A Spin Through the Rotation

I have an admission to make, I’m not watching every Nats game this season. I try to catch parts of games here or there when I can, but so far I think there’s more games I’ve entirely missed than watched in full. If you are like me, you also have probably found that when it comes to the Nats’ starting rotation there are some narratives in your head due to which games you happened to have caught, that are not entirely based in reality. So I decided to dive into the numbers to catch up on what’s really going on with the Nats starters.

To start (eh? eh?) here’s the bread and butter chart of Nats starters by the classics ERA and FIP. I’ve left Brad Lord out because I want to hone in on the planned rotation.

If you’ve caught any MacKenzie Gore game you are probably not surprised by these numbers. He has been consistently great all year with only one real outlier against the Orioles where he somehow combined 10 hits with 9 strikeouts to only make it 3.2 innings. And even then he only gave up 2 runs. Gore has become exactly what the Nats hoped and dreamed he would be when they traded Juan Soto for him three years ago, a Cy Young contender who can lead their rotation for the next decade.

Outside of Gore the story is more muddled. Irvin has been completely inconsistent with some great outings mixed well with outright stinkers. He has 5 games where he pitched into the seventh, giving up just 5 runs total. He also has 4 games where he couldn’t get through six innings and gave up 14 runs in those. I admit I had only seen his good games and did not realize how badly he had done outside of those. His 3.93 ERA still looks nice but it’s not supported by his FIP.

Parker is another guy where the narrative in my head did not match reality. He started off the season great, giving up just 5 runs in 32.1 innings across his first five starts of the season. As you can tell from the 4.71 ERA and 4.41 FIP he was not able to keep that up in his next seven. He’s pitched into the sixth inning only once since April and has given up at least three runs in all seven starts. With his start to the season I thought Parker had established that last season was not a fluke, but he has quickly and thoroughly disabused me of that notion.

The last two are fascinating, Soroka was briefly out with an injury and has only made half the number of starts as the others, but has not done much with them by ERA and FIP. While with Williams, by ERA he is by far the worst and from what I’ve seen of him and reactions to games I’m not watching, he’s looked the worst on the mound too. But by FIP he’s the second best and it’s a genuinely decent 3.93. Is this a case of an early season small sample size playing a trick on us?

To try to answer that I turned to Baseball Prospectus’ more advanced measures of pitch quality StuffPro and PitchPro. These statistics offer a bit of a trade off. On the one hand, they are better at isolating a pitcher’s performance from the team around them and become useful at much smaller sample size than ERA and FIP. On the other hand, they utilize machine learning models to better approximate value and are not purely measurements of what occurs on the field. I know there are mixed opinions of the use of models, but I think the Baseball Prospectus team have made a good case for their effectiveness. If you’re interested in learning more about that you can read about them here.

To paraphrase StuffPro and PitchPro predict for each pitch a given pitcher throws the probability of a swing, the probability of a take, the probability of contact, and the value of contact. Each model is fairly similar with some small differences in the inputs they use mostly related to how the pitch moves. The great thing about both, especially so early in the season, is that they do not need a lot of pitches before giving accurate results. So let’s look at the Nats starters by PitchPro and StuffPro keeping in mind that the lower the number the better.

No surprise that Gore is yet again far and away the best pitcher on staff. Baseball Prospectus says a pitcher with an elite arsenal would have a PitchPro/StuffPro of around -1.25 runs. So while Gore does not look quite as elite by this measure as he did ERA and FIP, he is still solidly in the good range.

The reason for Irvin and Parker’s inconsistencies show up here as well. Both are just below average with their arsenals, with no individual pitches standing out. That allows them to have games here and there where their pitches are a little tougher leading to a good outing, but overall they don’t have the pitches to be consistently good. These are the arsenals of a good 4^th/5^th starter instead of their current positions of 2^nd and 3^rd.

You were probably wondering why I said Soroka’s ERA and FIP were fascinating above when they were both bad, but that was just a bit of foreshadowing for his solid PitchPro and StuffPro values. Despite the terrible ERA and FIP, Soroka’s pitch values are closer to Gore’s than his teammates’. The Nats should continue to be patient with Soroka in the rotation and give him some more time to see if he can get his ERA and FIP aligned with his pitch values.

One person the Nats do not need to be more patient with is Williams. His arsenal is exactly as terrible as his ERA and any hope that lower FIP might have given you should be dashed. As soon as the Nats have someone who looks remotely ready in AAA they should jettison Williams. Honestly the Brad Lord experience might be better at this point.

Overall I was hoping this was going to be a more positive outlook, but outside of Gore there’s not a lot to be excited about. Irvin and Parker have not developed further and look like they might be maxing out as 4^th/5^th starters. Williams has been exactly as awful as expected and while Soroka has some interesting signs in his pitch values, he eventually will need the results to back that up. If the Nats want to be a playoff team they’re going to need a lot better behind Gore next year.