Carlos Frias is suddenly a very important man. Coming into spring training, he was buried deep on the Dodgers’ starting pitcher depth chart. He filled the fifth starter role in a mostly-capable fashion last year, but his long-term destination has always looked like the bullpen. In February, it looked like that could finally happen. However, after injuries to Brett Anderson, Mike Bolsinger, Frankie Montas, and Brandon Beachy, Frias is on the inside track to winning back the number five starter position (if it can be so won at this point, instead of the revolving door approach).
If you want to make a statistically-oriented case against Frias getting that role, it’s simple:
This is a leaderboard sorted by lowest strikeout rate, among starters with at least 70 innings pitched last season via Fangraphs. Frias is 13th-worst out of 157. By K%-BB%, he was 12th worst. Frias’ peripherals were nearly identical to Kyle Kendrick‘s. Frias’ ball-in-play numbers were better, which is why he wasn’t a complete disaster last season, but it wasn’t pretty.
Despite this, when Beyond The Box Score looked deeper into Frias’ season last week, they found some reasons to be optimistic:
Frias’ 2015 season as a starter is plotted above as the red dot. He is the largest outlier amongst all 2015 qualified starters. When starting, he caused batters to whiff on 21.0 percent of their swings and induced swings on 47.3 percent of all pitches — good for a 9.9 percent whiff rate. This would make him an above-average starter when it comes to fooling batters given the league average of 9.3 percent.
In simpler terms, Frias was getting a lot of swings and misses, but they weren’t translating into strikeouts as much as they should have. There’s obviously a strong relationship between the two numbers and Frias broke that relationship last season. The article concluded that Frias is due to improve significantly this year as a result. However, this conclusion lacks some detail, which is worth exploring. In order to understand if Frias is going to improve this season, first we must understand why those swinging strikes did not turn into strikeouts in 2015.
The first place to look is pretty obvious. In 2016, Carlos Frias recorded three looking strikeouts, to his 40 swinging strikeouts. The league-average ratio of looking strikeouts to total strikeouts is around 23%. In 2015, Frias’ strikeout looking to strikeout ratio was 7%, the second-lowest among the 194 pitchers to throw at least 70 innings. If you give Frias an average ratio while keeping the swinging strikeout total the same, he would gain roughly nine strikeouts, bumping his strikeout rate to almost 16%. According to the trend line on the Beyond the Box Score figure, that’s still short of the expected rate of 20% or so, but it’s a significant gain.
This raises the obvious question: why is Frias not getting looking strikeouts? Is it luck, or something else?
The first place to look is two strike swing tendencies. Last season, Frias has a pretty normal swing rate on all pitches, both in-zone and out-of-zone. However, breaking up his in-zone swing rate by count reveals something dramatic:
|Carlos Frias||League average*|
|Behind in count||66.7%||65.9%|
The league swings a little more against Frias than they do for an average pitcher. This is true across most situations (except even counts), but with two strikes there’s a huge change. The entire league’s in-zone swing rate goes up with two strikes, which isn’t a surprise given that hitters want to protect the zone. However, this was especially true for Frias. Last season, Frias threw 88 pitches into the Pitch FX strike zone with two strikes. Batters swung at 86 of them. The two they watched accounted for two of Frias’ three looking strikeouts.
The results of this high percentage are obvious: if batters don’t watch the strikes, they’re not going to strike out looking. So the next question becomes: is this a trait specific to Frias, or is it one that is prone to regress? While there wasn’t time to run a fully statically robust test, a quick one will have to do. The way I did this was to look at the 40 pitchers with the highest two strike in-zone swing rates in 2014 (minimum 50 opportunities). If that pitcher had another 50 opportunities in 2015, the numbers for both years were compared. The average in-zone two strike swing rate (for pitchers who qualified in both years) in 2014 was 96.5%. The average of those pitchers in 2015 dropped to 91.9%, almost back to league average.
For Frias, this is good news. This means that batters not watching potential called third strikes is not necessarily due to anything he’s doing in particular. If his in-zone two strike swing percentage regresses back the same 4.6% as the sample group, suddenly he gets four extra looking strikeouts. Using the assumption of adding them to his overall strikeout total**, Frias’ strikeout rate climbs to 14.1%. This isn’t a huge difference – instead of Kyle Kendrick, now he’s Aaron Harang – but it’s something. There are other areas still worth exploring.
Onto the next question. What about just plain bad luck on borderline called balls? Here are all of Frias’ two strike pitches that were taken last year, via Baseball Savant:
The three pitches circled in red are Frias’ three called strikeouts this year. The other 124 were balls. A lot of these pitches were clearly called correctly, but a lot of them are right on the edge. It’s worth noting that the drawn zone isn’t perfect; for example that cutter in the lower left of the box is not considered an in-zone pitch for whatever reason.
Framing could be a factor here. Frias threw 66-2/3 of his innings to Yasmani Grandal, 6 to Austin Barnes, and just 5 to A.J. Ellis. Ellis is the only below-average framer of the bunch, but there may be something about Frias that makes him particularly difficult to receive. Or, maybe he’s just unlucky. Breaking that down further is beyond the scope of this article. The strike zone in 2015 was quite large (though smaller in two strike counts), so a few of those pitches, especially on the bottom edge, could/should have been called strikes. “Bad luck,” especially without any quantification of that luck, is an unsatisfying way to explain things, but it’s certainly at play.
Those two factors combined are probably enough to account for the lack of looking strikeouts, but recall that there are missing swinging strikeouts too. There’s one factor I wanted to look into here: pitch selection and distribution, especially with two strikes.
According to Brooks Baseball, here are the pitches Frias used with two strikes last year:
|vs LHH||vs RHH|
Frias leans very heavily on his cutter with two strikes, and it makes sense because it was his best pitch by whiffs last year (other than the limited number of changes). Still, it isn’t what you’d call a traditional “out pitch.” The swing and miss rate for Frias’ cutter last year was about 18%. That’s a lot less than a well-executed slider, for example.
This brings up an interesting point, though. Many pitchers turn to a high whiff out pitch pitch to convert two strike counts into strikeouts. Frias doesn’t have that to the same extent, so he may have more trouble than an average pitcher converting count leverage into strikeouts. This could explain part of his strikeout deficit noted above.
One way to test this hypothesis is to compare Frias to other pitchers of similar swinging strike rates, and see what the difference is between their swinging strike rate on all pitches compared to their swinging strike rates with two strikes. Assuming two pitchers have the same overall swinging strike rate, the pitcher without an out pitch will be better at getting ahead of hitters earlier in the count, but will have more trouble putting them away.
Carlos Frias’ swinging strike rate goes up with two strikes with respect to his overall swinging strike rate, as the table below shows. The whiff rate is about 2.5% higher when it comes time to put a batter away. Since that number doesn’t mean much on its own, the table below also shows the swinging strike rate differential for pitchers who have overall swinging strike rates very close to Frias’ level.
|SwSt%, total||SwSt%, 2 strikes||Difference|
On average, the pitchers in this table (including Frias) have a 3.7% difference between their swinging strike rates in all counts compared to their swinging strike rates with two strikes. Frias is about 1.2% below that. Since Frias threw 317 pitches with two strikes last year, that accounts for four missing whiffs with two strikes. That can’t be directly translated into four extra strikeouts since it doesn’t account for the results before two strikes changing, but it’s still a significant change which limits his strikeout ceiling by a couple percentage points. A lot of this seems inherent to Frias’ pitches and overall style, so this is not something that will be easy to change in 2016.
Overall, it seems like Frias’ strikeout rate could be due for a rebound, due to his bad luck with takes on two strikes. A few percent of this shift is easily attributable to two strike in-zone swing rate, while a few percent may be dumb luck. However, Frias’ repertoire may lead him to difficulties actually getting a whiff with two strikes compared to an average pitcher. This will put a cap on whatever regression towards the trendline that could happen. He’ll still end up below that line next year, barring any big changes in how he approaches batters.
One other concern: what if Frias’ swinging strike rate drops in 2016? That would throw this whole thing out the window (good thing this post is only 1800 words). Only one spring training start from Frias has been recorded on Pitch FX, but in that start his velocity was significantly down from last year. If the team does anoint Frias the fifth starter, it could be a sign that this has gone away, since they surely have more information on Frias’ velocity than we do. It’s still worth keeping an eye on. There’s also the adjustment factor, which was not addressed here but is a real concern given Frias’ lack of pitches with a bend.
Ultimately, after this stupidly long exercise, the diagnosis on Frias is probably about the same: he’s an okay spot starter. He may be more than that, but only a significant amount of results lean in his favor in 2016 when they did not last year. He’s probably good enough to give the job for now, but there is some risk in doing so. Frias’ value will be determined by the adjustments that both he and his opponents can make next year. And, in the end, he’s probably still a reliever long-term.
*A few liberties were taken with minimum pitch count to keep the number of pitchers around 200 for each set due to Baseball Savant query limitations
**Only partially true, since some portion of the four results which “change” ended in strikeouts already. This could be responsible for a few tenths of a strikeout but at this magnitude the assumption isn’t that large.