I wrote this just as the 2007 MLB season was coming to a close. Matt Cain was on my fantasy team that year, and let’s just say that Joe Morgan was probably condemning him to pitchers’ purgatory for his 7-16 record and “inability to pitch in the clutch situations.” I am of the school of thought that clutch performance in baseball is largely a myth, so I set out to determine where Matt Cain’s luck faltered, and perhaps why.
Matt Cain was, as mentioned above, paradoxically 7-16 in 32 starts in 2007 with an impressive runs against per 9 innings of 3.78 (this includes unearned runs as well, giving us an idea of exactly how many runs crossed the plate on his watch). In an attempt to find a reason for this inconsistency, other than randomness and spontaneity, I delved further into the Giants’ and Cain’s statistics.
First of all, the obvious answer is that he received very little run support, about 3.1 runs per 9 innings. Using his teammate, Noah Lowry, for comparison, we can observe the unfair treatment of Cain. Lowry had an above-average 14-8 record with a 4.38 runs against average and 4.58 runs of support, obviously with the same offense scoring his runs. Intuitively, Lowry got more run support, and therefore he had a better record. Case closed. But why did Lowry receive better run support if the same guys are hitting behind him?
To add some confusion to the disparity in records, 74% of Cain’s 32 starts were against teams with winning records versus a roughly-equal 77% for Lowry. And furthermore, Cain pitched 9 more quality starts than Lowry, and has a higher percentage of quality starts: 68% vs. 50% for Lowry. Why is it, then, that the Giants are able to come up with almost 1.5 more runs per game for Lowry than for Cain? Why is Cain’s record so bad?
Cain’s runs against average is very good, better than his 7-16 record would indicate, and better than Lowry’s. Lowry’s defense actually made more scoreboard-affecting errors than Cain’s. Lowry gave up 68 earned runs versus 8 unearned while Cain gave up 81 ER vs. 3 unearned. Furthermore, looking at Cain’s gamelog, there is only one game in which unearned runs had a direct impact on the outcome of the game (he got a loss instead of a likely no-decision). So errors are not the primary culprit of Cain’s woes.
Pitchers who don’t go deep into ballgames sometimes lose out on win opportunities, but Cainwent a solid 6 1/3 innings per start, and Lowry a nearly-equal 6 innings on average. To put things in perspective, Cy Young winner Jake Peavy, from the same division, averaged between 6 1/3 and 6 2/3 innings per start (6.56), and finished the 2007 season with a record of 19-6. (P.S. The Padres weren’t hitting so hotly either scoring just .25 more runs/game than the Giants).
Investigating the ERA breakdown by inning didn’t tell much either. Giving up a lot of early leads might put too much pressure on the offense to score runs, possibly leading to flailing at bad pitches. But in the first three innings, Cain’s ERA was 4.13 and Lowry’s 4.03, not enough of a difference to matter.
Looking at more split stats, the Giants tended to score more runs during the day (4.5) than during night games (4.0), but Cain actually had two more day-starts, and a slightly higher % of day-starts than Lowry. Like day versus night games, home versus away didn’t seem to explain the difference in records. Each player pitched roughly half his starts at home, and each player’s runs-against averages were not significantly affected by that split.
My next step was to look at the opposing starting pitcher’s ERAs in each of Cain and Lowry’s matchups during the season. Cain’s counterparts averaged a collective 4.39 ERA on the season (as of 9/24/07) while Lowry’s averaged 4.56. This is not as big a difference as I had hoped, but might account for a couple close wins in Lowry’s favor, or conversely a couple close losses or no decisions for Cain.
While looking at those stats, I came across something interesting: The Giants’ bullpen did not blow a single one of Lowry’s potential wins, but blew 5 of Cain’s. With zero blown saves for Cain, his record would have improved to 12-16, a reasonable improvement but still no 14-8. Looking more closely at these blown saves, in 3 of those five, the reliever that blew the save pitched the night before, possibly affecting his ability to assure Cain a win. And Brian Wilson—no, not of the Beach Boys—blew his only save in seven opportunities to that point on Cain’s last start. Wilson’s season ERA before that outing was 0.84 in 33 1/3 innings. Nothing but bad luck for Cain.
Another potentially telling stat to look at is how many times each pitcher took a loss in a game where the Giant’s final tally either matched or exceeded their pitcher’s earned run total. This would show how many times each pitcher took a loss partly because of badly-timed run support. Cain has a slight edge here in that 4 of his losses could have potentially been, at least, no decisions (using this statistical definition) compared to 2 for Lowry. So let’s adjust records again assuming “fair treatment”: Cain is 12-12 while Lowry is 14-6.
Looking to see if opposing pitchers outperformed their own abilities turned in some interesting data. I define a “fluke start” as an outing where an opposing pitcher with a season ERA over 5.00 allows 2 runs or less in an outing of 5 or more innings stealing a potential “easy win” from either pitcher in question. Using this definition of “fluke start,” Cain was “robbed” of three potential easy wins—receiving two losses and one no-decision—compared to zero stolen wins for Lowry. Readjusting his record once again (always relative to Lowry’s stats, assuring “fair treatment”), Cain would then have a 15-10 record. At this point the pitchers’ records are beginning to look a little more equal, and it seems poor luck is the only culprit.
The Giants tended to score more runs when Barry Bonds was in the starting lineup. In fact, the Giants scored over 4.5 runs when Bonds got at least 3 plate appearances, but only managed about 3.2 runs per game when he didn’t. So possibly Cain was the victim of unlucky “Bonds-timing.” Bonds got to the plate at least 3 times in just 21 of Cain’s 32 starts, or roughly two-thirds. On the other hand, Bonds stepped into the batter’s box 3 or more times in 20 of Lowry’s 26 starts, good for just under 80% of his outings. However, Cain went just 4-10-7 (W-L-ND) with Bonds’ bat helping him, and 3-6-2 without the slugger. Lowry’s splits were 11-7-2 and 3-1-2, respectively. Each pitcher actually did slightly better than his averages when Bonds didn’t get his 3 plate appearances.
Finally I examined the two pitcher’s records in games where the final score was within 2 runs. In these games, Cain had a 3-10 record (with 9 no-decisions) versus a 5-5 record (4 ND) for Lowry. The difference in wins is easily covered by the blown saves discussed above, but it’s hard to ignore the disparity in “close losses.” So many close losses and no-decisions for Cain indicate that some timely runs could have significantly improved his record much more so than Lowry’s record. This stat was partially accounted for in the paragraph about poorly-timed runs. Cain’s 3-10-9 record in these close games could have been 3-6-13 if he were treated by his hitters in the same manner that Lowry was. This close-game-record comparison shows that Lowry got a little more help—maybe as a result of some timely hitting—than Cain did in the tight ballgames.
In reviewing these masses of stats, one can argue that unfortunate pitching matchups, a bullpen with a Cain-grudge, and untimely Giants hitting had something to do with Cain’s sub-par record and his team’s lack of offensive production. To argue that a pitcher has absolute control over his wins and losses, or that we should even use win-loss records from a single season as a barometer for his true ability, seems ridiculous. There are many more statistics out there to judge pitchers on, and win-loss records have proven to be misleading in more cases that just Matt Cain’s in 2007.