Hindsight Is 20/20: Game 4 Managerial Decisions

Saturday night, the Rays and Dodgers played one of the wildest World Series games ever. Leads changed hands, runners slipped, pitchers crumbled, and the Rays walked it off in spectacular fashion. At the time, I criticized several managerial decisions, and I wasn’t alone. With the benefit of a few days of thinking, however, I wanted to look back at a few key decisions each manager made and decide whether they were blunders or merely tough decisions that looked worse in hindsight.

For the Dodgers, the key managerial decision was the relief pitcher hierarchy. After a spectacular pitching performance from Walker Buehler the previous night, Dave Roberts had the entire bullpen available. His first decision came with two outs in the fifth inning, when Julio Urías began his third trip through the Rays’ lineup. Urías had been up and down on the night; he had nine strikeouts, but he’d also allowed some loud contact and two home runs. The Rays stacked their lineup to challenge him; the first four hitters were all right-handed.

Roberts went to Blake Treinen, and I think that’s a reasonable choice. The Rays had a bench full of lefties, which means any stretch of righties in their lineup can turn into lefties at the drop of a scorecard. Despite that fact, however, Randy Arozarena probably wasn’t leaving the game, and guaranteeing a Treinen/Arozarena matchup, plus forcing Tampa Bay to use some left-handed pinch hitters, is as close to a positive platoon matchup as the Dodgers were going to get.

That leads us to a pivotal pitching change in the sixth: two runners on, one out, and Brandon Lowe stepping to the plate. Behind Lowe, the Rays had Willy Adames and Hunter Renfroe due up. In theory, that’s two righties and a lefty. In practice, Lowe is the only Tampa Bay hitter who the team couldn’t substitute. That left Dave Roberts with three decisions, in my mind — all of which he would have had to make several batters earlier to allow the pitchers time to warm up.

First, he could bring in a lefty to limit Lowe’s production and live with the subsequent lefty/Adames matchup. In the current Dodger bullpen pecking order, that would be Victor González. Lowe/González is a good matchup for the Dodgers; Lowe has been pretty good against lefties in a limited sample in his career, but there isn’t nearly enough data to see him as anything other than a league-average lefty hitter, roughly 8% better against righties than lefties.

On González’s side, there certainly isn’t enough data to think of him as anything other than a league-average lefty pitcher; much better against lefties than righties, again roughly 8%. He’d then get Willy Adames, a much worse hitter who would have the platoon advantage. Adames has actually shown sharp reverse platoon splits in his career, but again, small sample — our best guess is still that he’s somewhere close to average.

The Dodgers could also have gone to Brusdar Graterol, conceding the platoon advantage to Lowe. The Rays would then have an option of their own: if Lowe’s at-bat ended in anything other than a home run or double play, they’d have a high-leverage spot for a pinch hitter. That involves some defensive wiggling — Adames is the team’s primary shortstop, which would probably leave Joey Wendle to play the position — but given that the Rays played Hunter Renfroe at first base in this game, they’re not above novel positions.

Graterol is a better pitcher than González, but he’d face two opposite-handed batters, while González would only face one. It’s not an obvious decision either way; limit Lowe, and you’re helping Adames, and vice versa.

Next, the Dodgers could do what Roberts chose: go with Pedro Báez, a righty who actually does have unique platoon characteristics. Unlike everyone else in this discussion, we have a mountain of data on Báez, 566 plate appearances against lefties to be precise. He’s been better against lefties than righties in his career, and even after regressing him to the mean, he’s roughly equally effective against both hands.

The Rays would still likely pinch hit for Adames if Lowe didn’t empty the bases one way or another, but that’s less damaging against Báez. Assuming the Rays would use Yoshi Tsutsugo as the pinch hitter, I used regressed platoon splits, 2021 projections, and a healthy dash of nested odds ratios to produce wOBA estimates for each batter/pitcher confrontation:

Batter/Pitcher Matchup Grid (wOBA)
Matchup Lowe Adames/PH
González .318 .326
Graterol .333 .332
Báez .338 .337

The results of this aren’t particularly surprising; with Tsutsugo available as a pinch hitter, letting him hit with a platoon edge isn’t where you want to be. The worst hitter of the three Rays is Adames, and righties have smaller splits in general, which leaves him as the best batter to concede an edge against. Báez has those nifty reverse splits — but he’s worse than Graterol overall by enough that he’s still worse against lefties, and much worse against righties, per Steamer’s 2021 projections and some translation.

Pitchers like Báez can mislead. “No platoon splits” makes him sound like a righty who crushes lefties, and it’s true that he’s better than a similar-talent, neutral-split righty. But left-handed pitchers don’t have no platoon split; they have a huge advantage against left-handed batters.

Giving up that free boost in favor of a middling pitcher, particularly given how the Rays lineup was set up, seems foolhardy. It’s not quite so bad as this grid makes it, because Renfroe, due up third, is also a righty, but he too could be pinch-hit for. Simply put, committing to a right-hander against an excellent left-handed batter and an available left-handed pinch hitter is a poor idea.

How much did that affect the results of the game? Negligibly! We’re getting into extrapolation territory here, but by plugging in the results grid I calculated, I came up with just under a 1% change in win probability by using Báez, the worst option, instead of González, the best option. It’s a tiny difference — Báez’s actual results lowered the Dodgers’ win probability by a total of 50%, more than 50 times more than the managerial decision cost. González might well have given up a home run to Lowe or Adames.

Still, I feel justified in saying it was an error. Báez wasn’t the best pitcher available, and he was the worst matchup against the Rays’ best hitter in a high-leverage spot. It wasn’t worth much, but it wasn’t the right decision.

The next inning, Roberts had another decision to make. He could leave Báez in to face the bottom of the Rays’ lineup or go to another reliever. He left Báez in, and it bit him in the form of a Kevin Kiermaier home run. I don’t hate the decision, though — it was a one-run game, but someone has to face the two worst hitters on the Rays, and it might as well be Báez, saving the better relievers for more important at-bats.

Roberts probably overplayed his hand, though. After Mike Zunino struck out and Kiermaier homered, the top of the Rays’ order went to work; Yandy Díaz drew a walk, which brought up Arozarena against a tiring Báez. Roberts stuck with his man and Arozarena grounded into a double play, but that could have easily gone wrong; a mediocre and tiring reliever against the hottest hitter in baseball with a runner on base doesn’t sound like a great outcome for the Dodgers. Rodgers got unlucky in the sixth and lucky in the seventh — but in both cases, he didn’t put his team in the best position to succeed.

Because Roberts used Báez instead of Graterol in these high-leverage spots, he got less out of his bullpen than he otherwise could have. After Adam Kolarek faced three lefties in the eighth (great decision!), Graterol entered to face two righties. The Rays now had less pinch hitting leverage — they’d used Wendle as a pinch runner and promptly deployed Brett Phillips as another. Graterol allowed a duck snort single from Adames but rallied to retire Renfroe, escaping the eighth.

Roberts went to Kenley Jansen to start the ninth, which means Graterol only faced two batters. Seven batters for Báez and two for the superior Graterol? It’s strange managing, to say the least. The fact that Jansen promptly coughed up the lead is neither here nor there; it simply seems like a waste to get only two outs from one of your best bullpen weapons while stretching another worse pitcher for two innings.

In the opposite dugout, Kevin Cash had some decisions of his own to make. I’m not convinced he fared any better, though his decisions were all on the batting side. The first one is a lineup construction mistake, and while those are generally overblown, I still think it’s worth talking about.

The Rays deployed a righty-heavy lineup:

Rays’ Starting Batting Order
Batter Handedness
Yandy Díaz R
Randy Arozarena R
Mike Brosseau R
Manuel Margot R
Brandon Lowe L
Willy Adames R
Hunter Renfroe R
Mike Zunino R
Kevin Kiermaier L

The general idea of stacking the top of the lineup with righties makes a ton of sense. Cash was forcing the Dodgers’ hands; there’s no question of running Julio Urías through the lineup a third time against a gauntlet of four solid right-handed hitters. Two of those hitters are also platoon pieces, so when Roberts inevitably countered with a righty, the Rays could seize the platoon advantage back by using their bench lefties.

One problem with this plan: the way the Rays designed it, Brosseau and Margot were the two batters who would likely be replaced. With Lowe batting after them, that would leave a three-lefty stretch in the Tampa Bay lineup, an easy way to let the Dodgers deploy a lefty with no chance of counterplay — the Rays had no right-handed pinch hitters.

Naturally, this exact event occurred. It almost had to — the Rays were always going to replace Brosseau and Margot. Adam Kolarek got a three-lefty slate, the optimal outcome for LA. How could the Rays have countered this? It’s a small change, but they could have swapped Margot (or Brosseau) and Arozarena in the initial lineup.

Sure, it moves their best hitter out of the two hole, but that’s small potatoes in a single game — lineup optimization is worth only a handful of runs over a full season. On the other hand, it takes advantage of the three-batter minimum; avoiding left-left-left pockets of a lineup is simply smart planning given the current state of baseball.

This was a small error; it essentially cost the Rays two platoon-advantaged plate appearances across the entire game. That isn’t huge in expectation, but it’s not nothing, and it wouldn’t have cost the team much to construct their lineup in a way that anticipated their own later moves.

The decision that I have a bigger problem with is Cash’s use of pinch runners. The first pinch running decision — sending Wendle in for Yandy Díaz in the bottom of the seventh — is understandable, if not obvious. Scoring a run in a tie game is quite valuable, and Wendle was likely to enter the game for defense before long anyway. Additionally, the Rays’ machinations had already removed some good pinch hitting spots; if Cash wanted to keep Adames at short, Díaz was one of the few batters Wendle could replace. Tsutsugo was still on the bench to hit for Zunino at some future point, which left only Adames, Díaz, and Renfroe.

Still, pinch hitting for Renfroe against a right-handed pitcher is a valuable chip. Wendle and Tsutsugo were the only two good hitters left on the Tampa Bay bench at this point; sacrificing one for a marginal running edge in a low-leverage spot — Wendle’s speed would only be the difference between scoring and not if Arozarena doubled — seems strange. As it turned out, Arozarena grounded into a double play, and Wendle’s speed didn’t matter at all.

In the next inning, Cash made a decision I like even less. With Ji-Man Choi on second base and two outs, he deployed Brett Phillips as another pinch runner. This left the Rays with no natural first basemen — Renfroe, who was at the plate, had to play first base the remainder of the game, and he’d played only nine innings at first in his professional career. He also had to bat against Graterol because Wendle wasn’t available to pinch hit.

A pinch runner on second base is a good spot for one. With two outs, however, it’s less valuable, because Choi could leave at the crack of the bat. I haven’t done the math on what percentage of singles are in the exact right area that Choi couldn’t score but Phillips could, but it’s a small subset when there are two outs.

More importantly, the Rays were setting themselves up for disaster in the ninth. At the exact moment that Phillips entered the game, the Rays were quite likely to send Arozarena to the plate in the ninth inning. If the eighth ended with Renfroe’s at-bat, Arozarena would be due up fourth. Should any runner reach base, the Rays could send their most fearsome hitter to the plate representing the winning run.

By pinch running with Phillips, they blunted that edge. He’s the worst hitter on the Rays’ roster, potentially including the catchers; his career slash line is .202/.284/.347. With him following Arozarena, it left an easy escape valve for opposing teams. Don’t want to face the hottest hitter in the postseason? Pitch around him and face the worst batter on our team. Put Choi in that situation instead, and it’s a different story entirely.

Naturally, that exact situation came to pass. Kiermaier reached base, Jansen did his best to avoid Arozarena, and Brett Phillips, the worst hitter Kevin Cash had access to, took the highest-leverage at-bat of the series. Of course, he won the game after a comedy of errors in the outfield and on the basepaths, then air-planed around the outfield in a glorious victory lap. But Cash couldn’t have known that, and he didn’t put his team in the best position to succeed, though succeed they did.

How much did these two managers put their thumb on the scales in Game 4? Only a little, despite their best efforts. That’s the nature of managerial decisions — the choices are never “use my best hitter or call up an 18-year-old from the DSL.” The worst hitters in the game still get on base nearly 30% of the time, and the best hitters make outs more often than not. The worst reliever in the Dodgers bullpen is still a competent major leaguer. Brett Phillips literally won the game, and Pedro Báez retired the Rays’ best hitter (Arozarena) in a high leverage spot.

Just because the maximum effect of manager moves is small doesn’t mean that moves can’t be bad, however. It wasn’t the reason the Dodgers lost and the Rays won; in fact, I’d argue that both managers cost their team a roughly equal amount. That doesn’t mean we should ignore the decisions, however. Both teams could have been fractionally better in Saturday’s game, and they simply chose not to be. That’s always worth recapping, even if the part of the game we’ll remember is Arozarena’s mad dash and Phillips’ wild celebration.

Ben is a contributor to FanGraphs. A lifelong Cardinals fan, he got his start writing for Viva El Birdos. He can be found on Twitter @_Ben_Clemens.

newest oldest most voted

7 batters for Baez and 2 for Graterol. Yes, I agree completely that this was terrible decision-making by Roberts. What reliever gets to blow TWO leads in the same World Series game? And all along you had Graterol and Gonzalez, and even Dustin May in the bullpen as better options. Losing with your best players is one thing, but losing by giving such a long leash to Pedro Baez, who has been blowing playoff leads since 2014 was really hard to stomach. Glad they simply pitched their best pitchers in game 5.


Correct me if I am wrong, but nobody had a worse night than Trenien in game 4. Yanking Urias was a mistake. They have repeatedly yanked their starters early and paid for it. Nobody was worse in game 4 than Treinen.

They pitched their best in game 5? Kershaw was their best option and they went away from him unnecessarily. They literally let him face the best hitter on TB in a bad match-up (righty) whom he retired and then they put in May vs an easier righty later in the inning. In doing that the Dodgers prioritized getting May a soft match up over using their best player. If you let your lefty face the best righty, then you need to let him face the next one too if he is cruising (he was – 7 straight I think) . LAD chose helping May over helping the Dodgers. If May was their best option vs a righty, then why not let him face RA? They let May face the soft underbelly of the lineup – essentially just padding his stats and burning the bullpen for fun. Typical Dodgers move, but that is how you make it a series I guess.


I’d disagree. May’s filthy but he had not been pitching well, and they had just made mechanical tweaks to try and get him right. They need him to contribute. By giving him a scenario where it is hard to snowball out of control – none on, 2 out – they were able to see if his mechanical fix would carry over into a game situation and get some useful outs out of him with a relatively low risk of building a big inning because of the base/out situation. If he looks shaky, he’s still likely to retire 1 of 3 batters and get out of the inning, and you can move on to someone else. If he looks better, as he did, you can let him have the next inning as well (which is what happened). Letting Kershaw face Arozarena was a calculated risk: May’s better at that matchup, but much worse vs Lowe. They decided that the Kershaw vs Lowe gap was large enough to be worth letting Kershaw face Arozarena, and that gamble paid off (and it was a gamble). If you go to May to lead off the inning, if things go poorly you have to get 3 outs to end the threat, and whatever you might gain vs Randy A, you give back in RHP vs Lowe.

Your analysis ignores the reality that the Dodgers need May to contribute to win this series easily. Even if they lose Game 6, they need to do so without completely shredding their bullpen for Game 7. This requires getting some outs from May and Gonsolin. Finding out if they’ll be able to do so was valuable, and also spared the rest of the pen after it had been used fairly hard the previous night. They now go into Game 6 in a situation where only Treinen is likely to be limited (having had 1 day off after 3 in a row), which is a much better position overall.

Treinen’s also been one of their most consistent relievers all season, and they’re not going to get side-tracked by a single bad outing, especially when you’re dealing with a heavy groundball pitcher so you know a certain portion of the time those ground balls are just going to find holes and they’ll give up a few hits. Treinen didn’t look bad in Game 4, and there was no reason to think he was wrong or would be ineffective in Game 5 – in fact, he’d pitched very well vs Atlanta in a similar situation used 3 in a row, so there was actually more evidence he could handle that situation than most pitchers as he just did so recently.

The thing about ‘7 straight’ is that what it tells you is that the guys he’s about to face are the last ones he did NOT retire. How Kershaw looked vs Adames and Zunino is far less important than how Kershaw looked the last time he faced Margot and Renfroe, and rather famously those guys both reached base in their previous trip to the plate.

Treinen and Gonzalez definitely represent the most consistently effective relievers for the Dodgers this season, and if you weren’t aware of that you’ve just never clicked and looked at either of their stat lines. Gonzalez was exceptionally effective, and he did that while facing more RHB than LHB and while sometimes taking on more than 1 IP per outing (he pitched 20.2 IP in 15 appearances).

You’re just grousing to grouse because you don’t like modern pitching strategy, but there’s ample evidence to support every decision the Dodgers made in this game.