Alek Manoah’s Steamer Projection Is a Feature, Not a Bug

© Nick Turchiaro-USA TODAY Sports

For the most part, projection systems fall in line with public perceptions of players. Yordan Alvarez is going to be very good next season, but Raimel Tapia won’t be. Shohei Ohtani is the eighth wonder of the world, and so on. But once in a while, they produce a head-scratcher that becomes the subject of debate. This leads to a lot of takes, some of them good but many of them bad. The worst are variations of “Projection X thinks poorly of Player Y, whom I like, and therefore it must be illegitimate.” They’re sometimes funny to read, though they’re mostly annoying because they stem from a fundamental misunderstanding of what projection systems are trying to achieve.

Let’s cut to the chase. The reason I’m writing about this is because Steamer projects Alek Manoah, who placed third in Cy Young voting and served as the Blue Jays’ ace, to put up a 4.09 ERA next season. That seems outlandish, even with the knowledge that projection systems are conservative by design. Manoah isn’t just a one-season wonder. His excellence extends back to his rookie campaign in 2021, and his sophomore effort seemed like a natural progression. The narrative is there: A great starter blossoms into a phenomenal one. Asserting that Manoah will go from an ERA in the low 2.00s to one in the low 4.00s is more or less a rebuke of it.

Of course, Steamer doesn’t think Manoah will land precisely on a 4.09 ERA – more on that later – but considering it’s the expected middle outcome, the shock is understandable. And while I’m not here to endorse it, I do want to point out that it’s not an indication the system is broken, or holds a grudge against your favorite player. You have your reasons, and so does Steamer.

First, a little bit on Steamer’s methodology. There are quite a few projection systems out there, and the most basic of them, at least among those that are widely known, is Marcel. It uses a player’s last three years of performance, gives more weight to the recent ones, then bakes in a regression towards league average to determine said player’s future performance. Meanwhile, PECOTA and ZiPS forecast how someone will fare in the future based on past players who share similar characteristics: body type, position, minor league statistics, and much more. They also factor in variables such as park factors and league-wide environment, which arguably provide a better assessment of what a player has achieved so far – and what he’s capable of.

If Marcel occupies one end of the spectrum and ZiPS the other, Steamer is somewhere in the middle. Its main focus is on past performance and regressing it towards the mean, but unlike Marcel, the weights and the amount of regression are based on the player in question rather than a fixed set of numbers. Basically, Steamer takes elements from both types of projections and creates its own proprietary blend.

For our purposes, the takeaway is that although Steamer is far more intricate than a projection named after a monkey, is still adheres strongly to the belief that (1) a player should be given the most credit for what he’s achieved recently, and (2) that a majority of players, given sufficient time, will see their production drift towards the average.

But wait, isn’t Manoah’s most recent body of work a Cy Young-caliber season? True, but Steamer cares far more about what’s under the hood. Compared to his debut, Manoah walked fewer batters in 2022. He also struck out fewer batters, allowed fewer home runs despite virtually no change to his fly ball rate, and stranded more runners on base. And for the second season in a row, Manoah ran a positively parsimonious BABIP – one of the lowest of any starting pitcher this year. Here’s another comparison, this time against the league as a whole:

Manoah vs. League Pitching
Statistic 2022 Manoah 2022 League
K% 22.9% 22.4%
BB% 6.5% 8.2%
HR/FB% 7.1% 11.4%
LOB% 82.6% 72.6%
BABIP .244 .289

Manoah’s walk rate was great, and his command is one of the many things that separates him from a run-of-the-mill pitcher. But when we look at his strikeout rate, we see it was barely above league average. Not much of Manoah’s run prevention came via whiffs. Steamer doesn’t think relying on batted ball outs is a reliable method of keeping runs off the board, and that view is reflected in its projections. Maybe Manoah has a genuine knack for suppressing hard contact, and in that case, we’ll be able to scoff at this debacle down the road. The thing is, metrics like left on base percentage and home run per fly ball rate take forever – and I mean forever – to “stabilize.” Not even Manoah’s 300-plus big-league innings are enough to determine whether he has actual control over them.

So how does Steamer adapt when confronted with a lack of information? It fills in the blanks with numbers regressed towards the league average. The result is largely what would have happened if Manoah hadn’t gotten just about the best outcome in every volatile aspect of pitching imaginable. If he really is the second coming of prime Zack Greinke, then Manoah deserved to have those outcomes. But Steamer, a methodical and probabilistic projection system, has little to no reason to believe that is true. Using player comparisons, ZiPS might have found a higher likelihood of Manoah following in the footsteps of all-time pitchers, but I still think it would have been bearish relative to the average fan. The bottom line: In terms of areas that are meaningful, Manoah didn’t really improve on his rookie season.

I’m inclined to split the difference. Based on his stuff and control, I see Manoah’s strikeout rate rebounding. And this is anecdotal, but seeing how he alternates between his four-seamer and sinker to keep hitters off-balance, I do think he’ll command a low-ish home run rate. But I sure as heck don’t see him maintaining anything close to a .244 BABIP, especially when the new restrictions on infield positioning have rendered Toronto’s defensive ingenuity less useful. Because he doesn’t rack up strikeouts like some of his peers, I also find it harder to believe he can keeping stranding baserunners. Manoah isn’t an “ace” in the sense that he’s one of the best pitchers around, but he seems like a number two starter, a true workhorse. Number two starters don’t put up low-4.00s ERAs!

One more thing I’ll say about that eyesore of an ERA projection is it may not reflect the full range of possible outcomes. A 4.09 ERA isn’t a death sentence – it merely means that 50% of projections are below it, and 50% are above it. Yes, Steamer thinks Manoah might do even worse. But let’s consider the cup half-full for now. There probably are scenarios where, once again, he authors a marvelous year. If they’re outliers, though, they wouldn’t move the median, which is simply the middle value. The fact that they’re assigned to Manoah in the first place speaks highly of his raw, possibly still untapped potential. Most pitchers are paired with an assortment of tightly clustered outcomes, which reflect how unspectacular they are. This is speculation on my part, as I don’t have access to specific data, but it makes sense. Steamer wouldn’t flat-out ignore a season Manoah closed out two months ago. It might have influenced the top percentile outcomes, just not the median everyone gets to see that much.

Above all, projection systems aren’t some almighty edict, nor do they proclaim to be. They’re one of the many tools at our disposal to evaluate and analyze players, which makes them just as important as everything else in the toolbox. They’re here to inform our decisions, not enforce them. You don’t have to agree with Steamer to acknowledge that there is a concrete and rigorous process behind its prognostications. I certainly don’t think Manoah will turn into Marcus Stroman next year, but I can understand how Steamer came to that conclusion. It’s skeptical, and for good reason. If projection systems crowned every breakout starter a forever-ace, they wouldn’t be very accurate. Manoah could be an exception, but decades of data tell us that he’s likely a good player moving forward, plain and simple.

At the same time, blindly trusting a projection is about as silly as not considering one at all. Doing so reduces baseball discourse into a recitation of figures, with nuance and perspective thrown out the window. Steamer is hardly perfect, and there’s a good chance it’s dead wrong about Manoah. This article is neither a defense or a refutation of this particular Alek Manoah projection. It’s more of a reminder – that no, Steamer hasn’t lost its marbles, and yes, it’s okay if you’re outraged at its pessimistic outlook. Just don’t go dismissing the whole system because of a single statistic. That’s not what projections are about, and it definitely isn’t what baseball is about.





Justin is a contributor at FanGraphs. His previous work can be found at Prospects365 and Dodgers Digest. His less serious work can be found on Twitter @justinochoi.

27 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
The Strangermember
2 months ago

Why do you hate my favorite pitcher?

(Excellent breakdown of how projections work/refutation of hot takes. Much appreciated.)

richwp01member
2 months ago
Reply to  The Stranger

I also like the description of other projection systems on a spectrum, which is very helpful.