Invention Shouldn’t Require Necessity
Necessity is said to be mother of all invention. It continues to be the impetus for creativity and movement away from tradition in Major League Baseball.
Jeff and I participated in the first postseason chat, a four-hour and three-minute affair Tuesday night that had moments of comedy, drama, soberness — and which featured 2,300 questions from a wonderfully engaged and spirited FanGraphs audience.
The subject of bullpen-ing came up early in the game. This isn’t surprising: it’s been a story of some interest heading into this postseason. I, for example, recently proposed that the Yankees ought to bullpen the Wild Card game. The Yankees, of course, have a dominant bullpen, the first major-league relief corps to feature five arms to have recorded strikeout rates of 30% or better. Aroldis Chapman looks like he’s back, hitting 103 mph, and he’s supported by Dellin Betances, Chad Green, Tommy Kahnle, and David Robertson.
But Joe Girardi went traditional on us. He started his ace, Luis Severino. When asked about the notion of using relievers exclusively for the Wild Card game, Girardi told reporters he wasn’t comfortable with the idea.
“I think that’s pretty risk,” Girard said of bullpen-ing. “I think that’s awfully risky.”
Of course, Girardi’s traditional, Plan A didn’t last long.
Brian Dozier began the game with a home run and an Eddie Rosario two-run shot made it 3-0 in the first.
The Twins!
Upset alert!
The Indians celebrating!
The environment appeared to get to Severino. Velocity wasn’t a problem — he reached 100 mph with his first pitch of the game — but his command was lacking. The slider Rosario hit out was a hanger over the plate. Perhaps sensing that his starter was overwhelmed, Didi Gregorius visited the mound at one point to calm Severino. Not long after, Girardi also visited the mound, motioning toward the bullpen for a reliever. Severino had recorded just a single out.
Among the online chat polls created by Jeff before the game, one asked readers to predict during which inning Girardi would go to his bullpen. I returned to the poll when it became clear that Severino’s night was over. “First inning” had received zero votes.
Green entered the game in that same first inning and limited the damage to three runs. The Yankees responded with three runs in the bottom of the first. With the score tied, and the Yankees in possession of a dominant bullpen, the Twins’ win probability plummeted.
Girardi let Green throw 45 pitches, near his season high of 48.
Robertson followed Green with 52 pitches over 3.1 mostly dominant innings.
So Girardi, who had days ago suggested that the bullpen-ing strategy was “awfully risky,” was forced to quickly embrace that strategy, pivoting to what became, in effect, a bullpen game. He wasn’t at all conservative with how he employed Green and Robertson, allowing the pair to throw a combined 97 pitches, and removing each from the game much earlier their normal, regular-season usage would dictate.
Really, bullpen-ing was everywhere in the play-in round, as the four starting pitchers combined to go 7.1 innings. An average of fewer than 2.0 innings each!
Joel Sherman wrote in the New York Post that, like Andrew Miller a year ago in Cleveland, Robertson expressed his willingness to be deployed when necessary.
Robertson has always been a very good reliever. But since his return to the Yanks, he has combined the ability to attack early, and for multiple innings like Andrew Miller, with a greater economy and effectiveness that hints at Mariano Rivera. His willingness to accept the ball whenever despite being a proven closer has made it easier for Girardi to ask others to sacrifice.
“We are all committed to win the World Series,” Robertson said.
Give Girardi credit for making the call to remove Severino and go to a sort of bullpen game when he did. Nor was it just Girardi: all four Wild Card managers were quick to move on from their starters.
But it’s interesting that it took an emergency, a necessity, to produce the move. And it’s interesting that Green did not melt in the moment when called upon to pitch in the first inning. Ditto for Robertson when employed earlier in the game than is typical. The relievers seemed better equipped to handle the early innings than the starting pitchers.
Yes, Girardi has the advantage in bullpen-ing because Yankees GM Brian Cashman has given him an elite bullpen to work with. But Girardi and the Yankees showed, out of necessity, that bullpen-ing can work on the most important of stages: an elimination game. Green and Robertson shows us that relievers can, in fact, pitch in about any inning and maintain their effectiveness. The Yankees paved the way for true bullpen games to follow.
Not bullpen-ing from the get-go is evidence that fear of the unknown is holding decision-makers back from choices that make sense in theory but have never been put in practice. It’s fear of being second-guessed, fear of what others think, fear of how players will handle situations. Fear has often limited the pace of innovation. But there’s opportunity if you can get beyond fear.
Invention shouldn’t require necessity.
A Cleveland native, FanGraphs writer Travis Sawchik is the author of the New York Times bestselling book, Big Data Baseball. He also contributes to The Athletic Cleveland, and has written for the Pittsburgh Tribune-Review, among other outlets. Follow him on Twitter @Travis_Sawchik.
Even in a bullpenning game, the expectation should still be for the elite starting pitcher to go once or preferably twice through the order. Put him on a short leash, absolutely, but “bench our ace” in an elimination isn’t going to happen any time soon.
I agree. It’s certainly not obvious that having David Robertson go for 3 innings is less risky than having Severino go for 3 innings.
There’s no doubt that having Severino pitch 3 innings is no riskier, and likely less risky, than having Robertson pitch 3 innings, but that misses the point of bullpenning at all. If you’re going to throw Severino 3 innings then he might as well go 6 or 7 since he’s spent enough to at least prevent him from coming back for a full game on Friday in the ALDS. If you want a starter to lead you off then you consider someone whose stuff will play up in a short stint. Tell Montgomery or CC to go give you 40-50 pitches (or use both) but don’t use Severino for only 3 innings. It’s the worst of both worlds.
The worst of all worlds is Mike Matheny insisting his starter pitch 5 innings, regardless of the score, so that he can be eligible for the win and then finding your team down 7-2 or so. Using a reliever early in the game isn’t the worst case scenario by any stretch.
The point of a bullpen game is not to save the starter for the DS. It’s to get the most innings from your best arms. The elite starter is one of those best arms, but only for a few innings.
Actually, if you read the original article a main premise is saving Severino for the divisional series, hence my point of not using Severino and burning him in limited innings. The idea is that the Yankees bullpen is so good and deep that they are as good or better than Severino collectively so the risk is minimal.
That is correct, it was part of the premise for the original article. But if the point is that you’re going to tax a good chunk of your bullpen by pitching them multiple innings in the first game of the entire playoffs (instead of one starter) then that’s not exactly maximizing your pitching ability in that first game of the divisional series either.
The main advantage is that it should help you win the game. And there is some evidence it would do that, although it would be better to actually have the pitcher who is used to pitching in longer stretches pitch in longer stretches.
I’m going to argue that the worst of both worlds is you lose the game. Really, it’s the worst of one world, but it’s so bad your second world (the ALDS) doesn’t exist anymore.
I agree with your premise. I was speaking to the point made in the original article. The ultimate goal is to win the game so if there is evidence that a certain pitcher has a lot of trouble in that second or third run through the lineup but is really strong the first time through then using your deep bullpen to get you 6 innings is a good option.
Severino isn’t really that guy though. The evidence is that he holds his velocity and effectiveness deep into games. That makes him an odd choice to use for this tactic. Tanaka, on the other hand, shows a pretty big drop off that second time through the order so if you didn’t have Severino available then the strategy makes a ton of sense to max your win probability without your ace. However, in the situation presented you had both Severino and the bullpen options available and there’s not much evidence that you gain anything pulling Severino after 3 innings over letting him go 6 or 7 so the bullpen option really only gives you a big advantage if you use it to save him for the DS. If you lose then obviously the gamble backfired, but you can argue it is the better long view strategy if your goal is to win the WS not just a wild card game.
I find it telling that through 8.2 innings of relief pitching in a mostly close game, not one of those innings were pitched by Betances. I think he’s lost Girardi’s trust
He didn’t even warm up. I’m sure he’ll get into a game in the ALDS (probably game 1, considering Robertson and Green won’t be available), but he’s been really shaky lately.
Or he was saving Betances for extras if someone else spit the bit. I mean someone has to be available if bad things happen
That’s why Sabathia and Gray were on the roster.
I don’t think it’s a big loss of trust in Betances, but it’s pretty clear that Girardi would rather not use him for multiple innings. He did that much less this year than he used to, and just once after June 15.
All the moves make sense without reflecting on Betances beyond that issue, given that the pen had to get 26 outs and there happened to be 4 other really good guys available. It made sense to try using as few as possible, leaving as many as possible fresh for the LDS opener. Betances just wound up the odd man out.
really,you sure about that. well, maybe… ok then.
It worked out great, but I don’t agree that it didn’t make sense to try to get some innings from Severino. Because they were stretched out of necessity, each of Green and Robertson tired at the end, which could have led to disaster. I’d rather avoid that risk. On a per inning basis, he may not have been as good this year as Green and Robertson, but he was comparable to the next rank of relievers (Kahnle, Chapman, Betances), so even if you were truly “bulpenning” the game, Severino might have been one of the guys you’d try to get two innings from. And if that’s the case, why not make it the first two innings, and if he’s on, just let him stay in until the first sign of fatigue?
I agree. Also, is the chance of Severino pitching a bad inning any less than Warren, Betances, Robertson, or Kanhle doing the same? Maybe I’m being too simplistic, but in a one game playoff, pitch your aces, both starter and relivers.
Using relief pitchers for most of a game isn’t an OR strategy, it’s an AND strategy. Bullpenning a game is an OR strategy. If you start that way, you are committed to it for the rest of the game.
If anything, this game proves you can start with a starter and switch to relief pitchers after only one out. So, this Yankees game seems more to argue against a bullpenning strategy than to argue for it. Yankees did not embrace a bullpenning strategy, had a horrible starter, used relievers and won the game.
A bullpen game is, and should only be, a “break glass in case of fire” strategy. Yes, it can work for one game – no argument there. But what happens the next game (or the game after that) when you’ve already overtaxed most of your relief corps? The Yankees (and I am a fan) better hope Sonny Gray can give them 7 innings tonight because the Klubot is lurking for Game 2 (OK, it might rain on Friday which would be an incredible stroke of luck but not something you can plan on) and losing Game 1 is probably going to seriously reduce their chances to win the series.
An alternative strategy for the WC game would have been to go to a guy more used to multi-inning outings like Warren if he hadn’t just come off the DL. If Warren were healthy I think that’s what Girardi might have done after using Green to put out the fire in the first inning.
Green and Robertson should be available for Game 2. Neither Chapman or Kahnle worked that hard so they should be available tonight, as would Betances, Adam Warren, and Jordan Montgomery.
Green always gets 2 days off after a long outing.
Robertson has literally never pitched 3 innings before. He’ll get two days as well.
The Yankees had the third best starter in the American League. There was no reason for an all-bullpen game to be the plan. Not to mention now Green, Robertson and Kahnle are unavailable.
Show me a team with a best starter not of that caliber AND a great bullpen and fine, there’s a case of fear or conventional wisdom holding them back. The Yankees are certainly not that. Girardi played it right.
agreed.
Yanks, and Fangraphs have the luxury of considering this option only by way of That Bullpen. Now, look at how it worked out for the Twins after Santana faltered.
This was a bullpen game. Someone had to nominally be the starter, but clearly Severino wasn’t used in a traditional starter’s role. The Yankees planned to use their five best pitchers in succession, and Severino was one of them.
So if Girardi had started with a reliever, say Chad Green, and he gave up three runs in the first inning, would people be saying that the Yankees should have started Severino?
First of many hindsight articles this playoff season? I don’t feel like you’ve made any statements to back up the title, of course invention and risk-taking requires necessity in-game in any sport. Especially in a one-game playoff to validate your whole season, why throw convention out the door at that moment? Severino obviously earned that start with his season, what manager would take that away from him because a projection model says it’ll give you a slightly better chance? That talk is for pundits, not for the people actually determining the outcome of the game.
I agreed with starting Severino. But a do-or-die game is exactly when I’d be most inclined to break from convention. And Girardi did break from convention with that early hook.
Severino throws as hard as nearly anyone in the league, and may be the third best starter in the AL. We don’t know what was going on in Girardi’s mind before the game, but having Severino toss four or five innings, and then turning it over to the BP, would have been a viable strategy on most nights. Give Girardi credit for realizing that Severino was in over his head very early. But I wonder what you would do with a Severino if you were Bullpenning. Stick him in the pen, let Green “start” and then bring him out later in the game?
It seems like all four wildcard teams used their pitching staffs in non-traditional ways, and it worked out for one of them (NY). Arizona, Colorado, and Minnesota all also used relievers early and often (and even put starters in relief roles), and did the ‘all hands on deck’ routine with their pitchers, and they gave up 8, 11, and 8 runs, respectively.
The story to me isn’t that this is some revolutionary idea of using pitchers in non traditional ways; the story to me is the Yankees have really good pitchers.
It’s only *because* the Yankees have so many outstanding relievers that the idea was put forward. The Twins and Rockies have pedestrian bullpens.
One of the drawbacks bullpening a game is it lies on the assumption that all those relievers are going to pitch to expectations. Even elite pitchers have bad days. This strategy greatly increases the chance of encountering a bad day, which in a close game could be disastrous. If you have a super-elite bullpen like the Yankees with 10 pitchers available, then maybe that is a risk worth taking. Still, it seems to make more sense to start the elite starting pitcher you have. If the starter is great, then he eats a bunch of innings and the bullpen margin for error becomes a lot greater. If the starter is terrible, as what happened in the Wild Card, then the bullpen game is the fallback position.
Perhaps the compromise position would have been to tell Severino to pitch like he is only going to pitch 3 innings. He has proven he can be elite in the bullpen in the past. Of course, trying to change a pitcher’s routine in a high pressure situation may not be optimal.
I’m not sure we could assume that Robertson could go 3 and Kahnle could go 2 and be kickass. We know that they could under the right conditions, but that is a big assumption in a playoff game.
I dislike Travis’s implicit methodology here of “bullpenning is a good idea because it worked in this sample size of 1.” In last year’s NL Wild Card game, Bumgarner and Syndergaard combined for 16 shutout innings, but we’d rightfully push back against anyone arguing that game shows that we should just ignore the third times through the order penalty for ace starters.
There’s still the difficult math of an all-bullpen strategy, even with a very deep bullpen, of counting on 4 or 5 relievers to get 27 outs, implying that the relievers need to average approximately 2 innings worth of outs.
Bullpenning worked in this case, and good for Girardi in showing flexibility to adjust once Severino struggled in the first inning. I struggle with the before the fact math, however, of assuming that something like 3+ innings of Robertson, 2+ innings of Kahnle, and 2 innings of Green will necessarily be better than Severino once to twice through the order (3-5 innings) and then shorter outings from each reliever.
If we look at the Fangraphs Depth Chart projections, Severino’s forecast FIP (3.60) is in the same range as Kahnle (3.40) and Green (3.44) and about a half a run higher than Robertson (3.18). Those projections also appear to assume, however, that Severino pitches some for the third time through the order (about 6 innings per start) and that the 3 relievers each average 1 inning per outing. Lower Severino’s projected FIP a bit for not having to pitch into the 3rd time through the order, bump the relievers’ projected FIP’s up a bit for longer than normal outings, and I’d say that Severino’s forecast FIP for 3-5 innings is probably better than the incremental innings needed from those relievers.
The before the fact strategic choice of “full bullpenning” isn’t about whether to plan on using those relievers at all in the middle of the game. Rather, it’s about not using Severino at all and replacing any innings from a starter with greater than 1 inning outings from relievers, so it’s about replacing what we’d expect to be the starter’s most effective innings with reliever innings that we’d expect to be less than those relievers’ average effectiveness.
Right, but it’s also about maximizing the chance of winning the division series by starting Severino twice there, even if doing so reduces the chance of winning the wildcard game by a tiny margin.
It’s necessity the reason we’re talking about bullpening in the first place?
The Twins were also forced to bullpen and it didn’t go well
D-backs will beat the Dodgers in 5 games or less, and they will score at least 7 runs a game. Go ahead and down vote this, it will give me more pleasure when Kershaw takes a dump on the pitchers mound like he does every October.
“D-backs will beat the Dodgers in 5 games or less, and they will score at least 7 runs a game.”
I would hope it would be 5 games or less.
D-back will beat Dodgers in five games or less, with at least 7 points per game. I’m making progress and voting, it will give me more joy when it takes a pot on the floor, as well as every day.
perhaps a point already made.
to change the way you deploy your pitching staff is the risk here. It means a pitcher needs to approach the game with a different mind set and preparation. Not a good bet in any situation, but in a playoff of one game?!
I don’t want to be a hater but could it be that many sabermetric writers Focus too much on single game expectancy when talking about BP Management (closer vs bp ACE, earlier reliever use…)?
in the Playoffs there are more of days but still wear and tear matters. for a single game bullpenning makes sense but it is also harder for the good BP pitchers.
We could see that very well last WS in game 7 when both miller and chapman were gassed from the heavy usage and lost effectiveness.
you Need to save your bullets and over a season you Need to get the innings and have enough off days for the BP and more after a Multi inning stint.
if you want to win a single game and not worry about tomorow then yeah, you bring the closer in the 7th but that puts also more stress on his arm.
that is why having one or two Starters who can go 7 once in a while is still valuable, it gives the BP some rest. you just can’t have 4+ BP innings every day.
Perhaps we should assess value to a pitcher based upon how many times he can successfully navigate the batting order of the opponent. Twice through the batting order means your elite, once through, very good, 6-8 batters, good. Of course, you would place a high value on a couple of pitchers who can destroy 6 batters or less. Whaddya think?