Pebble Hunting: Would Pedro Martinez Have Gone Undefeated?

Here’s the thing about pitcher wins: It’s not that they don’t have value in this sport. It’s that their value in this sport is limited to one very specific, very rare situation. That situation is not “what should we say about this Anthony Raunado performance, in one word or fewer?”, and it’s not “how do we decide who’s better, Fernando Abad or Christian Bergman.” But, then, the pitcher win shouldn’t aspire to be such a stat. The stat that exists to answer those questions lives the boring life of an accounting operations manager at a third-tier gas station franchise. The stat that exists to answer those questions lives steeped in the tepid banality of everyday. The pitcher win is limited and stunted and has a funny voice and it exists in case of just one scenario, and that scenario, if it happens, will be of the utmost importance. The pitcher win is basically Owen Meany.

That scenario almost happened, and we’re going to see if the pitcher win would have been up to the task if it had.

In 1997, when the Expos were shopping Pedro Martinez around, it wasn’t the Red Sox who had the first bite at him; it was the Indians. Multiple writers relate this, including Peter Gammons (writing about the Indians passing up Brian Giles for three months of Randy Johnson!), Tony Massarotti (in his book Dynasty: The Inside Story of How the Red Sox Became a Baseball Powerhouse), and Danny Gallagher and Bill Young, in Ecstasy to Agony: The 1994 Montreal Expos:

“We were looking for good, young pitchers not yet arbitration eligible. We tried to get Jaret Wright in a package with Cleveland for Pedro but they wouldn’t even talk to us.”

You could take this what-if any direction you want: What if the Indians had Pedro in the next two Octobers instead of Wright, who allowed 17 runs in 13 postseason innings after the non-trade? What if the Red Sox had Carl Pavano instead of Pedro? What if Pedro had hit free agency after a single season in Cleveland and landed with the Yankees, or any other team you might imagine? But the far better what-if is this one:

What if Pedro had spent a year supported by that Indians offense? That Indians offense, after all—the 1999 group specifically, with Thome and Manny and Lofton and Alomar and Justice; a lineup that produced an OPS higher than .790 at each of the top seven spots in the lineup, that hit .309/.388/.513 with men on base—was perhaps the best of our lifetimes. That 1999 group scored 170 more runs than the Red Sox; indeed, that 1999 group is the only offense to score more than 1,000 runs in a season since 1950. And that 1999 Pedro had the fifth-best ERA+ since 1950, the third-best in a non-strike-shortened season. He had the lowest FIP- in history. Is it possible that he would have gone undefeated?

We’ll answer that, after the caveat: I’m going to lay Pedro’s pitching line in each start over the Indians’ offense in the same game (e.g. Red Sox game one —> Indians game one), and duhhhhh I know this doesn’t actually work. I know that if Pedro Martinez had started Opening Day of the 1999 season in Cleveland, the Indians wouldn’t have scored exactly five runs in the first six innings (as they did without Pedro), and that Pedro wouldn’t have pitched exactly six innings and allowed exactly two runs (as he did for Boston). I also know that we’re mixing park factors and so on, but I don’t want to deal with fractions of runs so we’ll ignore that. You can think about all this throughout the rest of this post if you like and I’ll understand; you’ll have more fun if you don’t, though. Just roll with it. You focus too much on butterflies flapping and you’ll convince yourself each flap is the one that’ll give you cancer.

So here we go, 1999. We’ll break the outcomes down by game temperament.

The dominant starts with tons of offense.
No suspense in these.

• May 7th (team game no. 28): Leaves after eight, up 20-0.
• July 2nd (team game no. 79): Leaves in the middle of the eighth, up 9-1.
• August 24th (team game no. 126): Leaves in the middle of the ninth, up 12-1.
• September 4th (team game no. 136): Leaves in the middle of the ninth, up 15-0.

These include the fourth, fifth and sixth-best starts of the season for Pedro, by Game Score, and they would have been thrown in games decided, basically, by the second inning. In the fourth one, incidentally—the July 2nd start—Pedro threw eight innings, allowed only one run, struck out five, and walked nobody. His game score for that start, 69, was actually lower than his median performance in 1999. His average start was more dominant than 8/7/1/1/0/5.

Pedro is 4-0.

The quality starts with tons of offense.

• April 10th (team game no. 5): Leaves mid-eighth, up 12-3.
• April 15th (team game no. 8): Leaves after the seventh, up 11-3.
• May 18th (team game no. 38): Leaves mid-eighth, up 13-2.
• June 15th (team game no. 63): Leaves after the eighth, up 10-2.
• July 7th (team game no. 84): Leaves mid-eighth, up 9-3.
• September 27th (team game no. 156): Leaves mid-ninth, up 11-2.

These mark the last of the starts in which he got 10 or more runs of support while he was in the game. Most pitchers would be undefeated in these 10 starts, but give Pedro credit for always being good enough to pitch through five and get the win—plenty of pitchers might have gone 8-0 or so, but he didn’t leave any on the table here. Also: In more than a third of starts the Indians would have provided him 10 or more runs of support! That offense, man.

Pedro is 10-0.

The shutouts.
No suspense here, either. Throw a shutout, get a win. Pedro threw only one shutout in 1999, interestingly enough:

• September 27th (team game no. 157), wins 3-0.

Pedro: 11-0.

The complete games.

• April 25th (team game no. 18): Wins 4-2.
• June 4th (team game no. 53): Wins 5-1.
• September 10th (team game no. 141): Wins 4-1.

Struck out 17 Yankees and walked none in that last one. There are no real what-ifs in any of these three, so it’s safe to call them all wins.

Pedro is 14-0.

The abbreviated shutouts.

• April 20th (team game no. 13). Pulled in the middle of the top of the eighth, up 5-0.

The Indians would have had about a 99.5 percent chance of winning this one, so even with the addition of the bullpen into the equation we’ll call it:

Pedro is 15-0.

Very safe leads handed over to the bullpen.

• May 1st (team game no. 23): Leaves mid-eighth up 5-1. Indians 99.2 percent likely to win.
• May 12th (team game no. 33): Leaves after the eighth up 6-2. Indians 98.9 percent to win.
• June 26th (team game no. 74): Leaves mid-sixth up 6-1. Indians 97.5 percent to win.

That last one, when he left after five—in real life, the Red Sox were cruising to a 17-1 win. Pedro threw only 59 pitches. In the Indians scenario, he probably goes a bit deeper and leaves with an even safer W. But all of these are safe.

Pedro is 18-0.

Pretty safe leads handed over to the bullpen.

• April 5th (team game no. 1): Leaves mid-seventh, up 5-2. Indians 88 percent to win.
• June 20th (team game no. 68): Leaves mid-eighth, up 3-1. Indians 95.5 percent to win.
• August 19th (team game no. 121): Leaves mid-seventh, up 6-3. Indians 95.8 percent to win.
• August 30th (team game no. 131): Leaves after six, up 4-1. Indians 90.4 percent to win.

At some point we’ll have to dock him a win, out of recognition that bullpens do blow things, but a) it still wouldn’t spoil his perfect record and b) given the win expectancies he bequeathed, his team would have been about 69 percent likely to win all of these games.

Pedro is 22-0.

Pedro had to grind it out.

• June 9th (team game no. 58): Leaves mid-7th, up 6-4.

It’s going to be one of three things that does him in, obviously. He has a bad start and loses. His team gets shut down and, despite him pitching well, he loses. Or he’s just normal and the Indians are just normal and he loses. This was the biggest risk of the latter category, as they didn’t score a ton of runs and he had his second-worst outing of the season, but he still left with his team 78 percent likely to win. In real life, the Indians bullpen that year gave up two runs in the seventh inning of game no. 58, which would have tied this contest, which would have cost Pedro the decision. So we’ll leave him at 22-0, barely.

The no run support games.
In all of the following, he got three or fewer runs of support while he was in the game.

• May 23rd (team game no. 43): Leaves after six, down 3-2.
• May 29th (team game no. 48): Leaves after eighth, tied 2-2.
• Aug. 3rd (team game no. 106): Leaves mid-sixth, up 2-1.
• Aug. 8th (team game no. 111): Leaves mid-sixth, up 3-2.
• Sept. 15th (team game no. 146): Leaves after seven, tied 2-2.

So we have two games he’ll certainly take no-decisions in, because he left tied. Two games he left leading, with the Indians 59 percent likely to win each of them—call it a win and a no-decision. And, finally, we have one where he might lose.

A little context on that start, which came (in real life) against the Blue Jays. Pedro threw six innings and allowed three runs, two of them earned, and was lifted after throwing 95 pitches. In the previous six starts, he had thrown 127, 135, 125, 125, 126 and 137 pitches, from which we can deduce that a) he was certainly capable of, and used to, throwing far more than 95 pitches, and b) he was probably given an easy start because he had been worked hard leading up to that game. The Red Sox were winning 7-3. They had a well-rested bullpen (Wakefield had gone 8 â…“ the day before) and an off-day followed, so with the game in hand, Pedro certain to get the victory, and some benefit to getting him out a little early it was a fairly easy call.

But put the same performance in the Indians schedule, and things are different. Now he’s trailing, and—importantly—he’s undefeated in late May, which would probably have been considered at least a little. The Indians bullpen was not particularly well rested, and they didn’t have an off-day following. So probably Pedro goes out for the seventh. Maybe the eighth. Maybe he allows more runs, or maybe he shuts the Blue Jays down for another two innings.

This matters, because the Indians bullpen allowed three runs in the seventh but their offense scored one in the seventh. If Pedro had been pulled after six, like he was for the Red Sox, the loss would have stuck—it would have been 3-2 when he left, and 6-3 after seven, and the lead would have held from there. But if Pedro had been in the game, had pitched one more scoreless inning, he would have been in position to leave the game tied after seven. He would have been in position to get a no-decision.

What do we do with this? I don’t know. I guess we choose the answer we want. Pedro is 23-0.

But we have one more bullet to dodge:

The terrible start.

• July 18th (team game no. 92): Leaves in the middle of the bottom of the fourth inning, down 9-0.

There’s no way to salvage this one. If you’re desperate, you might note that he was BABIP’d to death, that he didn’t walk anybody or allow a home run but managed to give up 13 hits, plus two batters reaching on errors, on just 20 balls in play. You might make up an escape clause involving excellent defenders Omar Vizquel and Roberto Alomar instead of hacks Nomar Garciaparra and Jose Offerman. But we’re not those kinds of people, are we? We know when we’re beat. Or at least we know when hypothetical alternate universe Pedro Martinez is beat.

Pedro is 23-1.

Ironically, Pedro did not lose this start in real life. The Red Sox won 11-9. Derek Lowe got the win.

***

Epilogue: What about 2000? Not quite as good an Indians offense, but an even better Pedro.

2000

• Start 1: Leaves mid-8 up 4-0 (7 IP, 0 R)
• Start 2: Leaves mid-bottom-8, up 17-1 (7 1/3, 1 R)
• Start 3: Leaves end-7 up 4-2 (7 IP, 2 R)
• Start 4: Leaves mid-8 up 4-3 (7 IP, 3 R)
• Start 5: Leaves end-7 tied 0-0 (7 IP, 0 R)**
• Start 6: Complete game, wins 10-1
• Start 7: Complete game, wins 16-0
• Start 8: Leaves end-7 up 7-0 (7 IP, 0 R)
• Start 9: Leaves mid-9 up 4-3 (8 IP, 3 R)
• Start 10: Complete game, wins 12-0
• Start 11: Leaves mid-9 tied 0-0 (8 IP, 0 R)**
• Start 12: Leaves end-6 tied 1-1 (6 IP, 1 R)
• Start 13: Complete game, loses 3-1
• Start 14: Leaves mid-top-7, up 8-5 (6 2/3 IP, 5 R)
• Start 15: Leaves mid-8 up 5-2 (7 IP, 2 R)
• Start 16: Leaves end 8, tied 1-1 (8 IP, 1 R)
• Start 17: Complete game, wins 1-0
• Start 18: Leaves mid-8, tied 1-1 (7 IP, 1 R)
• Start 19: Complete game, wins 5-2
• Start 20: Leaves end-8, tied 2-2 (8 IP, 1 R)
• Start 21: Leaves mid-5, up 6-3 (4 IP, 3 R)
• Start 22: Leaves end-7, up 10-0 (7 IP, 0 R)
• Start 23: Leaves mid-9 down 6-1 (8 IP, 6 R)
• Start 24: Complete game, wins 5-0
• Start 25: Leaves end-8 up 5-1 (8 IP, 1 R)
• Start 26: Leaves end-7 up 9-3 (7 IP, 3 R)
• Start 27: Leaves mid-8 down 3-1 (7 IP, 3 R)
• Start 28: Leaves mid-9 up 8-1 (8 IP, 1 R)
• Start 29: Leaves end-5 up 2-0 (5 IP, 0 R)

The two starts with asterisks: Those look like no-decisions, but we can’t really count them as such because the real-life Indians were batting against Pedro. In our scenario, they wouldn’t have been, so those are likely wins. But it’s still not enough to get Pedro to a perfect season, though. Most likely he’s around 20-3. Better than his actual record that year—18-6—but who cares. I don’t care. Pitcher wins are stupid.

Many thanks to Jim Walsh for inspiring this inquiry.