The sainted Mac Thomason defined an episode thusly:
When Tim Hudson suddenly loses the ability to get anyone out, to the degree that it is surprising he has full control of his limbs and his bowels, he is having an Episode. Usually in the sixth or seventh inning.
Now we all know that Tim Hudson was a good pitcher, so the Episode is intended as an aberration — you can’t say that a pitcher like the 2008 Elmer Dessens had episodes. He gave up 10 runs in 4 innings pitched, but that’s just who he was.
Watching the Braves makeshift staff this season, there seem to be a lot of pitchers who give you a few pretty good innings but then yield a crooked number. Is this sort of pitcher worse than a pitcher who gives up the same number of runs per inning, but whose path is steadier? Is it worse to give up a run an inning for fice innings or one complete meltdown of a fifth inning after four shutout innings?
I want to start by saying that from the fan’s viewpoint, episodes are particularly disheartening for a host of reasons: (a) you thought this pitcher “had it” today, and he didn’t; (b) as long as an episode is a possibility, no lead is safe; (c) variance in results lead fans to question their own perspicacity.
But does it really make a difference? I made a little study of starting pitchers to get a first look at this. (Relievers don’t have episodes… they just occasionally suck.)
I created an Episode Index for starting pitchers. This has a fair number of steps which most of you won’t care about. For the three others, here’s what I did:
I took every pitcher with 10 starts or more in a season. (I used the seasons 2021-2024.) This gave me 716 pitcher-seasons. I then looked at every complete inning pitched by these pitchers and kept track of runs allowed. (I didn’t use earned runs, but the results don’t change very much when I do.) From this frequency chart of runs allowed by a pitcher by inning, I calculate the standard deviation. Thus, to take an example, here’s Max Fried, 2021:
| Runs Allowed | No Runs | 1 Run | 2 Runs | 3 Runs | 4 Runs | Sum |
| # of Innings | 147 Innings | 34 Innings | 6 Innings | 2 Innings | 3 Innings | 192 Innings |
| f x X | 0 | 34 | 12 | 6 | 12 | 64 |
| f x X squared | 0 | 34 | 24 | 18 | 48 | 124 |
Std = sqrt(124/192-(64/192)^2) = 0.7312The std represents a deviation of runs allowed, but this alone cannot serve as a measure of Episodes, because it is highly correlated with ERA. Episodes are the extra bad innings you have holding your ERA constant. To get the Episode Index, we then create a linear regression predicting Std.
Using the regression line, the Episode Index is just the Actual Std vivided by the Expected Std from the linear regression times 100. The Indices range from a high of 136 (Adrian Houser, 2022) to a low of 56 (John Brebbia, 2022).
But now let’s look at pitchers who are pretty good pitchers and see if we can draw any judgments holding ERA constant. Here is the database of pitchers with (adjusted) ERAs between 2.95 and 3.05, ordered by Episode Index.
| id | Year | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | ERA | Starts | Episode Index |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Framber Valdez | 2024 | 149 | 14 | 4 | 4 | 1 | 1 | 0 | 1 | 0 | 2.99 | 29 | 126 |
| Kodai Senga | 2023 | 127 | 22 | 8 | 3 | 2 | 0 | 0 | 0 | 0 | 2.98 | 29 | 106 |
| Marcus Stroman | 2021 | 136 | 22 | 13 | 5 | 0 | 0 | 0 | 0 | 0 | 3.02 | 33 | 103 |
| David Peterson | 2024 | 94 | 17 | 7 | 0 | 2 | 0 | 0 | 0 | 0 | 2.96 | 22 | 102 |
| Adam Wainwright | 2021 | 172 | 18 | 15 | 3 | 1 | 0 | 0 | 0 | 0 | 3.02 | 33 | 97 |
| Tanner Bibee | 2023 | 108 | 21 | 6 | 2 | 1 | 0 | 0 | 0 | 0 | 2.98 | 25 | 96 |
| Cody Bradford | 2024 | 57 | 9 | 3 | 2 | 0 | 0 | 0 | 0 | 0 | 3.03 | 13 | 94 |
| Tobias Myers | 2024 | 105 | 19 | 9 | 2 | 0 | 0 | 0 | 0 | 0 | 3.00 | 26 | 92 |
| Cole Ragans | 2024 | 159 | 19 | 10 | 4 | 0 | 0 | 0 | 0 | 0 | 3.03 | 34 | 90 |
| Seth Lugo | 2024 | 175 | 20 | 13 | 3 | 0 | 0 | 0 | 0 | 0 | 3.00 | 35 | 88 |
| Lance McCullers | 2021 | 137 | 18 | 11 | 2 | 0 | 0 | 0 | 0 | 0 | 3.02 | 30 | 88 |
| Edward Cabrera | 2022 | 59 | 6 | 3 | 0 | 0 | 0 | 0 | 0 | 0 | 3.01 | 14 | 67 |
This is a set of closely matched pitchers by ERA, but at the top and bottom we have two pitchers who pitch for the same team: Framber Valdez, whose 2024 season had some truly appalling innings, and Lance McCullers, who had only two worse-than-two-run innings in all of 2021. And guess what? Both of them finished 7th in Cy Young voting in their respective years. Their WAR totals? 4.4 for Valdez and 3,3 for McCullers. If anything, the “episode guy” looks better.
You make up for a few a few blowup innings with extra zero and one run innings and I think the episodes get forgotten. You can’t lose more than one game with an episode, but you can lose a lot of games with a lot of ill-timed single runs.
There’s more I think I can do with this, but that’s about all the energy I have for now. Are “episode” pitchers more homer-prone? More walk-prone? Does the Episode Index help explain any observed difference between ERA and FIP? Who is the most Episodic pitcher of all time? Are pitchers with a lot of Episodes in one year likely to higher Episode Indices in other years? (That’s the best sign that it’s a feature and not just luck.) Was Tim Hudson particularly episodic? What the hell happened to Foltynewicz in Game 5 in 2019. Is there something worse than an Episode?
As always for these ramblings, comments welcome.
What I always wonder with these questions is how much is just randomness that we see a pattern in. I would expect that pitch-to-contact guys like Hudson are more prone to episodes because BABIP is a lottery. I remember prime Venters being almost unhittable, except for the fact that his sinker action turned most contact into a swinging bunt. Every now and then, opponents would luckbox into a bases loaded situation.
As a stats guy, JF, how do we tell whether some perceived phenomenon is just randomness vs an actual thing?
The answer to that was a major piece of my working life. Humans are (at least for the moment) the greatest pattern-matching devices known. This facility translates into a persistent and replicable proclivity to see patterns where none exist. Humans are terrible at spotting randomness, and actively resist seeing it even when they are shown that an underlying randomness would produce identical results.
Roger,
Harris’ second half adjustments seem to be largely due to lowering his hands to where they were before, making him shorter to the ball. Therefore he has more time to decide if the curveballs in the dirt are indeed curveballs in the dirt. And I think when pitchers figured out that he would sell out early for that pitch, he would get it a lot. I can only assume that he had his hands higher to sell out for more power, which further makes me wonder if young, 24-year old Michael Harris knows what kind of player he wants to be. He’s putting on weight, he’s losing weight, he’s hitting for power, he’s now getting shorter to the ball and hitting more line drives. He seems like a young player still figuring out his game.
Is he so slow to adjust that he has to have a bad first half?
This was the same adjustment that Seitzer suggested that turned Harris into a rookie sensation—lowering the hands. Harris obviously fancies himself a home run hitter, but what he needs to do is just be the same guy he has been for 3 previous seasons. The single most important thing he could do to make himself more valuable is take more pitches. He is so good at getting bat to ball when he is going well that he doesn’t have to worry about deep counts.
Well, only FORMER Braves make ESPN’s top 50 players list: https://www.espn.com/mlb/story/_/id/45947026/mlb-rank-2025-season-update-top-50-baseball-players-judge-ohtani-skubal-raleigh
Ouch.
Listening to Mets and Yankees panic is so soothing here in mid-August.
Good job, AA. You signed a player who compels your potential Rookie of the Year to do something stupid.
What was that?
So, uh, Spencer Strider gets a $16 million raise next season lol
It’s OK. Austin Cox will make less.
Recap is up:
Recap is up: