Dear Phillies Phans,
I understand that rookie manager Gabe Kapler has already become quite the lightning rod in the city of Ben Franklin with some rather strange-looking decisions. For those of you interested in more than just first-week overreactions, let’s look at them one-by-one and see if they actually make sense.
Warning! Gory Mathematical Details Ahead!
On Opening Day, with his Phillies up 5-0 against the Braves in the bottom of the sixth inning, Kapler went to the mound, somewhat by surprise, to tell starter Aaron Nola that his day was over. Nola had only thrown 68 pitches and was the owner of 5 1/3 shutout innings. Left-handed reliever Hoby Milner replaced Nola and faced Freddie Freeman. Freeman promptly hit a Milner pitch over the right field wall to drive in Ender Inciarte (who had doubled to lead off the inning) and make the game 5-2. This eventually became relevant because the Braves plated three runs in the bottom of the eighth to tie the game, and then Nick Markakis sent the SunTrust Park fans home happy by hitting a three-run walk-off home run in the bottom of the ninth.
What Was Kapler Thinking?
The recurring theme of this article is that had Milner gotten Freeman out (and also gotten the left-handed Markakis out, which he did), the inning would have been over and it’s likely few people would have mentioned it. There also would have been plenty of tongues clucking if Kapler—who had just watched Nola give up a leadoff double to Inciarte and a loud fly out by Ozzie Albies as he began his third trip through the lineup—had left Nola in to face the Braves’ best hitter, to whom he would have been spotting the platoon advantage. Maybe Freeman hits that home run anyway.
We don’t fault a manager for “good process, bad luck,” but was it a good process?
At the time that Kapler made his decision (bottom of the sixth, one out, runner on second, up by five), the Phillies had a win expectancy of 94.6 percent. The game isn’t in the bag at that point, but Kapler had to feel good about how Opening Day was going. But here we run into a place in sabermetrics that’s a bit more art than science. There still isn’t a grand unifying theory of starter usage that takes into account …
- The fact that starters tire as they throw more pitches, and …
- There comes a point when someone in the bullpen is probably just a better idea on a percentages basis, but …
- You can’t always go to that guy because if you do too often, you’re going to over-extend your bullpen by the end of June, and …
- Whatever else starters do, they are trained to provide bulk innings, so taking them out before they have a chance to do that is a bad idea, but also …
- You want to make sure not to over-extend them within a game because high pitch counts are a risk factor for injury.
Presumably, Nola had another 30 or so pitches left in him, which could have bought Kapler another inning or two if he was lucky. Or did he? In his final spring tune-up, Nola went only four innings (of shutout ball). Pitch counts weren’t available, but he faced 17 batters that day. The time before that, he’d thrown five innings and faced 21 batters. Maybe Kapler figured that with the game (mostly) in hand, he should err on the side of not over-extending his starter because he didn’t have to, although it would also mean erring on the side of potentially over-taxing his bullpen.
The problem with being a manager is that you are constantly having to make decisions where you end up being wrong and the options aren’t always so cut-and-dried. Or maybe it’s something a little more philosophical. Here’s a table of Phillies starting pitcher appearances so far in 2018, by innings, runs allowed, and pitches:
|Aaron Nola (3/29)||5.1||1||68|
|Nick Pivetta (3/30)||4.0||3||73|
|Vincent Velasquez (3/31)||2.2||7||69|
|Ben Lively (4/3)||5.2||2||91|
|Aaron Nola (4/4)||5.0||2||87|
|Nick Pivetta (4/5)||5.2||0||97|
|Vincent Velasquez (4/7)||6.0||1||98|
|Jake Arrieta (4/8)||4.0||3||73|
|Ben Lively (4/9)||5.2||5||100|
The first time around, with each starter, Kapler was a little more reserved with pitch counts. Only Ben Lively broke 90. We didn’t know this on Opening Day, but in retrospect it looks like Kapler had it in mind to go easy in the early going. Other than Velasquez’s not-so-nice outing, no one was having just an awful day at the point they were taken out.
What’s weird, though, is that when you start thinking this one through, later in the season, the numbers would suggest doing the opposite of what Kapler did on Opening Day. Assuming a pitcher is fine to throw 100 or so pitches, numbers 70-100 may not be his best of the day, and there may be someone better in the bullpen, but in a lower-leverage situation—like, say, up five runs in the sixth inning—you might as well leave the starter in there and let him eat innings. He might not be your best pitcher in the moment, but you don’t need your best pitcher in the moment if the game is in hand (or out of hand). If it had been a one-run or tied game, going to the bullpen would make more sense. And maybe in the middle of the season, in the exact same set of circumstances, Nola would have stayed out there.
In the third game of the season, jack-of-all-positions Pedro Florimon pitched the eighth inning of what ended up as a 15-2 blowout. Florimon gave up two runs on a walk to Nick Markakis and a home run by Lane Adams, but given that his teammates had given up 13 runs in seven innings, he wasn’t that far off the team average.
What Was Kapler Thinking?
Again, this one appears to have been planned. Florimon, who had never pitched in the majors before, threw bullpen sessions during spring training in anticipation of this moment. Last year, the Twins did the same with Chris Gimenez. And why not? If the game is already out of reach, why send out another pitcher to add more pitches to his arm-o-meter? Yeah, it looks funny and Twitter will gawk, but Twitter does that with random stock photos.
In some sense, this is a brilliant move, specifically because it means that Kapler can be a bit more aggressive in his use of the bullpen, knowing that he has a fallback option to fill in five or six innings over the course of a season, as long as those innings don’t really matter. And that eighth inning against the Braves did not matter. There were those who criticized Kapler for over-extending his bullpen to the point at which he was forced to use Florimon. The opposite was true. He over-extended his bullpen because Florimon was just part of the plan.
On April 3, Kapler employed a shift against Jay Bruce. This is nothing new for Bruce, who is nearly Ortizian in how often teams shift against him. But Kapler left third baseman Scott Kingery on the third base line to defend against the bunt. With a runner on first and no outs, Bruce hit the ball to the right side of the infield, where both second baseman Cesar Hernandez and shortstop J.P. Crawford were playing. Crawford probably should have broken for second to cover the bag and potentially start a double play, letting Hernandez take the ball, but it turned out that Crawford fielded the ball and had to go to first for the out. The next batter, Todd Frazier, doubled in the runner (Yoenis Cespedes), who was not doubled off, and scored himself with two outs on a Travis d’Arnaud single. The Phillies lost on those two runs, by a score of 2-0.
What Was Kapler Thinking?
Let’s go back to the Bruce play. For one, that’s poor execution of the shift by Crawford and Hernandez, and at some point, that comes back on Kapler’s plate. But why was Kingery guarding the line? Usually in the shift, the fourth infielder plays near the regular “shortstop” position. Kingery was nominally the third baseman that day, but has played shortstop for the Phillies already. Had he been in the shortstop zone, he could have come across to take the double play throw. With two outs and no runners, Frazier’s double wouldn’t mean a run, and Adrian Gonzalez’s ground out following Frazier’s double would have ended the inning. Why on earth was Kingery guarding the line against a bunt by Bruce of all people?
Bruce may have pulled one over on Kapler. He has regularly put in time practicing bunting and maybe more importantly, made a fuss over it, in spring training. The math says he should try it once in a while for real. But Bruce has had one bunt hit against the shift (and he sees it a lot) since 2014. Not one per season. One. If Kapler had reason to believe Bruce was going to bunt, then it would be correct to leave Kingery there to guard against that from happening. If a batter can bunt the ball fair against a shifted infield, he’s got a really good chance at a base hit.
Is it possible that just the idea of Bruce bunting got into Kapler’s head?
On April 4, with runners at first and third, and two outs in the bottom of the sixth inning of a tied game, Kapler brought right fielder Nick Williams in to play shallow against Mets hitter Amed Rosario. After the game, Kapler stated that he was trying to optimize the outcome of the play (which I assume means he was trying to pick the positioning for Williams that had the best expected value). Rosario, who apparently didn’t get the memo, tripled over Williams’ head and plated the two runners that were the eventual difference in a 4-2 game.
What Was Kapler Thinking?
This one is a bit harder to justify, but I think I see the basic premise. It’s important that the game was tied, there were two outs, and there was a runner on third. The two most important runs in a game are the run that ties the game and the run that un-ties the game, and I think Kapler was starting from there. The Phillies were in specific danger of giving up the un-tying run. If Rosario got a hit (any kind of hit) that run would score.
Part of the unfairness of a baseball field is that it’s so large that there’s no way the nine men tasked with defending it can cover it all. There will always be pockets of the field unguarded. The reason outfielders don’t play shallow all the time is that while it probably allows you to catch some balls that would otherwise drop in front of you, it also will allow some to go over your head. If you’re playing deep and you guess wrong (i.e., the ball is hit shallow), at least the ball is in front of you and headed toward you, rather than hit over your head. If we over-simplify Kapler’s choice to playing Williams either shallow or deep, the reward for guessing right as to whether the batter will hit the ball shallow or deep is the same (you can catch a fly ball for an out), but the punishment for guessing wrong is greater if you are playing shallow.
You’d need some other circumstance—for example, if you somehow knew for certain that the batter was going to hit the ball to short right field—to justify playing shallow. In this particular case, the fact that—for the purposes of scoring that un-tying run—a single is functionally equal to a double or a triple probably weighed heavily in Kapler’s thoughts, but I think he may have been over-playing that hand a bit.
If Kapler had absolutely no idea where Rosario was going to hit the ball (whether shallow or deep), and he guesses right, his reward is the same third out. If he plays the fielder deep, and Rosario hits a short line drive or fly ball that lands for a single in front of the runner, then you’re looking at the runner at third scoring, and there now being two outs and runners and first and third. (Maybe first and second, but let’s just go with first and third for now.) That’s bad, but the deep-playing outfielder can stop the damage there. If Kapler plays his fielder shallow (which he did) and guesses wrong (which he did), then he’s now got a two-run deficit and a runner on third because the ball is over his head.
This table shows how this same set of circumstances (bottom of the inning, two outs, first and third) plays out on the win expectancy graph, depending on the inning. We’re going to look at the win expectancy for the initial state, the change in that win probability for the first part of the hit (single to right, one run scores, runners now at first and third), and the change in win probability for the second part (where it becomes a two-run triple).
|Inning||Initial Win%||After 1-run single, runners to 1st and 3rd||Delta from initial||After 2-run triple||Delta from 1-run single|
“Delta from initial” is the penalty for allowing the ball to drop in the first place. “Delta from 1-run single” is the penalty for letting the ball get past the outfielder.
The initial run scoring is much more damaging to the Phillies’ chances of winning the game compared to the ball getting by the right fielder and knocking in the second run. But that ratio gets more severe as the game wears on, until the ninth inning when it literally doesn’t matter whether the hit is a single, double, or triple. The game is over, no matter what. In the ninth inning, the correct approach is to play the fielder in the place where he would have the greatest chance of turning a batted ball into an out. If the spray chart says there’s a 51 percent chance of the batter hitting it shallow, then play him shallow.
In the sixth inning, though, the ratio between the the two columns isn’t quite as big. Yes, it’s still going to be very important to try to stop that ball from becoming a hit, but the danger from the ball getting past a fielder playing in too shallow also has a big penalty attached. In the ninth inning, even a single is a death blow. In the sixth, there’s still some game left to play. You have to think about that downside risk in the sixth inning, and so you have to be a bit more careful.
Looking at Rosario’s brief MLB spray chart, there’s a bit of a strange outcropping of balls hit into short right field, which depending on where you set the cutoff of “shallow” and “deep” looks to be about half of his balls to the rightward part of the outfield.
Rosario was never a power hitter in the minors and Kapler had the added benefit of knowing how the Phillies planned to pitch to him in upcoming plate appearance. Maybe he felt very confident that Rosario was more likely to hit the ball into short right field than deep right field, though the math says he needs to be particularly confident before it makes sense to pull Nick Williams in.
Once again, we’re talking about this because it didn’t work in a very loud and obvious way. There are variables we can’t calculate that might have made this the correct decision, but given what we know, I wonder if Kapler made the correct strategic decision for the eighth inning, but not the sixth.
On April 5, Kapler took Nick Pivetta out of a game in which he had thrown 5 2/3 shutout innings.
What Was Kapler Thinking?
Like it or not, abusive pitch counts are a real thing. Pivetta wasn’t going to go nine shutout innings with his pitch count already at 97. The game was still 3-0 at that point, there was a runner on first, and Kapler got the Adam Morgan-on-Justin Bour lefty-on-lefty matchup he wanted. Win the game first, worry about the trivia questions later.
What Was Gabe Kapler Thinking?
Well, look. Williams is a defensively limited corner outfielder. He can hit. Maybe in some cosmic sense, he deserves better than this, but let’s take stock of the Phillies’ current roster situation. Kapler appears committed to three players every day in the lineup: second baseman Cesar Hernandez, first baseman Carlos Santana, and “left fielder” Rhys Hoskins. That leaves center and right field for Odubel Herrera, Aaron Altherr, and Williams to fight over. (Scott Kingery has also seen time in right field so far.)
With a defensive liability already in left field, it’s tough when you have one in right field to go with him. The downgrade in defensive performance from Williams to Altherr or Herrera is worth a lot, and given that Altherr and Herrera aren’t slouches with the stick, the downgrade in defense might be worth more than any potential upgrade in offense. Kapler is doing the logical thing with his resources. Williams makes a dandy “first bat off the bench” pinch-hitter, particularly against right-handed pitching.
This pretty much tells you what you need to know. Williams has started three games so far in 2018. Two of them backing up ground-ball generator Aaron Nola and one backing up noted foe of worms Jake Arrieta. Kapler is playing Williams when he knows that his defense isn’t going to be needed as much. Around here, the idea that defense matters is something we take for granted. Once in a while it’s worth remembering that there are plenty of fans (and um, players) out there who haven’t heard that message yet. That’s not computers making lineups. That’s a candid look at what resources are available and optimizing with what you have.