Pitching by the Numbers: Model starters, revisited

A follower on Twitter (@MichaelSalfino) asked me to update my model that I used early this season to assess pitchers.

But I don’t want to rank everyone again. Instead, I set filters to limit how many pitchers we would pull. My filters were pitchers with at least seven starts AND a K/9 of at least 7.00 (actually below average this year, slightly, but we don’t want to be too tough) AND a SO/BB of at least 2.0 AND an ISO allowed of under .130 (average this year is about .147). ISO is slugging average minus batting average so we isolate the hits that are most damaging (extra base hits) when trying to prevent runs.
I’ve sorted the chart in reverse order of ERA. It’s not because I think the guys on the top are the best pitchers, obviously. It’s just that they are the most gettable (because their ERAs are relatively high). With filters like this, you are not going to expect pitchers to give up too many runs. And for the most part, they do not. The median ERA is slightly over 3.00. And we have only four pitchers who have ERAs in the first half over 4.00 (none over 5.00).
But before we get to the chart, a couple of points.
It’s pretty well documented that strikeouts minus walks are better for us than K/BB ratio. I have no issue with that. It’s not a stat I get, however, and not one that most are familiar with, and I do not think it’s advantages are strong enough to include it given that I not only use K/BB but also K/9.
While I could easily calculate it myself, I have one funky stat in there already. Though I don’t think ISO allowed should be funky. I mean, we all agree that ISO is a great stat for hitters. So why shouldn’t we care about it for pitchers? But to be clear, my use of ISO allowed did cause a minor uproar when I introduced it last year and cited it again this year. The general view by some colleagues I do respect is that it’s not a good stat for us because it’s not projectable. Basically, we’re told, it takes too long to stabilize to be of in-season use. Of course, I disagree. But I’ve included 2013 ISO allowed and career ISO allowed for these pitchers so you can make your own determination. But I note that only a quarter of our 28 pitchers have ISOs this year that significantly differ from their career ISOs.
I thank Baseball-Reference for the ISO allowed stats, which are not tracked by anyone else to my knowledge (though you can do it for any pitcher by hand if you just have their slugging allowed and batting average allowed).
Once more, the requirements to be on the below list: seven-plus starts, seven-plus K/9, two-plus K/BB, ISO allowed less than .130:

Obviously pursue the injured pitchers at your own risk. We won’t discuss them.
Those with the worst ERAs are most gettable. So that starts with Porcello. He sure looks like he should have a better ERA to me. The control is top-shelf, but nothing else is. Still, he should be a serviceable low-level mixed-league starter in the second half and he’s currently available in 87 percent of leagues. I’m tired of being wrong about Porcello. But the purpose of this column is to come up with a model and then follow it because models consistently beat instinct/guessing.
Straily is even more available (89 percent) and is in a better pitching environment. His recent minor league demotion was just All-Star break related.
Lincecum is victimized now by his high walk rate. But there’s a lot to like in that profile. He’s only available in 15 percent of leagues though.
Nova is widely available (not owned in 75 percent of leagues). He’s such a strange pitcher, once scorned by sabermetricians because his K rate and other foundational stats did not support his too-good ERA. Lately, especially last year, those foundational numbers radically improved, yet his ERA cratered. Go figure. Clearly, Nova is not going to sustain a .077 ISO allowed, about half his career rate. But it’s reasonable to think he can be above average in limiting extra-base hits going forward. And his ERA right now is way too high when you look at his numbers, so there’s a decent chance regression in ISO will not cause a decline in his ERA. And I think the Ks are bettable now. So go get Nova.
Cobb has a concussion and is available in 29 percent of leagues, so you’d probably have to trade for him and why do that when you can just get Nova?
Liriano is the other interesting guy on the list. You can’t get him, of course. But should you believe him? The .067 ISO is not going to last. But his career ISO allowed is above average, though not quite great. His big problem is limiting walks and staying healthy (and those two things are often related; pitchers throw a lot of innings sometimes before we know they’re hurt). I would trade Liriano to a true believer but I’d want 75 cents on the dollar of what he’s earned, which is a lot –- a top 30 hitter. If he’s healthy, he’s probably going to be good. And I blindly said in response to a question that he’d collapse in the second half given his career numbers. But as often happens when I actually see the data, I’ve changed my mind.

What to Read Next