Wednesday, August 22, 2012

LaTroy Hawkins and the 2004 Cubs

I was reading a bit about the Cubs today, and the subject of the 2004 Cubs came up. And, unsurprisingly, LaTroy Hawkins was prominently featured as one of the reasons that team did not make the playoffs, even though it was in many ways better than the 2003 Cubs, who did. When this happens I usually like to make a little comment that Hawkins wasn't actually as bad that year as many perceived, and that he never really deserved much of the criticism he received from fans and the media that year. But blog comment space doesn't allow for a full analysis of Hawkins' 2004 season, and the reasons that different stats paint a very different picture of the sort of season he had. This format does, so here goes.

The popular narrative is that LaTroy Hawkins was a talented pitcher that couldn't perform under pressure and was thus uniquely poorly suited to be a closer. Indeed, in 2004 as he closed for the Cubs, his ERA was stellar at 2.63 and his strikeout and walk rates excellent but he blew 9 saves.

Aggregate stats like ERA and strikeout rate have the weakness that they don't use the individual game as a unit. A closer's save record, like a starter's W-L record, hold the individual game up and attribute its result to a pitcher. That's why they resonate in ways that aggregate rate stats don't. Unfortunately they can also be pretty inaccurate measures of a pitcher's impact on the games he pitched in. And sometimes, as in Hawkins' case, they're even misleading indicators of what happened while he was pitching.

Measuring a single player's actual impact on a game is impossible, but measuring what happened while he was pitching is done regularly with WPA. Hawkins' WPA didn't shine as brightly as his rate numbers, but it was pretty good, around 1.2. It's better than any other Cub reliever that year, and in line with what many other good relievers did. You'd expect a reliever that blew so many saves to have a bad WPA. His decent WPA suggests that his blown saves were not as bad as those of many others. Indeed, my recollection is that several times he blew saves in games the Cubs won; he held the tie and often was credited with the win himself. Even if he performed poorly he left with the game still in reach. WPA will rightly say this sort of blown save is better than the stereotypical reliever meltdown, where the closer gives up the tying and winning runs.

So what do the game logs say? Let's look at the 9 blown saves of LaTroy Hawkins in 2004.

  1. April 28: Hawkins entered the game with one out in the 8th and a one-run lead. He gave up a home run to tie the game and finished the inning with the game tied. The Cubs happened to score one in the 9th and win (this does not bear on Hawkins' WPA, nor on any rational analysis of his performance, though he was credited with a win). Although it occurred before Hawkins was installed as the closer, this fits the pattern of a blown save that was, while not good, about as good as a blown save can be.
  2. May 28: In the second game of a double-header Hawkins entered in the bottom of the 9th with a two-run lead and gave up a two-run homer to tie the game. He finished out the 9th and was replaced for the 10th by Francis Beltran, who threw one pitch that was hit out for a home run and a Cubs loss. Again, Hawkins allowed the tying run but not the winning run to score, though this time he blew a larger lead.
  3. June 13: In the bottom of the 11th Hawkins entered with a one-run lead. He allowed three singles (one which he fielded; it is possible though not certain that he fielded poorly -- either way, WPA counts it the same as any other single) and the tying run, finished the 11th, pitched a 1-2-3 12th, and exited with the game tied. The Cubs went on to win. The pattern holds.
  4. July 4: A one-run lead going into the top of the 9th, a solo home run, a blown save, a tie-game departure, a Cub run (on an RBI walk by Todd Walker), and a win. The pattern is four-for-four.
  5. August 21: Another one-run lead in the bottom of the 9th. Hawkins recorded only one out before allowing two runs and picking up the loss. A throwing error on Aramis Ramirez contributed; still, the pattern is finally broken.
  6. August 25: Exactly like July 4, except in the details. Hawkins gave up the tying run on a double to a guy named Magruder, and the Cubs won it on a solo shot from Corey Patterson. Pattern is 5-for-6.
  7. September 21: Another for the pattern (6 out of 7), and another blown save-win for Hawkins. This time it was the bottom of the 9th, and Patterson (another Cub that had a pretty solid season that many remember as disappointing) scored the winning run on a wild pitch, setting up a save for Ryan Dempster.
  8. September 25: Hawkins entered with two on and one out in the bottom of the 9th and gave up a 3-run jack to tie it up. That's -0.4 on the WPA scale, which is exactly halfway between 0.1 (what he'd have got for the save) and -0.9 (what he'd have got had he given up another donger instead of finishing the inning). That's the very mathematical signature of the pattern. 7 out of 8. Oh, yeah, the Cubs lost in extras, holding onto a slim half-game wild-card lead.
  9. September 29: I'll have the regular. The real orthodox regular. In a one-run save opportunity in the top of the 9th Hawkins gave up one run. The Cubs went on to lose in extras and fell a half-game back in the wild-card race.

So the final score: in nine blown saves LaTroy Hawkins finished out the inning with the game tied eight times, giving the opponents the lead only once. On top of that he picked up losses after entering tie games three times. This does not amount to a great season, and the stretch run was particularly mediocre, but it does amount to a better season than the blown-save count and popular narrative would lead you to believe. And that bit about a decent-but-not-great season with a mediocre stretch run? That describes the 2004 Cubs to a T, the whole team. Singling out any one player is silly.