It won’t be remembered this way, but last night’s Braves/Phillies Game 2 clash provides an interesting bookend to the interminable Blake Snell discussion we’ve been having every October since the moment it happened in 2020. Let’s set the scene: Zack Wheeler looked absolutely dominant to start the night, bowling the Braves over to the tune of five no-hit innings, with an error the only blemish on his pitching line. He started to wobble in the sixth, with a walk and a single leading to an unearned run. The Phillies led 4-1, and Rob Thomson had the bullpen working overtime, but Wheeler struck out Austin Riley to end the threat and keep the bullpen at bay.

Clearly, the Phillies were considering going to a reliever, and you can understand why. They showed a ton of trust in their bullpen in the first game against Atlanta, and the ‘pen delivered: 5.1 scoreless innings fueled a 3-0 victory. After an off day, the gang was rested, and today is another off day, which meant there would be more time to recover, particularly considering there were only three innings to cover.

On the other hand, it’s pretty easy to understand why Thomson left Wheeler in. He’s Zack Wheeler! He’s been one of the best pitchers in baseball for half a decade now. He was rolling; the first 18 Braves hitters managed legitimately nothing against him, and Atlanta is a historically great offensive team. Let your aces eat, the thinking goes.

We all know what happened next: Wheeler gave up a single, recorded a strikeout, and then surrendered a two-run home run to Travis d’Arnaud that tightened the score to 4-3 and sent Wheeler packing. Riley hit a go-ahead home run the next inning, and the Braves held on to win after one of the craziest endings to a playoff game I’ve ever seen. Clearly, none of that was preordained when Wheeler came back out for the seventh. But what were the tradeoffs involved in sticking with him?

First and by far most important, there’s the times through the order penalty. This isn’t how it always works, but consider last night’s game. The first 18 batters Wheeler faced accrued a wOBA of zero. The only baserunner reached via an error, and while some wOBA formulations give the offense credit for that, it’s more fun for this example to zero it out. The next six Atlanta hitters racked up a .744 wOBA. If you’re not familiar with the wOBA scale, the best single-season mark since baseball was integrated is Barry Bonds’ outrageous 2002 season (.370/.582/.799), which works out to a .544 wOBA. In other words, the Braves turned into titans.

That’s not how Wheeler’s average third time through goes, obviously. But he’s historically struggled when facing a lineup the third time, in the same way that all pitchers do. He allows a .262 wOBA the first time through, .281 the second time, and .328 the third time. Even if you only look at his numbers since joining the Phillies, he checks in at .237/.265/.318. Limit it to only the top four hitters in the batting order, since the average competition the third time through is tougher (you face the bottom of the order less frequently), and it’s .245/.299/.361.

If you’re curious, the Phillies relievers who have pitched in this series are pretty good, wOBA-wise:

For his part, Wheeler allowed a .277 wOBA overall in 2023. You can think of him as delivering Craig Kimbrel-like performance over a starter’s workload, which is phenomenally impressive. Kimbrel might one day be in the Hall of Fame because he’s so good at pitching one-inning stints, and Wheeler is doing it over whole games. But he gets worse as the game goes on – he’s been far better than Kimbrel in his first inning of work, and worse by the time he’s laboring with the lineup for a third time.

There’s nothing wrong or weird about that – that’s how baseball has gone since time immemorial. John Smoltz, the current king of decrying the times through the order penalty, allowed an OPS of .688, .719, and finally .770 as he faced an order multiple times. It’s not a fault of modern pitching. It’s not a matter of grit or want-to. It’s just harder to pitch to batters who have seen your offerings more times.

That’s the argument for pulling Wheeler: for one batter or one inning, in this context, he hasn’t been as good as the relievers the team could have used instead. That’s not revisionist history, it doesn’t include last night’s game; it’s just tougher to face a lineup a third time through. He’s a better pitcher than the bullpen, but his task was also more difficult, and I think that the first-order math supports that conclusion quite clearly.

Of course, we don’t play baseball by lining up first-order math arguments and then declaring a winner. There are plenty of other factors at play, and I think a lot of those support leaving Wheeler in. First, there’s a “this specific night” argument. Sure, those statistics are hard to argue with in the aggregate. But Wheeler was cruising! He hadn’t worked very hard. Surely, that makes things easier for him.

Eh… try as they might, baseball analysts haven’t been able to figure out whether these things matter. Russell Carleton is all over it. If you only have time to read one link, read the last one – even cruising pitchers and ace pitchers underperform late-inning relievers. Ethan Moore laid out a good theory of why, but the why matters less than the what, and the what is that relievers are the better option barring extenuating circumstances.

This stinks. It’s no fun that baseball is this way. But as best as a wide variety of good and careful researchers can find, you can’t hand-wave your way to leaving your starter in over a rested and qualified bullpen if all you care about is winning the game in front of you. By the time you get this deep in the game, and by the time the opposing hitters have seen your ace that many times, the ugly truth is that a reliever is probably the better option.

Okay, then, let’s talk about the bigger picture. The Phillies don’t care about winning the game; they care about winning the series. Does that change the math? I think it argues in favor of leaving Wheeler in, that’s for sure. The more times the Braves see Philadelphia relievers, the more familiar they presumably get with their offerings. Ben Lindbergh looked into a similar question – whether starters perform worse the second time they face a team in a series – and found no effect. But that isn’t quite the same, and I haven’t seen a conclusive study of this. Let’s give Rob Thomson a little credit there; maybe he was protecting his guys from overexposure.

What’s more, it was a three-run game. Per our win expectancy charts, the Braves stood only a 13.5% chance of winning the game when Wheeler stepped to the mound to start the bottom of the seventh. Maybe Thomson was willing to give up a little bit of equity in this game, which felt well in hand, to give his guys more rest in future games. More rest, less familiarity, Wheeler with only 85 pitches through six innings; if you want to steal a little rest for your bullpen, there are worse times to do it. The fact that today is a rest day argues the other way, of course, but there were still clear positives to leaving Wheeler out there.

What’s more, this decision simply didn’t matter that much. Calculating the exact change in win expectancy is a little more complicated than I want to get into here, but even if you assume the bullpen had a .265 true talent wOBA ignoring opponent quality, while Wheeler was at .318, that works out to something like an eighth of a run’s difference across the three batters he faced. Sacrificing an eighth of a run of equity – or perhaps an 8% chance of one run and a 2% chance of two runs – usually doesn’t flip the outcome. Even after d’Arnaud’s homer, the Braves only stood a 32.3% chance of winning, and that was close to the worst that could have happened over those three batters. Wheeler getting tired was hardly the end of the game.

I think that’s the best argument for leaving him in: the actual cost in win expectancy was small, and there were enough fringe benefits that Thomson was able to convince himself that the tradeoff made sense. This was hardly a select-Wheeler-therefore-lose situation; it’s all small edges.

Of course, the Blake Snell decision was all about small edges, too, and that hasn’t stopped people from treating it like Kevin Cash incinerated a World Series trophy just to stick it to old school baseball thinkers. These decisions are far less important than what the players do afterwards. Baseball is a game of execution; the edges managers can manufacture are dwarfed by what the players on the field do or don’t do on a given night.

I guess that would be my takeaway: I don’t think this decision was open and shut in either direction, and I don’t think the Snell one was either. I don’t think either “caused” a loss. I do think that this goes to show that no matter what you decide, it can backfire on you. Sometimes you pull your cruising starter and the bullpen donks away your lead. Sometimes you leave him in and he just doesn’t have it.

One last fun note on this: I was looking for signs that Wheeler was fatigued, but I ended up just being impressed by the Braves. Wheeler only threw six middle-middle pitches all day. That’s phenomenal command. The Braves swung at all six and made contact with every one. They fouled two off. They hit the other four 100 mph or harder. The last two of those were Olson’s single and d’Arnaud’s homer. Goodness gracious, is this lineup good. This series has been electric so far, and last night’s game was just another reminder that playoff strategy is great, but playoff baseball is so, so, so much better.