Analyzing Trevor Williams Pre and Post-Injury 2019

In 2018, Pirates’ pitcher Trevor Williams had the best season of his young career. The right-hander started 31 games for Pittsburgh, which was second on the team behind Jameson Taillon. Williams started his season on April 1 with a game in Detroit. He tossed six innings, giving up no runs, while striking out a batter and walking five. But there was one intriguing number that stood out when Clint Hurdle decided to close the curtain on his starting pitcher: zero hits.

Williams isn’t a strikeout pitcher, which can be cause for concern in today’s game. It’s well-known among fans that he is a pitch to contact type of thrower. Sometimes that works, sometimes it doesn’t. The reason Williams is able to find some success with this strategy is because of how particular he is about where he spots his pitches. He tends to try to throw exactly to a spot on the black. It’s for that reason he walked five during that game in Detroit.

That was the beginning of a season in which Trevor Williams showed plenty of promising signs. Take, for example, how well he pitched after the All-Star break, starting on July 23 at Cleveland. Between July 23 and his final start on September 27, Williams started 12 games, going 7-3 with an ERA of 1.38. Batters were only hitting .217 against him, with an OPS of .578. He was pitching on a superstar level.

If we take a look at other 2018 statistics, according to Baseball Savant, the percentage of batted balls against Williams for balls which were barreled or were solidly hit was below league average, which was 6.3 percent and 5.6 percent, respectively – Williams, on the other hand, came in at 5.8 percent and 4.3 percent. To go along with that, batters made weak contact at a rate higher than league average, at 6.8 percent to the league’s 4.6 percent. Finally, the average exit velocity against Williams was 85.6 mph, which put him in the top 10 percent in the league.

So, when 2019 rolled around, hopes were high regarding Trevor Williams, particularly because of how strongly he finished the previous season. While the expectation was that there would be some amount of regression following his tremendous second half, it looked promising that his progress would be minimally diminished – at the very least, he would’ve still been a pitcher well above league average.

For nearly two months, that was the case. Prior to his injury, Williams started nine games between March 31 and May 16. Over that time, Williams commanded a 3.33 ERA, while hitters registered a .678 OPS. Both of those metrics fell below league average. Where ERA is concerned, he was pitching over one run better than league average, which was 4.51 for the season. As for OPS, hitters were posting a league average of .758.

---

During the AT&T SportsNet broadcast on May 16, Williams can be seen clutching at his side after striking out the Padres’ Wil Myers. The camera panned to Williams’ father, a concerned look on his face, sliding the scorecard back into a black binder. After 3.2 innings pitched, Williams’ outing in his hometown San Diego was done.

Williams would be diagnosed with a “side strain.” Once he returned to the mound on June 19 against Detroit, what was the outcome? Five innings pitched and seven earned runs on nine hits. While that start was far from good, it was his first start in over a month; for that reason, it’s hard to critique Williams at the conclusion of his first outing post-injury. For the remainder of the season, though, he failed to get much better.

In 17 games started post-injury, Williams posted a record of 5-8 with a 6.58 ERA and an opponent OPS of .947, registering nearly half as many walks as strikeouts over that time (34 BB to 71 K). Hitters also batted .302 against him over those 91.2 innings.

It can be tough to return from an injury. Performance may dip as a result. So let’s shift our attention to the data to see if those numbers paint a picture of a quantitative regression. Because Williams mostly sticks to two pitches, the four-seam fastball (51.9 percent) and slider (20.1 percent), according to Brooks Baseball, those are the numbers we’ll fixate on.

Pre-injury, Williams saw a marginal, nearly negligible increase in average fastball release speed, going from 92.15 to 92.64 mph. Meanwhile, the average release speed of his slider increased from 81.45 to 84.3 mph, a nearly three miles per hour increase over the course of two months.

So, what was the average release speed for those two important pitches after his injury? For his June 19 return, his fastball averaged 91.77, while the slider clocked in at 83.52. Both of those numbers are below what we came to expect Williams to average, but nothing too out of the ordinary and shouldn’t really raise any red flags.

By the end of the season, his fastball climbed up to 92.46 mph, while his slider actually regressed somewhat to 83.11 mph. Eventually, the fastball and slider came to rest at approximately the average of those two pitches during the pre-injury, three-month run.

Because the average release speeds haven’t produced any overtly valuable information, it’s important to begin to look elsewhere. Looking at Brooks Baseball’s horizontal movement data on Williams’ slider, it actually showed signs of decreasing before the injury. The April 2019 average horizontal movement (in inches) was 4.30 but then decreased in May to 3.98. Once he returned, the dip continued, going from 3.89 in June to finally rounding out at 3.49 in September.

Could this decrease in slider horizontal movement be a big contributor to Williams’ regression after his injury? It’s hard to say. When you begin to break down the post-injury data game-by-game, the picture becomes even more complicated. We’ll continue to look at Williams’ horizontal slider movement.

The above chart is Williams’ horizontal slider movement by game after his injury. While for many of the starts the movement stayed relatively stable, there are two obvious outliers. First, the August 29 game at Colorado. The dip there is the most noticeable, and it registers as his biggest movement downgrade of the season. How did Williams pitch during that game in the hitter-friendly Coors? He got the win after tossing seven innings, while only allowing three runs on seven hits, while picking up six strikeouts. If you hypothesized that a reduction in slider break would yield a decrease in performance, that game isn’t exactly a promising data point. That game, of course, is one sample of many — the greater the data set, the more accurate the picture.

As we know, a pitch’s location may be the most important variable in considering the effectiveness of a pitch. According to Baseball Savant, during that game, Williams threw 12 sliders on the edge or out of the zone. Of those 12 pitches, only three resulted in a ball.

The other outlier would be his final start of the season, an afternoon game at PNC Park against the Reds. In that game, he managed seven innings, allowing two runs on five hits, while walking a pair and picking up five punch outs. During that game, Williams only got batters to chase a slider out of the zone one out of six times. Despite that, he still managed to pitch effectively. He was, however, able to induce swings and misses and weak contact on fastballs up in the zone.

Let’s compare that to the Detroit game when he first returned, where the slider horizontal movement was a little bit closer to his average, which might provide us with a somewhat more accurate point of analysis. In that game, Williams only threw four sliders out of the zone, and only two of those resulted in a strike. The rest of his slider attempts were in the zone. This suggests that not only was he failing to properly locate the slider, he wasn’t very effective on piquing batter interest in the pitch.

The only thing that’s been made clear through all of this is that seemingly independent of slider movement, Williams’ results have been an amalgamation to produce a mixed bag.

So, could Trevor Williams’ lack of productivity come from a reduction in the movement and the effectiveness of his slider? Right now the results are inconclusive; it may be a case of spotting the pitch effectively, or it could be fastball dependent. Or could it possibly have been a mental affliction after having been injured, completely irrespective of Williams’ “stuff” on the mound?

On April 6, The Athletic hosted a Q&A event between readers and Williams. When asked about his lack of production post-injury, despite his promising numbers pre-injury, Williams had the following to say:

“During the rehab process I was so focused on how my side felt and not my overall body. Everything felt great but I was creating bad habits subconsciously in my bullpens because (sic) for whatever reason (sic) my body was trying to protect my side. Unfortunately (sic) it lasted most of my starts when I came back.”

Trevor Williams has been an odd case in trying to determine and then analyze his success. For example, according to Baseball Savant, his fastball spin is only in the 22^nd percentile, while his fastball velocity is in the 26^th percentile. These can be concerning numbers in a league that’s becoming dominated by the high fastball. For that pitch to be effective, it generally requires a high velocity, or at the very least, some amount of deception.

How did Williams do with the high fastball in his successful 2018 campaign? Here are some graphs. The first of which shows swing rate on fastballs, the second shows whiff rate on fastballs, while the final graph shows slugging percentage against the fastball.

In 2018, Williams did a pretty good job of working the zone up near the top, generating a large amount of swings, and a decent percentage of whiffs, while keeping hitters somewhat at bay when they did make contact.

How did his charts look in 2019? Here are the same three graphs for the ’19 season:

What’s interesting with these charts is that for 2018 and 2019, both swing rate and whiff rate are fairly similar; in fact, the whiff rate in a lot of those higher strike zones is better in 2019 than 2018, while still generating approximately the same amount of swings; meanwhile, the slugging percentage, mostly for lefties, increased bountifully in 2019.

In 2018, for the categories xwOBA, xBA, and xSLG, according to Baseball Savant, Trevor Williams fell between the 40^th and 50^th percentiles. Despite that, his 2018 FIP (3.86) is considered above average. Meanwhile, in 2019, his 12^th to 19^th percentile rankings in those three categories fall much more closely to his 5.12 FIP.

---

So, should we expect Trevor Williams to produce like he did in 2018? When taking all the relevant data points into consideration, probably not. Even for 2018, his xFIP was 4.54, which is a far cry from his 3.11 season ERA. Furthermore, FanGraphs’ three-year ZiPS projections have Williams performing consistently, but not quite near his 2018 highs. With FIP totals ranging from 4.26 to 4.29, the projections have Williams as a roughly two win per year player.

An odd case to accurately predict and assess, Williams has had some highs and lows with the Pirates, and if there’s anything that can be garnered from that track record, it’s that we’re likely to see much more of the same moving forward; which is to say, mild unpredictability and a slew of different outcomes.

But in the end, I think the question is a lot more nuanced than that. Can Trevor Williams be a top-of-the-line, marquee starter? Probably not. Can he be a serviceable big leaguer that tosses some really good games? That is where I would put my money. I would expect Williams to perform at the level of a three or four in the rotation. While his pitch-to-contact philosophy doesn’t bode tremendously well in the modern game, Williams has shown enough to prove he has the ability to stick around in the major leagues.

As Dan Szymborski of FanGraphs recently tweeted, “...all models are wrong, some are useful.” Since baseball can be such a finicky, hard-to-predict sport, maybe we’ll get much more of ’18 Williams than we thought. But maybe the predictions are right.