Explaining Roy Halladay’s Futility

Phillies ace Roy Halladay, perceived by many to be the best pitcher in Major League Baseball, was knocked around to the tune of seven runs (six earned) in five and two-thirds innings yesterday in the series finale against the Boston Red Sox. Many in Philadelphia had yet to see Halladay as anything other than the epitome of perfection, so the shellacking was startling to say the least. Good pitchers don’t give up seven runs and fai to make it through the sixth inning, after all.

The Internet was instantly ablaze with excuse-making for Halladay’s first truly bad start as a Philadelphia Phillie. As expected, the most common explanation was that Halladay pitched poorly due to manager Charlie Manuel overworking him — he had thrown triple-digit pitches in eight straight starts and averaged over 122 pitches in his four starts prior to yesterday. If Roy Halladay was overworked, we would expect to see a decline in his velocity, no?

This table shows his average velocity on his three fastballs in each inning:

Inning FF FT FC
1 91.9 92.3 91.9
2 91.8 92.5
3 92.4 92.0
4 91.4 92.6 92.2
5 92.2 93.0 91.4
6 91.9 91.8 91.8
GM AVG 91.8 92.4 91.9
2010 AVG 92.3 92.4

FC = Cutter | FT = Two-seamer/sinker | FF = Four-seamer

While Halladay’s four-seam fastball was 0.5 MPH slower than his 2010 average, his two-seamer stayed the same and his cutter actually had more velocity.

In graph form:

The dip in velocity for his two-seamer in the sixth inning is likely going to catch the eye of many, but it dropped to about 91.8 which is only about 0.5 MPH slower than his 2010 average. Given the small sample size, this certainly should not raise any eyebrows. The standard deviation on his 2010 two-seamer is about 1.5 MPH.

While it is certainly rational to want to limit a star pitcher’s workload in seemingly meaningless games in May, Roy Halladay may simply be an anomaly. Last year, after three consecutive starts in which he threw 119, 117, and 117 pitches, he dominated the New York Yankees in his next start — a complete game victory on May 12 in which he allowed only one run on five hits and did not issue a walk. He also started off September with five straight starts in which he threw 111, 108, 112, 115, and 114 pitches. He finished September with three complete game shut-outs in six starts.

There are more rational explanations for Halladay’s struggles yesterday. Let’s examine them.


A Crashburn Alley article that doesn’t cite random statistical variation? You’re not going to find it. J.C. Bradbury of Sabernomics introduced me to a great quote by Leonard Mlodinow, author of The Drunkard’s Walk:

We miss the effects of randomness in life because when we assess the world, we tend to see what we expect to see. We in effect define degree of talent by degree of success and then reinforce our feelings of causality by noting the correlation. That’s why although there is sometimes little difference in ability between a wildly successful person and one who is not successful, there is usually a big difference in how they are viewed.

When I read that quote, I think of Cole Hamels of course, but I also think it can be applied to Halladay’s performance yesterday. Halladay could throw the same exact pitches in the same exact locations to the same exact batters in the same exact situations and he will almost always experience drastically different results due to factors completely out of his control, even beyond BABIP and HR/FB%.

Boston’s Familiarity

The Boston Red Sox are very familiar with Roy Halladay since he spent so much time in the AL East as a member of the Toronto Blue Jays. Boston batters have compiled 1,159 plate appearances in 275 innings against him in his career. Going into yesterday’s game, eight members of the Red Sox had stepped to the plate at least 20 times against Halladay. Their book on Halladay likely resembles Tolstoy’s War and Peace.

Combining with “randomness” explained above, Halladay’s career 4.39 ERA against the Red Sox is a lot higher than his estimated 3.16 xFIP.

Halladay Was Not Good

Roy Halladay is human, after all. His start yesterday earned him a game score of 26 which is pretty bad. However, of his 297 career starts, he has finished 21 of them (7%) with a game score of 26 or worse. Of course, the distribution of those games is heavily weighted towards the beginning of his career in 1999-2000.

Cliff Lee, to be forever compared to Halladay in Philadelphia, started off his Phillies career with a 0.68 ERA through his first five starts. However, he slowed down at the end of August and into September. In three starts from August 29 to September 9, Lee allowed 17 runs in 15 innings. Was he fatigued? After all, he averaged over 112 pitches in his first five starts as a Phillie.

Red Sox Hitters Were Good

Simply put, a lineup consisting of Victor Martinez, Kevin Youkilis, Dustin Pedroia, Adrian Beltre, J.D. Drew, and David Ortiz is capable of hanging a lot of crooked numbers even on the best pitchers in baseball. Let’s not forget they hung five runs on C.C. Sabathia on Opening Day, four on Zack Greinke on April 10, and five on Francisco Liriano last Thursday. Credit the Red Sox for putting some good swings on Halladay.


The tenet behind metrics like xFIP and SIERA is that there are many factors out of a pitcher’s control. One of those factors is the conversion of batted balls into outs by the pitcher’s defense, hence DIPS: Defense-Independent Pitching Statistics.

Of the Phillies’ 28 errors, nine of them (32%) have come in games started by Roy Halladay. The defense simply has not played well for him. That was exemplified yesterday when Greg Dobbs let a ground ball through the five-hole. It was a sure inning-ending 5-4-3 double play, but it got through Dobbs into left field and allowed two Red Sox batters to score, increasing the lead to 3-0.

While there is merit to wanting to limit Halladay’s workload, there is no evidence that the 490 pitches he threw over his last four starts reduced his effectiveness yesterday against the Red Sox. As the great game of baseball goes, yesterday was a combination of a lot of different factors — randomness, most importantly. Sometimes pitchers have bad days and sometimes hitters have good days. There need not be a deeper causal relationship beyond that.