Posted in MLB, Philadelphia Phillies, Sabermetrics | Print | 48 Comments »
The title is exaggerating, but in the discussion that ensued on my finger-wagging article about the misuse of Cole Hamels, reader and commenter Pete got the hamster wheel a-spinnin’. If you haven’t read the article, click here, or perhaps the following synopsis will suffice.
My argument was that Charlie Manuel was wrong to use Hamels with an eight-run lead going into the eighth inning, when the Phillies had a 99.7 percent chance to win. Many counter-arguments were made in the comments, such as that there was a double-header coming up where the bullpen would presumably be needed (the bullpen pitched a combined five innings in both games), or that Hamels needed to be “conditioned to throw more pitches”.
In reality, though, there is no counter-argument to the claim that using Hamels in that situation was unnecessary (sparing the obvious semantic debate about the definition of “necessary”). With two innings of regulation left and an eight-run lead, to abstain from using the bullpen implies that they were worse than a 36.00 ERA pitcher. I’m not exactly confident in David Herndon, but I trust him and his bullpen compatriots to not allow eight runs in two innings.
In the comments, Pete wrote:
Correct me if I’m wrong, but it seems you are a step away from advocating for having pitchers pitch towards a certain expected win percentage? What percentage would that be? 99%? 95%? X%? Any way you fixed it, it would be a minor revolution in the game if managers started following this line of thinking (and probably the last straw for those people who already hate pitch counts and pitching towards anything other than a win). I can imagine a scenario where the Phillies score 10 runs in the top of the 1st inning and a newly Baer-schooled Charlie Manuel doesn’t send Hamels out to pitch the bottom of the 1st. I am imagining also that Charlie’s decision would be a misinterpretation of the Baer rule, but we better start preparing now to see Charlie muck it up.
I like that way of thinking. Not so static, but the overall principle makes a lot of sense. Contract stipulations (money, years, incentives, etc.) should — and do — play a huge role in player usage; this is one such case. Hamels has one more year of arbitration before he is eligible for free agency, and the Phillies will want to lock up a player of his caliber before he is able to test the open market. As much as it may pain the traditionalists, Hamels should absolutely be pampered, as should anyone else to whom the Phillies do or could owe a lot of money over many years (e.g. Roy Halladay, Cliff Lee, etc.). Protecting investments is one big factor in the current and continued success of a franchise.
As for specifics on when to yank a starter, it should be a case-by-case decision. The manager will need to assess a lot of details. By how many runs does my team lead? What inning is it? Do I trust my bullpen? Which relievers will I use? What are the long-term effects of employing this strategy? (e.g. does my bullpen need more time to recover?) How many runs can I expect my team to tack on before the end of the game?
As an example, according to Tom Tango’s “The Book”, the home team has a 16.7 percent chance to win if it is behind four runs entering the top of the second inning. The road manager should expect his team to score more runs, making things worse for the home team. After all, the average National League team averages 4.1 runs per nine innings, and even the worst team in the league averages 3.3 runs per nine.
The biggest problem here is the uncertainty. With the thousands of innings of data we have from the current season, we can make fairly accurate assessments about what to expect in the big picture. Unfortunately, in the span of eight innings, our ability to make correct predictions drops precipitously. In a vacuum, where average players play on average teams and they face similarly average players on similarly average teams, we can improve our odds, but since — as the traditionalists will tell you — baseball is not played in a vacuum, those eight innings are highly prone to all of the variables that makes it such a great sport. Temperature, wind, strength of the opponent, the lineups that are being employed, the opposing pitchers, etc.
In that first inning scenario, I am much more confident pulling Homer Bailey of the Cincinnati Reds since they are more likely to tack on more runs (4.9 per game) than the Atlanta Braves (3.3). I am much more confident pulling Hamels if I am playing the Houston Astros (allow 5.0 runs per game) than if I am playing the Braves (allow 3.3). If it is cold and the wind is blowing in, I would expect to score fewer runs, so I would be hesitant to pull my starter again. If my bullpen is bad like the Los Angeles Dodgers (4.84 ERA), I would lean on my starters, but if my bullpen is great like the San Diego Padres (2.44 ERA), I would be quicker to make a call to the ‘pen. Those are just the broad strokes; we haven’t even gotten into the specifics, such as pitcher batted ball splits, or batter/pitcher platoon match-ups.
So, you can see the initial problems with this practical use of starting and relief pitching. Many situations will not be as cut-and-dried as the Hamels situation, where you have just 0.3 percent left to ensure a victory. However, this would be a gigantic step in the right direction. It would, undoubtedly, reduce the number of superfluous injuries that occur to pitchers throughout the course of the season.
Maybe Hamels doesn’t finish off the last two innings of that game, but he also doesn’t leave the game with a back injury. If the injury had been more serious, he lands on the disabled list and the Phillies have to rely more on Kyle Kendrick, as well as Vance Worley. A rotation of Halladay/Lee/Oswalt/Hamels/Kendrick leads to fewer losses than Halladay/Lee/Oswalt/Kendrick/Worley. Theoretically, being highly risk-averse leads to more wins. With players to whom you are paying many millions of dollars over the course of many years, more wins over many years means a more successful franchise, happier fans, and increased revenue.
In summation, the above was a long way of saying, “Don’t use your starting pitching superfluously, especially if they’re important.” Want to run Kendrick out there with an eight-run lead in the eighth inning? Knock yourself out. But Hamels, or Halladay, or Lee? Protect them! The implementation doesn’t have to be nearly as scientific as I have described it; all that is required is a broad recognition of your team’s chances to win the current game, the side-effects of your decisions on games in the immediate future, and the overall health of your team in the big picture. Yes, it’s so easy, even a Charlie Manuel could do it.