I’ve Never Seen Anything Like It! Unique Pitching Lines Come in All Shapes and Sizes

Jordan Montgomery put together a solid outing on Wednesday night. In 6.1 innings of work, he struck out six Rays and walked only two. He did get tagged for five hits, but avoided allowing any home runs, which made the whole package work admirably. He gave up three runs, but with a little defensive prowess, things could have gone even better; two of those three were unearned.

That kind of game happens all the time these days. On the other hand, that particular game has never happened before. That exact box score line — 6.1 innings pitched, six strikeouts, two walks, five hits, no homers, one earned run and three total runs — had never occurred in the more than 380,000 starts since 1913, the first year where earned runs were recorded, as James Smyth pointed out:

? Unique Box Score Line Alert ?

Jordan Montgomery 6.1 IP, 5 H, 3 R, 1 ER, 2 BB, 6 K

A first out of more than 1.1 million MLB pitching lines in the regular season or postseason since 1901 (thanks @baseball_ref @Stathead)

— James Smyth (@JamesSmyth621) June 3, 2021

I’ll level with you: I had a hard time believing Smyth at first. That line is so middle-of-the-road. Everything about it feels like a common enough occurrence. There are no truly strange parts in that score, nothing that stands out as an obviously rare feat. An easy example: Carlos Martínez also recorded a unique line on Wednesday. His was altogether stranger: 0.2 innings pitched, one strikeout, four walks, and 10 earned runs without a homer or an unearned run. That just sounds like an unprecedented start.

One thing I could do is shrug and say “Yep, Montgomery’s line isn’t very interesting.” I wanted to do more than that though, so I decided to put some numbers to it. First, I enlisted the help of our own Lucas Kelly to pull out a box score line for every start made in the major leagues since 1913, the first year where earned runs were recorded. Yeah, we’re going pretty far back on this one.

That gave me more than 106,000 unique lines, defining a line as IP, K, BB, H, HR, R, and ER. From there, I needed to come up with some way to define what’s unusual among these one-of-a-kind lines. It’s clear that pitchers produce Martínez-esque lines less frequently than Montgomery-style ones, but I wanted a quantitative test. How else could we compare Montgomery to another middle-of-the road start? As an example, what about Tim Hudson’s May 15, 2005 effort? He went six innings and gave up only two runs, one of which was earned. He did it despite plenty of traffic; he surrendered nine hits and two walks, and only struck out five.

To judge component-by-component rarity, I looked for how often a particular result had happened in my full sample of 380,000 starts. For example, starters have allowed exactly three earned runs 70,867 times. I did that for every statistic in the line score. That gave me seven observations for each start.

Next, I added in rate statistics for each category other than innings pitched (naturally enough). I made a slight modification here to make things more interesting, given that this is hardly rigorous analysis. Instead of an exact inning count, I counted how many innings the pitcher finished, unless they didn’t finish any. Why? Otherwise going a prime number of outs (4.1 and 5.2 are 13 and 17 outs, for example) would naturally mean those values would occur far fewer times.

Is that some scientific necessity, or even an obvious choice to make? It sure isn’t, but it felt necessary to keep all the most unusual lines from being simply a case of a pitcher throwing a prime number of outs, making their achievement appear artificially scarce.

Why include rate stats at all? Because they truly do capture something about a game. A pitcher who records six strikeouts and allows five earned runs isn’t particularly rare, but if they do it in two innings, that’s really something. By looking at both the absolute statistics achieved and the rate at which they were achieved, I hope I’m capturing a happy medium of the two.

With the categories set, I then took the inverse of each number and added them all up. In other words, if a particular number had only come up once — a theoretical 21 strikeouts, say, or 22 hits allowed, oof — that added 1 to the total for 1/1. If something happened 10 times, it added 0.1. If something has happened 70,867 times, like allowing three earned runs, it added 1/70,867, essentially zero. In this way, extremely rare occurrences move the needle, and commonplace results are worth nothing.

Next, the answer you’re dying for: how individually strange (stat-by-stat) was Montgomery’s start? It was tremendously mundane! It was the 103,080th-rarest unique start — I’m struggling for ways to avoid saying unique unique start, when what I really mean is that it has rare components — out of 106,245 starts. That’s the third percentile. Montgomery had one of the most mundane unique box score lines possible.

Martínez, on the other hand, had a wild one. His start checks in as the 879th most interesting in the entire sample set. That’s the 99th percentile, naturally enough. You don’t see starts like that every day, or indeed every year. Perhaps even more shocking, Jeremy Guthrie accomplished almost the exact same feat in 2017, when he gave up 10 runs of his own, also in two-thirds of an inning. The only difference? He also walked four batters and gave up six hits with no homers. The only other difference? Martínez tallied a strikeout, while Guthrie didn’t even accomplish that.

The real point of all this is — well, there’s basically no real point in all of this. But while we’re here, I’m dying to know what the weirdest start of all time was. It’s by a Hall of Famer, because of course it’s by a Hall of Famer; that’s just how baseball works. On May 24, 1918, I hope Stan Coveleski had a hearty breakfast, because he put in a full, and I do mean full, day of work. He went a solid 19 innings to capture the win against the Yankees, 3-2. He only struck out four in his 19 innings of work, and amazingly gave up a home run and allowed more than a baserunner per inning despite only allowing two runs. Just about every ratio statistic in his line was unique, because 19 innings holy moly.

The top of the list is dotted with lovely lines. Here are the top 10, for your viewing enjoyment:

Unsurprisingly, the weirdest starts are dotted with pitchers going an ungodly amount of innings in the dark ages of the game. The highest-placed game that occurred after 2000 was an Edinson Vólquez start from 2016, and it wasn’t pretty; 12 runs, 11 of them earned, in only one inning of work. He did, at least, limit the damage to a single home run.

On the other end of the spectrum, here are the 10 unique games whose individual components are the most common:

Six innings, no home runs, and unexciting strikeout, walk, and hit numbers seems like the standard here, a standard Montgomery mostly matched. If he’d recorded one fewer out, he might have been in the running. Instead, he’ll have to settle for throwing a wildly uninteresting unique game, rather than a truly mundane one.

Do unique starting pitcher box score lines matter? I mean, no. No, they do not. Baseball loves its numbers and statistics, though, and I enjoy the fact that unique games come in such a wide array of shapes and sizes. Don’t forget Stan Coveleski’s unicorn — but don’t forget Bruce Hurst’s, either. And hey, while you’re at it, check out the inimitable Sam Miller, who was writing about the abundance of unique pitching lines in 2012. It turns out this article, like so many starts in baseball’s history, is itself not unique.