Is Bud Norris better than Greg Holland?

On February 12 this year, the Cardinals signed Bud Norris. It was a move that should have been mostly uneventful except it had the unfortunate timing of coming directly after the Chicago Cubs signed Yu Darvish. Also Norris had some, let’s say, unfortunate comments about foreign players. (It was more than unfortunate). Combine that with a 4.48 career ERA and the vast majority of Cardinals fans were opposed to a fringe bullpen signing.

On that same day, Jeff Sullivan of Fangraphs compared Bud Norris to Greg Holland. Before Norris wound up on the DL twice for knee inflammation, Norris had a pretty good first half. He pitched 33.1 IP, struck out 42 (30.7 K%), walked 13 (9.5 BB%), and had a 45 GB%, which all led to a 2.43 ERA, 3.16 FIP, and 3.43 xFIP. He pitched in 28.2 IP after that and his numbers fell to 32 strikeouts (24.6 K%), 14 walks (10.8 BB%), and while his groundball percentage remained stable at 44.6%, his HR/FB% doubled and he allowed five home runs in the second half in comparison to three homers in the first.

The very next month, Bud Norris turned 33-years-old. He’s also a pitcher. The injuries can explain away the decline in performance, but it’s just as easy to look at the rest of his career as a mediocre at best starter and think his first half was nothing but a fluke. Plus, even if it wasn’t a fluke, injuries can linger or a different injury can come or even the pitcher simply never recaptures the magic he found. Any number of things can happen to make it easy to distrust that first half.

It wasn’t quite as hard to believe in Greg Holland. Let’s ignore salaries here, because if Holland and Norris both signed for the same salary, there’s no doubt which player is the better bet going into the season. Holland rebounded from Tommy John surgery to help stabilize the Colorado Rockies bullpen. His numbers, to be honest, don’t look all that impressive, but he did play half his games at Coors Field. (Just look at Jake McGee before and after Coors for instance)

The theory goes that a player that returns from Tommy John is a copycat version of themselves the year after TJ surgery, and copycats are never quite as good as the original. The second year after, they return back to the form. As an example of this theory from a Cardinals perspective is Adam Wainwright in 2012 and 2013, though Wainwright’s 2012 is quite a bit better than you probably remember. But his 2013 was better than he had ever pitched before.

A less persuasive argument for Holland, for me anyway, was that he was very bad in a few appearances that dragged down his numbers. In six games last year, he allowed 17 of his 23 earned runs in just six appearances. Five of these appearances came in August. The theory goes that if he just limits those appearances, he’ll be better. That is true. It just so happens to be just as likely that those runs will be more spread out this year than limited to six bad appearances, with the caveat obviously that he should have less of those and runs in general at a place like Busch Stadium over Coors Field.

Before the season, Steamer and ZiPS release their full-season projections of every player in the MLB. Once the season starts, they also have a very handy rest of season projection (shortening to RoS for the rest of this post). I’m fascinated by these, because while it is true that two weeks of baseball is much too short to draw any meaningful conclusions, the RoS can change quite a bit based off these first few weeks. I’ll use Holland and Norris as an example, with their current 2018 performance factored into the RoS.

Here are Holland and Norris’ projections before the season:

Unfortunately, there appear to be a couple quirks with these projections. Most obviously, Bud Norris is a starting pitcher part of the time for ZiPS. Barring a disaster, he will not pitch 100 innings this year. I also have no idea why Steamer has Norris pitching only 45 innings, but has Holland pitching 65 innings.

Anyway, the point is that there’s a clear advantage for Holland in these projections. I wish I knew what ZiPS projection of Norris was as a full-time reliever, but those 100 innings include 15 starts, so at 5 innings per start, 75 of the 100 innings are as a starter. Depth Charts on Fangraphs combines Steamer and ZiPS but does not seem to account for the fact that ZiPS has Norris as a part-time starter. If you average ZiPS 4.49 and Steamer’s 3.75, it comes out to 4.12 which is what Depth Charts has his FIP as, but Depth Charts only has him as a reliever so it should be lower than that.

If you accept the common wisdom that a starter converting to a reliever loses a full run of ERA, ZiPS has his FIP as a starter at 4.75 and as a reliever as 3.75. Keep in mind I am doing some very simple math to figure this out, so there is a high probability ZiPS reliever only projection does not have him with a 3.75 FIP. But that’s the best I can do with the information I have. Also, the common wisdom may be wrong.

Bud Norris has since thrown 5.2 utterly fantastic innings - about as good as a reliever can pitch - while Holland has pitched about the worst 1.1 IP you can pitch. It is actually incredible how much Holland has damaged his RoS projection with his start this season in so little time. Similarly, with Norris’ start, if you add his RoS with his current stats, he is a lot, lot better than what his original projections say just based off a few innings.

There are again a few quirks here. Holland’s K/9 actually rises for ZiPS, because I’m guessing more walks leads to more strikeouts (more baserunners = more opportunities for strikeouts). Also I assume the RoS finally figured out that he was playing at Busch Stadium, which would explain both the lower HR/9 and the lower FIP. So I wouldn’t look at either of those as much of a positive. For instance, Holland’s ZiPS WAR projection jumps down to 0.6 despite superficially better numbers.

ZiPs is steadfast that Norris will still become a starter at some point, but he’s already lost two starts based off his early performance so hopefully, he will be projected for no starts at some point. Again, using my back-of-the-napkin one full run lower for a reliever estimate, ZiPS has Norris as a 3.60 FIP reliever now.

Now, let’s compare their original projections to their rest of the season projections combined with their start so far.

I thought it would be more interesting to give Norris and Holland the same amount of innings for a more direct comparison. I also only used Steamer because I frankly have no idea how to adjust K/9 or BB/9 for starter to reliever transitions, but if I’m at all correct, a difference of 3.60 and 3.55 is close enough. The WAR figure for Holland looks kind of wonky, but like I said, I’m pretty sure it was assuming another year in Coors. That’s how a 0.18 drop in FIP leads to your WAR getting cut in half.

The hot start by Norris and cold start by Holland explain most of the difference in the estimated 2018 end results, but it’s worth pointing out that Norris and Holland are now nearly indistinguishable from a value standpoint in expected performance going forward. Greg Holland’s combined RoS for Steamer and ZiPS is 3.54. Norris is likely only a little bit above that.

As of this very date, it looks like the Cardinals paid $14 million plus the cost of a draft pick for 0.5 WAR. That is probably significantly lower than the Cardinals were hoping for when they signed him.