Baltimore Orioles: Wei-Yin Chen Has Been Lucky


Don’t let Chen’s seemingly solid statistics fool you. Mandatory Credit: Joy R. Absalon-USA TODAY Sports

Aside from an indefensible deal for a reliever, the Baltimore Orioles remained dormant at the trade deadline last week. They did nothing to upgrade their rotation, which didn’t pitch all that well for the season’s first four months — with one, ostensible, exception: Wei-Yin Chen, who, in his third year in the States, has apparently taken a step forward. Even with his difficult outing Wednesday, he still has a respectable 3.90 ERA, with good peripherals to boot — a 3.70 xFIP and a 3.81 SIERA. Perhaps the Orioles didn’t go for a starter on the block because they felt that he would sustain this performance henceforth; sadly, this was an error on their part.

Chen has predicated his success thus far on walks, or the lack thereof, as his BB% has improved dramatically to 4.2% (from a previous career mark of 6.9%). Strikeouts have also aided his performance; his 2014 K% sits at 17.7%, not much lower than the 18.6% figure he compiled in 2012 and 2013. These two facets of his game, coupled with a decent 40.9% ground ball rate, have given him the aforementioned satisfactory peripherals. But there’s more to it than these.

Strikeouts and walks can tell us how lucky or unlucky a pitcher has been, but what about the processes that lead to those? In other words, while statistics such as xFIP and SIERA (based primarily on Ks and BBs) tell us what a pitcher’s ERA should be, can anything tell us what his strikeout and walk rates should be?

Last year, Mike Podhorzer devised an expected strikeout rate, or xK%, equation. To determine what a pitcher’s fan rate should be, it utilizes four variables: Str%, the percentage of all pitches that are strikes; L/Str, the percentage of all strikes that are looking; S/Str, the percentage of all strikes that are swinging; and F/Str, the percentage of all strikes that are fouls. With the updated coefficients, it correlates extremely well to actual strikeout rate (Adjusted R-squared of .913). This formula thinks his strikeout rate should be 18.2% — a bit higher than it has been. But we shouldn’t close the book on Chen there.

Around that same time, Podhorzer whipped up an expected walk rate, or xBB%, equation. It only takes into account three variables — Str% and K%, as well as I/Str%, the percentage of all strikes that are balls in play — but still gives a high correlation to BB% (R-squared of .770). This formula believes Chen should have issued free passes at a 6.2% clip — a good deal worse than his actual mark of 4.2%. With strike and in-play strike rates only marginally better than those of all pitchers in MLB, this should come as no surprise. At any rate, this certainly doesn’t bode well for Chen in the season’s final months.

But hey, let’s say you don’t trust Podhorzer. Let’s say you want other xK%s and xBB%s, ones that don’t depend as much on factors outside the pitcher’s control (framing, umpires, etc.), or more importantly, ones that offer more predictability. Later last year, after Podhorzer finalized his work, another statistician by the name of Chris Carruthers stepped into the arena, with his own regression equations to model strikeout and walk rates.

Carruthers founded these equations on a simple concept: Any pitch has five possible outcomes. If the batter didn’t swing, it could land outside the strike zone, resulting in a (theoretical) ball; or it could land in the strike zone, for a (theoretical) looking strike. If the batter did swing, he could make contact, resulting in either a foul or a ball in play; or he could fail to make contact, resulting in a whiff. Hence, the pitch can be: BIP, SwStr, Foul, OZ-Looking, or Z-Looking. His equation intakes all of these in rate form, and spits out numbers that provide far more predictability than Podhorzer’s — his (i.e. Carruthers’s) xK% and xBB% equations both have RMSEs of lower than .0500 for even the largest of sample sizes.

Back to Chen. How do the statistics with better forecasting ability feel about him? Much more poorly than the others: They think his K% should be 13.5%, and his BB% 8.9%. The latter mark comes as a result of his rather astronomical 39.4% OZ-Looking%, which is to say, nearly two of every five of Chen’s pitches have been theoretical balls. Those have come at the cost of looking strikes, his subpar rate of which (11.8%) is to blame for the feeble former number. Regardless of the causes, these statistics give very little hope for Chen’s near future.

As the Orioles make their final push for the playoffs, they will need all the help they can get. If they’re wise, though, they won’t rely on Chen for much of that, as he will most likely come crashing down to Earth.

All statistics courtesy of Baseball-Reference.