The Plan Worked for Seven Innings

Ninety-five pitches, one run. Fifteen pitches, two runs.

The first line belongs to Tyler Mahle, who was not certain he would pitch at all. The second belongs to Dylan Lee, who has been the steadiest arm in Atlanta's bullpen all season. Between them sits a 4-3 loss in ten innings, a Wild Card Series tied at one, and a season that now fits inside a single Thursday night.

Philadelphia 4, Atlanta 3. Game 3 is at eight o'clock at Truist Park, and whoever loses it goes home.

The argument held

This morning I argued that Mahle's last five starts told you more about this game than Cristopher Sánchez's postseason record did. Mahle had never thrown a postseason pitch. Sánchez had thrown 19.1 of them. The recent form pointed the other way — 2.35 against 5.12 — and I said to trust the recent form.

That argument was correct, and it is worth being precise about how correct.

Mahle went seven innings and allowed one run on six hits, with one walk and six strikeouts, on 95 pitches. Sánchez needed 116 pitches — a career high — to get through six and two-thirds, and he gave up three runs and walked five. The pitcher with no October innings was the better pitcher in October. The pitcher with the résumé threw the most pitches of his life and left trailing.

Atlanta led 3-1 after seven. Then the game went to the bullpens, and everything I know about reading a pitcher stopped working.

What the same method said about the eighth

Here is Dylan Lee's season: 6-1, a 3.09 ERA, 77 strikeouts against 17 walks in 67 innings, and a 0.97 WHIP. Fewer than one baserunner an inning across a full year. He was the winning pitcher in Game 1. He had retired the same part of the Philadelphia order the day before.

Walt Weiss was asked about going to him and did not hedge.

"Lee is our high-leverage lefty," he said. "[He] put down the same part of the order yesterday pretty easily. Threw the ball really well yesterday. So no, that's his inning every time."

By the method I used on Mahle this morning, that is the right answer. Recent form, strong season, correct matchup.

Kyle Schwarber hit Lee's 3-2 slider 447 feet. Bryce Harper hit the next one he saw 418 feet. Fifteen pitches, two swings, 3-3.

Now take the same method into the other dugout. Andrew Painter is a rookie. He went 3-10 this season with a 5.40 ERA and a 1.41 WHIP. There is no reading of that line that recommends him for three innings of a tied game his team had to win. He threw three scoreless innings, struck out three, and won it.

So the method got Mahle exactly right and got Lee and Painter exactly backwards, in the same game, in the innings that decided it.

I do not think that makes the method wrong. I think it makes the sample honest. A season line is a claim about a whole year of innings. The eighth inning of a playoff game is not a sample — it is an event. Lee's 0.97 WHIP was a true description of 67 innings and a useless description of fifteen pitches. Painter's 5.40 was the same thing in reverse.

This is, statistically speaking, not a comfortable thing to write eight hours after publishing the opposite.

The number that actually explains the loss

Set the pitching aside.

Atlanta had nine hits, six walks and eleven men left on base. Philadelphia had twelve hits, one walk and nine men left on base. Atlanta put more men on. Atlanta drew six times as many walks. Atlanta lost by a run.

The difference is that Philadelphia hit three home runs and Atlanta hit none. All three of Philadelphia's changed the score. Atlanta's three runs came on a single, a double that glanced off a right fielder's glove, and a sacrifice fly.

There is a tidy version of this where the top of the order is the problem, and the box score will not support it. On Tuesday Atlanta's top four went 1-for-15 and won, and I wrote that it was not a formula the club could repeat. Wednesday the top four went 4-for-16 and drove in all three runs. The order did its job. The club still scored three.

That is not a bad night. That is the shape of this roster in a close game. A lineup that gets on base and moves runners one station at a time needs more innings than a lineup that can erase a two-run deficit with two swings. In a best-of-three, nobody gets more innings.

Bryce Harper's misplay deserves a sentence, because it is most of the reason Atlanta led at all. Drake Baldwin's drive to right glanced off Harper's glove with runners on first and second, scoring Sean Murphy. Harper was not charged with an error. "Obviously a ball I should have caught," Harper said. An inning later he hit the ball that tied the game. October does not keep a ledger.

The part nobody had before first pitch

Mahle nearly did not pitch.

"We had a scare before the game," Weiss said. "He felt an abdominal pull. We didn't know if he was going to make the start."

The timing was as late as it gets. "We weren't sure. It was right at game time after he came out of the bullpen," Weiss said. "He ended up working through it, and it got better as the game went on. But we were scrambling early on."

A pitcher with no postseason innings, an abdominal pull in his warmup, and a club scrambling an hour before first pitch — who then threw seven innings and allowed one run. Whatever else Thursday brings, that happened.

ESPN lists him day-to-day. Nobody at the club has said what that means past Thursday, and there is no version of this where guessing is useful.

Baseball has a long memory, and so does the manager

Asked about Lee, Weiss did not reach for a chart. He reached for 1988.

"When you're a late-inning reliever, these things happen," he said. "I've seen some great ones that this has happened to. I saw one of the greatest. I was at shortstop when Kirk Gibson did his thing off Dennis Eckersley, one of the greatest closers of all time. It comes with the territory when you're pitching late in close games. Every once in a while, they get you."

A manager answering a question about his own reliever by describing a home run he says he watched from shortstop thirty-eight years ago is the most useful thing said in that room. Not because it absolves anyone. Because of the part Weiss added himself: he called Eckersley one of the greatest closers of all time. That is the whole argument. Being the best arm available has never been a defense against one swing.

What Thursday is and is not

Three days ago I wrote that Atlanta's 0-3 record against Philadelphia in October — the 1993 NLCS, the 2022 and 2023 Division Series — told you nothing about this series. I still believe that, and I am not going to start believing the opposite because the series got harder. A fourth loss would not be a pattern completing itself. It would be a fourth loss.

What is true is narrower and worse. Atlanta has one game left, at home, and across two games it has outscored Philadelphia by exactly one run.

Grant Holmes is expected to pitch. Whether he starts or works in bulk relief has not been settled, which is its own quiet statement about the 75 starts Atlanta already replaced this season. Aaron Nola is lined up for Philadelphia, 7-12 with a 4.67 ERA, a number that by this afternoon's evidence predicts approximately nothing.

The last pitch of the game was a fastball to Dominic Smith that plate umpire Adam Beck called inside, that Painter and J.T. Realmuto challenged on the spot, and that the automated zone confirmed had clipped the corner. Strike three, tying run on second, winning run on first. It is the first postseason game ever to end on a strike-zone appeal, which is both a genuine first and a slightly cheap one — this is the first season the system exists, so there was no earlier October in which it could have happened.

There is no grievance in it. The pitch was a strike.

There is just one night left, and a lineup that will have to hit the ball over a fence.