Does Pitch Data Support Pat Murphy's Claims About Joe Ross Late in Starts?

It wasn't the decisive moment of Monday night's loss, but it was the one that had the most Brewers fans talking afterward. With one on and one out in the sixth inning and the game tied at 1-1, Pat Murphy went to reliever Hoby Milner, lifting starter Joe Ross after just 79 pitches. Milner struggled, and a crucial error by Jake Bauers exacerbated the problem. The home nine not only brought around the runner Ross bequeathed to Milner, but scored twice more, and the game would end 4-2.

Afterward, there were questions for Murphy about taking down Ross so proactively. Early in spring training, Ross was one of the first pitchers to earn Murphy's full vote of confidence as a part of the starting rotation, and with injuries mounting, the next three days look to be fairly bullpen-heavy. Removing Ross, therefore, required the skipper to think there was significant trouble lurking just around the corner for him. Why did he feel that way?

Pat Murphy said Joe Ross had reached a point in the sixth where his stuff was starting to look flat. Here's the full back-and-forth on the decision to remove him from the game. pic.twitter.com/aDsofDpFWp

— Curt Hogg (@CyrtHogg) April 23, 2024

Murphy and the Brewers coaching staff believed they saw Ross's stuff flattening out, and with left-handed batters Rowdy Tellez and Jack Suwinski due to see him for a third time in the game, they didn't want to press their luck. Murphy knew the Pirates might counter with a pinch-hitter for Tellez if he brought in Milner (which they did), but felt that was still a preferable matchup to Ross against Tellez and Suwinski.

As he also said in the above exchange, though, part of that thinking was that the team knows Ross and his tendencies. Were they seeing something that wasn't there, because they expected it to be? And separately, is what they think they know about his stuff and the inflection points within a game when it changes for the worse true?

Happily, this has never been more testable. We can see most of what the team can see about Ross's performance, last night and across the season. Let's turn to the data.

First of all, yes, Ross's sinker was flattening out as the game moved along on Monday night. That's not unusual, especially for a pitcher his age, but it's the most important of Murphy's claims, and it's accurate. Here's his horizontal movement on the pitch throughout the start.

Though 'two-seamer' has lost the fight with 'sinker' for terminological primacy in pitching, horizontal movement is really about as important as vertical movement is. The 'sink' on a pitch that fools and frustrates a same-handed batter is just as often horizontal run as it is heavy dropping action, and Ross's pitch is more run than sink regardless of when he's throwing it within an outing. 

Now, you'll notice that he had an earlier cluster of sinkers within the outing with even less armside run than the last handful he threw, before coming out of the game. Why didn't those concern the staff as much? Partially, it's because they were thrown considerably harder.

Again, it's not at all unusual--especially for a pitcher who missed most of the last three seasons with various injuries--to lose velocity as the game wears on, but that's also one big reason why pitchers lose effectiveness within a game. Ross began dipping below 94 MPH (and even down under 93) at the end of his appearance; the line where a sinker goes from theoretically above-average to almost unavoidably below-average in the modern game is right around the 93-MPH threshold. Combine the falling velocity with the straightening-out process happening in terms of movement, and you have a whole different pitch from the first two innings to the sixth. Any of the newfangled stuff metrics on the market would grade the two very differently.

Zooming back a bit, though, let's talk about the average change in Ross's stuff within a game, on the sinker and in general. To do so, here's the movement plot for his first 35 pitches of each appearance this year.

Here's the same chart for pitches 36-70.

And finally, here's how his stuff moves after pitch No. 70 in an outing.

Ross's slider movement doesn't change much at all over the course of a start. He does, however, lose some run on the sinker, and he throws fewer four-seam fastballs as outings progress. That could be because the pitch is just meant to nestle itself in the heads of opposing hitters and set up his other stuff for the rest of the contest, but in watching him, it feels more like the four-seamer costs him something. He throws it hard, but his command of it seems effortful and inconsistent, and he might not feel he can return to the pitch once he starts to tire. The deterioration pattern of his sinker as a game goes on also drags it toward the same movement range as his four-seamer, so he might be trying to avoid throwing two different pitches that have basically the same effect but could interfere with one another in terms of his command of them.

Ross loses about one mile per hour on each offering from the first stage of an outing to the second, but not from the second to the third. He still sits around 94 (touching higher) after pitch 70. His release point doesn't change. His extension holds steady. His stuff certainly softens a bit, but no more than (for instance) Freddy Peralta's does, if you break down his starts similarly. So, why are the Brewers so much more anxious to get Ross out of there?

Firstly, of course, there are health considerations, but Ross said he felt good Monday night and was surprised when Murphy lifted him. The team doesn't want him to get hurt, but nor are they so invested in him that they're likely to treat him with kid gloves at any point. If they were confident he could have gotten the last two outs of the sixth inning last night, they'd have stuck with him.

The much bigger issue, then, is that stuff isn't just what comes out of a pitcher's hand. It's also what a batter sees, and because batters learn and adjust within a game, that changes, too. Check out Ross's pitch chart for pitches 1-35 in each appearance this year, broken down by the outcome of the pitch.

He's in the zone a lot, but hitters aren't consistently putting the ball in play when he is. With his lively stuff and the opponents still trying to find their cues against him, Ross gets plenty of foul balls and whiffs even on theoretically hittable offerings, early on. Now, here's pitches 36-70.

I've broken out these pitch bins because they best allow us to analyze the specific decisions Murphy talked about after Monday's game, but break them down into bins of 20 or 40, instead, and you'll still see much of the same. Here, it's Ross going through a mid-game phase in which (having set hitters up by being aggressive in the zone with his best stuff the first time through) he mostly stays out of the heart of the zone, or at least the upper half thereof. He's keeping hitters from locking in on him by changing things up from the previous encounters.

Now, here he is from pitch 71 onward.

It's back to the heart of the zone, now, because hitters are wise to him and not expanding the zone as much when he needs them to. That's trouble, because when he is in the zone, they're not missing as often. They've seen what he has, and they can handle it, especially as he loses a tick or two and a few inches of wiggle on the sinker. 

By no coincidence, 35 pitches is as good a proxy as you could choose for 'one time through the batting order'. Though we're not looking at this data through the lens of times through the order, we almost are. The league is averaging 3.90 pitches per plate appearance in 2024, so nine batters faced takes about 35.1 pitches. The Brewers lifted Ross after 21 batters faced Monday night because he was set to face two left-handed batters for a third time, and his recent track record says that he's not well-suited to do that. After pitch 51 this season, opponents are batting .371 against Ross, and he's struck out just four of 38 batters. The third time through, hitters have a .353 average.

It's not nearly as simple as saying that a pitcher's stuff gets worse later in games. Everyone's does. It's also not solely about batters seeing a hurler a third time. For starters like Ross, though, it's about the collision between those two dynamics. His stuff isn't far enough above the dividing line between success and failure to land him on the right side of it once it starts to decline. It would still be viable, although only just, but for the fact that hitters also get more familiar with him and anticipate him better right when his stuff is worsening.

Murphy was right to pull Ross in that instance. He's taken flak for his bullpen management this year, and he was in a tough spot because of the likely brevity of at least two or three starts in the next turn, but the Brewers are trying to win 90 games this season. Already at 14, they'd covered more of the ground (15.5% of the win total needed) than the calendar (12.3% of all scheduled games played) would suggest. Because of the construction of this roster, Murphy knows there's adversity ahead. Every win right now is precious, because it cushions the eventual fall and eases their recovery therefrom. A 15th win Monday night would have been worth the extra few outs on Milner's arm. Milner couldn't get the outs he needed, due partly to poor defense, but the decision-making by the team's dugout staff was sound.