Exactly how valuable has Robbie Ray been?

After his seventh straight outing of at least six innings with no more than two runs allowed Monday, Robbie Ray lowered his ERA to 2.71 over 159.1 innings. In doing so, not only has he been keeping the Jays fading postseason hopes alive, but he has emerged as a legitimate contender for the AL Cy Young Award.

When the stats were updated Tuesday morning, Ray’s Baseball-Reference page pegged him a whopping 6.0 wins above replacement for his 2021 efforts. On the MLB leaderboard, not only is that right alongside Zack Wheeler, Wade Miley and Walker Buehler for the best in baseball, it’s a full win clear of Gerrit Cole for the best in the American League. By this measure, Ray is not only a contender for the Cy, but a pretty clear frontrunner.

By contrast, at FanGraphs he’s credited with a decidedly less eye popping 3.4 WAR for those same results. That still places him along the AL’s elite, but in a cluster of five pitchers behind the top three of Cole, Nathan Eovaldi and Carlos Rodon.

While exactly where he stacks up along peers is ultimately trivial, from a value perspective it is no small beans. If wins go for something like $7- or $8-million on the open market, that’s upward of a $20-million difference.

At the heart of this difference is that valuing pitching performance is as much an art as science, and the two systems use very different approaches. Baseball-Reference (bWAR) starts with actual runs allowed, compared to league average, and applies adjustments for estimated opponent quality, parks, team defence. FanGraphs (fWAR) by contrast focuses on evaluating the things over which pitchers have the most control and starts with FIP (fielding independent pitching, which is scaled to ERA) before adjusting for park.

Given that Ray’s ERA has been well below his FIP since the early days of the season, I was aware that his bWAR was better than his fWAR, and it was logical that would be the case. Still, I was very surprised by the magnitude of the current gap, with one almost double the other.

Ray’s sensational ERA of 2.71 is well ahead of his FIP of his (still very good) 3.43 FIP. Over 159.1 innings, a difference of 0.72 runs (per nine) works out to about 13 runs, or about 1.3 WAR. Usually, when bWAR and fWAR diverge for a pitcher, this is most of the difference. In this case, 1.3 would only be about half the 2.6 WAR gap.

So I figured there had to be something unusual in one or more of the the other adjustment factors. But it turns out, there’s nothing exceptional there. Moreover, FanGraphs also publishes an alternative metric they call “RA9-WAR”, which is basically just a stripped down version of Baseball-Reference’s calculation, using actual runs allowed with just a park adjustment factor. And sure enough, here they have Ray at 6.0 WAR too.

So that still left the other half of the puzzle unexplained. And it turns out there’s something else remarkable about Ray’s 2021 season.

Across MLB in 2021, 16,335 of 17,838 runs scored have been earned. That works out to 91.5%, or conversely that about one-in-eleven runs are unearned. So while the MLB average ERA is 4.23, the MLB average RA/9 is 4.62 (for the AL, both are a little higher 4.31 and 4.69). So one would expect that a pitcher with a 2.71 ERA to give up a little over three total runs per nine innings.

But of the 49 runs Ray has given this year, just a single run allowed back on May 27th against the Yankees has been unearned (whereas one would expect more like four or five to be). His ERA of 2.71 only increases to a RA/9 of 2.77, and while his ERA is an excellent 1.60 runs better than the AL average, Ray’s RA/9 is even more incredibly just shy of two runs better.

Have his defenders just been incredible behind him? Perhaps, but there’s a more structural factor at play. Ray is a fly ball pitcher, and they tend to allow a smaller percentage of unearned runs because balls misplayed in the air tend to get scored as errors much less often than ground balls do when booted. If the fielder doesn’t get a glove on it, it rarely goes down as an error, even when it really should have been caught.

So to get the question posed at the outset, exactly how valuable has Robbie Ray been in 2021; which WAR ought to be believed? At the end of the day, this is mostly philosophical. At brass tacks, when Ray has been on the mound, runs have been kept off the board at a league leading rate.

While’s he’s certainly the person most responsible for that, it’s likely the case he’s been buffered by a tailwind of beneficial variation: allowing fewer hits than expected given the contact profile, more runners stranded than unexpected, maybe the dearth of unearned runs. Certainly on a go forward basis, the results would be more in line with 3.4 WAR than 6.0 WAR.

Personally, my basic inclination when there’s a big gap between results (bWAR) and process (fWAR) in a single season is to split the difference. Has he really been a ~8 WAR pitcher in a single season? Probably not. But equally, the 3.4 fWAR feels somewhat parsimonious. 4.5-5.0 WAR feels about right.