Bahnsen, 1972 (21-16) appears to be below average by any approach that I can find.  

Vida Blue and Paul Splittorff in 1973 were both very close to the "average" line, and could be above or below, depending on what method you use.

But interestingly enough, all of the other 20-game winners I can find who show as below average by other methods, seem to look substantially better by this line of analysis.   Lew Burdette, 1959, shows as well below average by some methods, but comes in right on the average line (5.000) in this approach.   Denny McLain, 1966, shows as below average by some methods, but well above average (5.84) by this approach; Warren Spahn, 1960, the same (5.67).   Jack Morris, 1992 (21-6) has a league-average ERA, but comes in at 5.54.   All of these appear to be pitchers who pitched well in the majority of their starts, but just got lit up a few times, ruining their ERAs.

33.  Back to Double D

                Not those Double Ds; I was talking about Don Drysdale in 1964.  The Dodgers outscored their opponents in Drysdale’s starts, 171 to 116—but lost 21 of the 40 games.   I was wondering, first, whether there was any parallel to that in the data.    That is the largest discrepancy between expected and actual wins in the data, but there is one other that is close. . ..just one.   With Phil Niekro on the bump in 1974 the Braves outscored their opponents 174 to 110, which should lead to a record of 28-11, but in fact they went just 20-19.    Other than that. . .nothing close. 

                This issue came back up when I was trying to figure out who should have won the Cy Young Award in 1964.   The answer suggested by a simple analysis is Don Drysdale.  Double D led the majors in most of the measurements derived from this analysis, and had a Good Game Record of 33-7.

                It thus becomes a key question:  Why did the Dodgers not win more of his starts?   It wasn’t run support; they scored 171 runs in the 40 games.    It was that they lost almost all of the close games with Drysdale on the mound.   Why?

                Two theories:

                A)  The Dodger offense for Double D was very inconsistent, scoring superfluous runs in games that were already won, so their offense wasn’t really as good as the total would suggest, or

                B)  Drysdale pitched just well enough to lose.

                Which is it?

                I set up a study.   I took the log of runs scored by the Dodgers in Drysdale’s starts, and a log of the runs scored by the opposition.   Then I made 1000 copies of each log, randomly scrambled them, calculated the wins and losses resulting from the randomly scrambled runs and runs allowed, and repeated the process several times.  

                When you outscore your opponents 171 to 116 you should win 68.5% of your games.   In the study, the "Dodgers" won only 63.0% of the games—clearly demonstrating that some of the problem is in (A) above.   The offense was inconsistent, to the extent that it was less productive than we would expect.

                But the study also shows that most of the problem was in (B).   Drysdale and his bullpen tended to give up just as many runs as they needed to to lose.    The inconsistency of the offense reduced the expected winning percentage from .685 to .630—but the inefficient matching of runs allowed to run support reduced it from .630 to .475.

                In no sense am I suggesting that this reflects poorly on Drysdale’s skill, or (God forbid) on his character.   It is simply something that happened.

                But let us suppose a player hits .350 with the bases empty, but .170 with men on base.   I wouldn’t suggest that the hitter lacked character, or that he choked in the clutch, or anything like that.   It is simply something that happened—but it happened.   He had less value than he should have had because it happened.

                Same thing here:   It happened, and Drysdale’s value is less than it appears to be because it happened.   It is not appropriate to pretend it didn’t happen.  Absent this information, I would argue that Drysdale deserved the Cy Young Award in 1964.  But because this happened, I could not make that argument.

34.  Turley Vs. Ford, 1958

                Bob Turley and Whitey Ford were teammates on the 1958 New York Yankees, who won the World Championship.  Turley won the Cy Young Award.   Among the most surprising findings of this study is that he actually deserved it.

                Let me explain first why this is surprising.   Turley was 21-7; Ford was 14-7.  They were the only two pitchers on the Yankees who won at least ten games, which is also quite surprising in that era, to have a World Championship team with only two 10-game winners. 

                But while Turley "won" more games, Ford’s ERA was almost a full run better, 2.01 to 2.98.   Turley walked 128 batters in 245 innings; Ford walked less than half as many.   Lee Sinins’ Complete Baseball Encyclopedia lists Ford as 45 runs better than league average, Turley only 24 runs better, and credits Ford with a Neutral Won-Lost record of 16-5, whereas Turley is at 17-11.

                Pete Palmer’s Pitcher Win statistic shows Ford at +3.9, Turley at +2.2.   Baseball Reference has Ford at 4.3 WAR, Turley at 3.6.   Fangraphs has Ford at 4.9 WAR, Turley at 1.8.  Turley went 21-7 because he was supported by 156 runs in 31 starts (5.03), whereas Ford won only 14 because he was supported by only 130 runs (4.48).   If Ford didn’t win the Cy Young Award, one tends to think, it could at least have gone to somebody else.

But hold on. First of all, there is a scarcity of good candidates, other than Turley and Ford.   There were three 20-game winners in the National League, Warren Spahn, Lew Burdette and Bob Friend, but two of those three, Friend and Burdette, were just about average pitchers.    Spahn was better than Burdette or Friend, but not dramatically better; Spahn’s ERA was higher than Turley’s despite pitching in the most pitcher-friendly park in baseball, with a park factor of an absurdly low 69.   The best pitcher in the National League was actually Sam Jones, 14-13 but with 225 strikeouts and a 2.88 ERA, but Jones was not better than Turley or Ford; he was merely better than Spahn and Friend.

                Back to Turley and Ford; Turley was the better pitcher in terms of Wins and Losses, which we tend to regard as a pretender statistic, but Ford was better in terms of ERA, which we in the sabermetric hovel tend to regard as the real deal.  

                But, in fact, Turley was better, and let me explain why.

                First, Turley made 31 starts and pitched 245 innings; Ford was at 29 and 219.   Turley pitched 12% more innings; that wouldn’t get him back to even, but it’s a start.

                Second, Turley allowed 1 un-earned run; Ford allowed 13.    Ford leads by .97 in earned run average, but .46 in total run average, earned and un-earned.   I don’t hold Ford fully responsible for the un-earned runs, but it is also inaccurate to entirely ignore them.   On August 13, 1958, leading Washington 4-3 after 7 innings, Ford gave up a two singles, a double and a home run in an inning, leading to five runs, but because of one error four of the five are listed as un-earned.    Oh, did I forget to mention:  it was Ford’s own error.    Ford is excused from any responsibility for four runs because of his own error.   On May 30, also against Washington, there was one error in an inning, putting a runner on base, but with two out Ford gave up a walk, a single, a walk, a single and another single—three hits and two walks, leading to four un-earned runs.   The earned run average system gives Ford a complete pass on those eight runs, because of the two errors, but I don’t; I hold him 50% responsible for his un-earned runs.

                Third, as you all know, Casey Stengel liked to "spot" his pitchers to get matchups against teams.   Traditional analysis is entirely blind to the effects of this, and the fan tends to assume that Ford was being spotted against the toughest competitors.   This method looks very specifically at who Turley was pitching against and where he was pitching.   When you do that, you can see that in fact the load being placed on Turley was significantly heavier than the load on Ford.

                The three toughest parks in the American League, for a pitcher, were in Kansas City (116 Park Factor), Boston (112) and Detroit (104).   Turley and Ford started five times each against Kansas City, so that’s not really a factor.   But Turley started seven times in Boston and Detroit.   Ford started zero times there. . .seven to zero in places that were tough for a pitcher.

                But in the three best pitcher’s parks in the American League—Baltimore, Chicago and Cleveland—Ford started 10 times, Turley 6.   Stengel was starting Ford against what he saw as the "left-handed" teams and in the left-hander’s parks, whereas he was starting Turley against the right-handed teams in the right-hander’s parks—but the collateral damage was that Turley was starting in the hitter’s parks, whereas Ford was starting in the pitcher’s parks, so the run context was not really the same.   Turley received "better" offensive support than Ford did, because Turley was pitching in Fenway Park and Tiger Stadium, whereas Ford was pitching in Baltimore, Chicago and Cleveland.

                Not done yet.   That would be offset if Ford was starting against the tough teams in the league and Turley against the easier teams, but he wasn’t.    If you weight the number of starts against each opponent by the opponent’s won-lost record, Ford’s opposition winning percentage in 1958 was .480; Turley’s was .483.   (That is, in effect, a difference of one run over the course of the season, if you’re curious.)

                Ford made 15 starts in Yankee Stadium, 14 on the road; Turley made 13 starts in Yankee Stadium, 18 on the road.   An interesting small point is that Turley, who started 38% more often on the road than he did at home, was substantially more effective at home than on the road (2.54 ERA to 3.42), whereas Ford, who had one more start at home than on the road, had a sensational 1.27 ERA on the road, 2.71 at home—and allowed almost all of those un-earned runs at home.

                So Turley pitched more innings than Ford, allowed 12 fewer un-earned runs, pitched in substantially tougher parks against slightly more difficult opposition, and pitched much more often on the road.   When you factor all of that in, it was in fact Turley who deserved the 1958 Cy Young Award. 

35.  The 29 Pitchers Who Probably Should Have Won the Cy Young Award

                Let’s be honest:  From 1956 to 1990, the Cy Young voting was terrible, and the chief culprit was over-reliance on the won-lost record.

                There have been 105 Cy Young Awards, going to 106 players—

                1 per year, 1956 to 1966 (11)

                2 per year, 1967 to 2013 (94)

                And in 1969 the American League voting ended in a tie, so we have three winners that year.

                Since 1991, the Cy Young voting has been substantially better.    By my count. .  .and my accounting here is absurdly generous, as I will explain in a moment. . .but by my count the Cy Young voters were 37-22 in picking the right pitcher through 1990, and have been 39-7 since 1991.

                First of all, as I have now said three times and will repeat several more, if an award was reasonable, I count it as being right.   I’m not here to quibble or to assert that I have the perfect answer to all of these questions; I just want some reasonable answer.   Was Randy Jones really the best pitcher in the National League in 1976?  Probably not; most indicators would say "No".   It was probably Tom Seaver—but there are some indicators that favor Jones, and that is all that we ask.   Was Steve Carlton really the best in 1977?   Arguable; could have been Carlton, could have been Seaver, could have been J. R. Richard or John Candelaria.   We accept it as accurate.  

                In 1987 Roger Clemens and Jimmy Key each made 36 starts, and each total up to 254 points, a flat-footed tie which makes them dead even in their margin over .500, or over replacement level.   Clemens had a Good Game record of 24-9; Key, of 28-6—so I would have to vote for Key over Clemens, based on what I know at this time.  But Clemens ties Key in most of the indicators, so it’s reasonable, so we accept it as valid.  

                Was David Cone really better than Clemens or Randy Johnson in 1994?   Debatable—so I accepted it as true.   Whenever I could mark the Cy Young voters down as "right", I did. 

                I scored all of the awards going to relief pitchers as "reasonable".   Yeah, I know; that’ s silly.   They’re not all reasonable.   But to compare a reliever to a starting pitcher introduces mathematical complications into what is already a research project of substantial scale, and I just didn’t want to get into that; I would rather concede them all than debate them all.

                The reality is that several of the awards to relievers were given in years when the best starting pitcher in the league had very bad offensive support, thus was stuck with a subpar won-lost record.   The award to the reliever was a kind of evasive maneuver taken to avoid awarding a deserving pitcher with a poor won-lost record.   Does anyone really believe that Steve Bedrosian was the best pitcher in the National League in 1987?   The best pitcher in the National League in 1987 may have been Nolan Ryan, but Ryan was 8-16, and they weren’t going to give the award to a pitcher who lost two-thirds of his decisions.

                Maybe it wasn’t Nolan Ryan; maybe it was Rick Reuschel, or maybe it was Tim Burke, another reliever.   It wasn’t Bedrosian.   If we throw out the awards to relievers, that would make the score 66-29, rather than 76-29, and also it would make the voters 30-22 up through 1990.  

                The 1969 American League Cy Young Award, where the voting ended in a tie. . ..I’ll count it as being right.   Probably the best pitcher in the American League in 1969 was Andy Messersmith, rather than either Denny McLain or Mike Cuellar, but I’m not here to quibble.   It’s a reasonable choice; we’ll count it as being the right choice.

                There are the curious votes of the National League in 1969 to 1971.    The Cy Young Awards for those years went to Tom Seaver, Bob Gibson and Ferguson Jenkins, and they should have gone to Tom Seaver, Bob Gibson and Ferguson Jenkins, but not in that order; as best I can calculate this, Gibson should have won in 1969, Jenkins in 1970 and Seaver in 1971, making it Gibson-Jenkins-Seaver rather than Seaver-Gibson-Jenkins.

                But really, is that a big deal, or is that a quibble?    I think that Gibson was better than Seaver in 1969, but he wasn’t much better.   I think that Ferguson Jenkins was the best pitcher in the league in 1970 (Seaver second, Gibson third), but Jenkins wasn’t much better than either of the others.  I think Seaver was better than Jenkins in 1971, but it’s close, and actually I’m not sure; Jenkins is ahead on some of the markers I laid out before, and also Jenkins hit .243 with 6 homers, 20 RBI in 1971; if you give him credit for that I suspect he is clearly ahead.  

                It’s a quibble; it doesn’t really matter in which order these three pitchers won their awards.   I decided to count them all as "right", even though my methods show that they’re all three wrong.   Take those away and the tie in ‘69, and the voters up to 1990 would score at 26-26.

                By the way, that’s an archetype of a kind of common problem; very often the right pitchers win the award, but in the wrong seasons.  But there are a couple of other problems.   Early Wynn was clearly not the best starting pitcher in baseball in 1959, but. ..who was?    I don’t know.    Based on the sum value of all of his starts, it would be Johnny Antonelli of the San Francisco Giants, with 229 points.  Based on the average value of a start, it would be Hoyt Wilhelm of Baltimore.  If you look at the margin above .500 or the margin above .300 (replacement level), the best pitcher was Camilo Pascual of Washington.   If you look at the Good Games record, it was probably Vern Law of Pittsburgh (22-7, .759), although his teammate Harvey Haddix actually had a better percentage (18-5, .783), and if you prefer the simple total of Good Games pitched, that would be Larry Jackson of St. Louis, with 23.     And Don Drysdale is in the mix, and Don Newcombe and Warren Spahn and Jim Bunning and Sam Jones.    The reality is that I don’t know who was the best pitcher of 1959, having studied it as carefully as I can study it.   I honestly believe that almost every pitcher I have mentioned here would have been a better pick that Early Wynn, but if I can’t say who did deserve to win the award, I don’t see how I can charge the voters with getting it wrong. 

                1964 is like 1959; I really don’t think that Dean Chance deserved the award, but I can’t say with confidence who did; it could have been Joel Horlen, Gary Peters, Koufax, Marichal, Drysdale, Bunning or Larry Jackson.   I scored those two as "right" because if I had scored them as wrong I would have had to tell you who should have won the award, but if I had scored them the other way the total up to 1990 could go under .500, at 24-28.    Still giving the voters credit for all of the questionable selections that could be argued to be right, like Randy Jones in ’76. 

                Let’s be honest:  Up to 1990, the Cy Young voters only got the answer right when it was obvious.   If it wasn’t obvious, they usually missed it.

                Then there is the case of the National League award in 2010, unanimously won by the great Roy Halladay.    My methods suggest that the award "should" have gone to Ubaldo Jimenez of Colorado.   Ubaldo was 19-8, leading the league in winning percentage (.704), and he posted a 2.88 ERA in Coors Field, which is equivalent to about 0.55 in a normal park.

                But the voters have a point here, too.  Ubaldo was 15-1 with a 2.20 ERA, then went 4-7 the second half of the season.   Roy Halladay pitched a no-hitter in the playoffs; I know it’s supposed to be irrelevant, but it’s hard to ignore.   It’s easier to ignore my calculations than it is to ignore the no-hitter in the playoffs.   I can’t mark that one down as wrong. 

By the way, did you know that Jimenez was not the only major league pitcher named "Ubaldo?"   In 1987 the Expos had a pitcher named Ubaldo Heredia, who got into two games. 

Giving the voters credit for all of the debatable selections, all of the relievers and the tie in 1969, there still appear to have been 29 bad selections in Cy Young history.   These are the pitchers who, in my current best judgment, would have been better choices:

                1956—Herb Score

                1957—Jim Bunning, rather than Warren Spahn.   Similar to the 1956 vote, Bunning was a young pitcher (in his first year in the starting rotation) going up against an established star, in Warren Spahn.   Spahn won the vote 15-1, with the rogue voter going for Dick Donovan.   Bunning’s average performance level was 7.03; Spahn’s was 5.94, and Bunning defeats Spahn in basically every performance indicator, except that Spahn’s ERA was .01 lower.

                1960—Don Drysdale, rather than Vern Law   (Won-Lost record)

                1967—Jim Bunning, rather than Mike McCormick  (Obvious)

                1970—Sam McDowell, rather than Jim Perry.   McDowell had a better ERA, in more innings, in a far tougher park for a pitcher.

                1973 AL—Nolan Ryan, rather than Jim Palmer

                1974 AL—Gaylord Perry, rather than Catfish Hunter

                1975 NL—Andy Messersmith, rather than Tom Seaver

                1976 AL—Frank Tanana, rather than Jim Palmer

                1978 NL—Phil Niekro, rather than Gaylord

                1979 AL—Dennis Eckersley, rather than Mike Flanagan.   If Eckersley had won the award in 1979—which he probably deserved—he would be the only pitcher to have won it both as a starter and as a reliever.

                1980 AL—Mike Norris, rather than Steve Stone

                1981 NL—Steve Carlton, rather than Fernando Valenzuela.   They’re even—unless you look at park effects.   Park factors of 129 vs 91 makes it not even close. 

                1982 AL—ABV, anybody but Vuckovich.   Best answer is tough to see, but it was probably Floyd Bannister, who had a 12-13 won-lost record for the Mariners.  

                1982 NL—Mario Soto rather than Steve Carlton

                1983 AL—Dave Stieb rather than LaMarr Hoyt

                1983 NL—Mario Soto rather than John Denny

                1984 NL—Dwight Gooden (rookie) rather than Rick Sutcliffe

                1985 AL—Dave Stieb rather than Bret Saberhagen

                1988 AL—Roger Clemens, rather than Frank Viola

                1990 AL—Roger Clemens, rather than Bob Welch  

                1990 NL—Ramon Martinez, rather than Doug Drabek 

                1993 AL—Randy Johnson, rather than Jack McDowell

                1998 NL—Greg Maddux, rather than Tom Glavine

                2001 AL—Mike Mussina, rather than Roger Clemens

                2004 NL—Randy Johnson, rather than Roger Clemens

                2005 AL—Johan Santanadanna, rather than Bartolo Colon

                2008 AL—Roy Halladay, rather than Cliff Lee

                2012 NL—Clayton Kershaw, rather than R. A. Dickey