Sorry, but that would be kind of a bummer. Our topic today is the crushing weight of statistical determinism. In researching this article, I learned something that increased my knowledge but also diminished my sense of what was possible, and made me feel a little sad. I would now like to share my grief with you. We will study Victor Robles problem.
You may not remember the days when Victor Robles was a potential star. After short and impressive stints in 2017 and 18, he had a breakout season in 2019, racking up 92 wRC+ and 3.5 WAR, and finishing 6th in NL Rookie of the Year voting. ZiPS predicts him for 3.3 WAR in 2020. If he can take the next step offensively, he’ll be a superstar; If his offense remains just below average, he will still be a very productive midfielder. Instead, he turned in three consecutive seasons with a WRC+ under 70. Here are the Statcast gearboxes for the rookie season and 2022:
There is a lot of blue in the top two rows. Tablet discipline was a primary concern when Robles He was called for the first time, This was certainly a problem, but the lack of strength stands out much more. Although his peak strikeout speed indicates he has the ability to hit the ball hard, Robles’ average strikeout speed has been in the first percentage in each of his big-league seasons, and his strikeout rate has been no better than the fifth percentile. Victor Robles’ problem is a question: Can a player who didn’t hit the ball hard as a beginner turn into a good hitter?
Here’s a quick refresher on why hitting the ball hard is a good thing. There have been 620 players with at least 400 balls in the Statcast era. They are grouped in the table below by average exit speed (EV) in 1-mph increments. The first line displays the wRC+ average, and the second line shows the percentage of players with a wRC+ of 100 or higher.
Average exit speed and wRC+
|wRC average +||71||77||86||85||92||97||105||108||119||131|
It is very difficult to be a good hitter if you don’t hit the ball hard. I collected data on each rookie since 2015 and compared it to their stats in the following years (hereinafter their veteran stats). As a note, rookie stats include full rookie player eligibility. For example, Robles included his coffee cups in 2017 and 2018.
As I often do, I started looking at correlation coefficients. It’s a quick way to get a sense of what’s connected, as well as how sample size affects the data. Below is a table showing the relationship between rookie performance and veteran performance across three stats. It also divides them into buckets with different minimum ball in the junior play events. Each group has at least 100 BIP as veterans:
The relationship between the stats of juniors and veterans
|Minimum Apprentice BIP||40||100||200||300|
|Average exit speed||.73||.69||.78||.81|
You’re probably not surprised that checkout speed and swipe rate stabilize quickly, or that they’re more subjectively predictive of time than wRC+. However, you might be surprised at the power of the association. If a player puts at least 300 balls into play as a rookie, we can predict his future power stroke rate with an astonishingly high degree of accuracy. The average player increases his damage rate by just over 1% after his rookie year. If we limit ourselves to players with at least 300 BIP as a beginner, only one player has increased his damage rate by more than 10%: Vladimir Guerrero Jr.which the entire baseball world expected to improve after he had a disappointing (though certainly not bad) 106 wRC+ as a 20-year-old.
Now is the time for the big question. We’re looking at how a rookie’s hit rate correlates strongly with his overall hitting skill as a veteran. We’ll take the same rookie stats (and add the max exit speed) and check if they relate to the player’s wRC+ as a veteran:
The relationship between wRC+ rookie and veteran stats
|Rookie BIP Min||40||100||200||250||300|
|Average exit speed||.44||.41||.55||.58||.66|
|maximum exit speed||.39||.39||.54||.43||.53|
EV and wRC+ statistics have a similar relationship even after only 40 BIP, but EV statistics improve further with increasing sample size. If you want to know how successful a rookie will be in the future, you might be better off ignoring how good he is now and just looking at how hard he hits the ball. Again, this may not surprise you, but it does show how stable the EV and rate are strongly affected, and how much they affect wRC+ on a longer sample.
Now that we’ve established the importance and immutability of exit velocity, it’s time to smooth ourselves with the icy touch of numerical certainty and condemn some newbies this year to the realm of eternal weak connection. Here are the rookie’s stats Geraldo Perdomo And the Stephen Kwan:
Geraldo Perdomo and Stephen Cowan – Rookie stats
|player||Palestinian Authority||BIP||EV||maxEV||h h%||BB%||K%||AVG||OBP||SLG||wRC+|
Perdomo and Cowan have had very different seasons, but they do have a few things in common. They’re the rookies who’ve played all year (and in Perdomo’s case, a bit of 2021), and hit the ball very softly. Both were original prospects who entered the season with questions about their strength (despite Kwan’s hit tool being rated elite). What jumps out at me is that Kwan’s WRC+ is twice as much as Perdomo’s, although Perdomo’s worst-hit rate is much higher. It doesn’t help to write off twice as much, but a second factor is at play here. The Perdomo had a higher hit rate, but the EV was identical. Here’s how that’s possible:
Perdomo & Kwan – Speed Exit
|player||strong hit||Not a hard hit||Babeb|
Kwan hits the ball hard even when he’s not hitting the ball hard. This helps explain why its BABIP score is 74 points higher than Perdomo’s. When Perdomo isn’t hitting the ball hard, his EV is only slightly better than the 75.3 mph that Victor Robles set as a rookie. Turning to Victor Robles is certainly not the way to solve the Victor Robles problem.
On balls that aren’t badly damaged, Kwan’s EV is much higher, as are wOBA and xwOBA. Cowan led the league with 99 hits between 90 and 95 mph. At the league level, those medium-hitting balls have a boba of .269, nowhere near .649 from hard-hitting balls, but certainly much better than .204 from balls hit below 90 mph. It’s definitely better to hit the ball hard, but it’s still a good idea to avoid hitting the ball weakly.
Cowan shares his penchant for middle contact with a very short list of players who have fared well as veterans despite extremely low hitting averages as rookies. From 2015-21, there were 75 starters with an injury rate of less than 28% (with a minimum of 40 BIP). Only nine of those players have gone on to have a MVP+ above 100. On not-hit balls, all nine had an EV above the league average of 80.7 mph:
100+ wRC veteran + despite the low EV
|player||rwRC+||rEV||rEV (other than HH)||rHH%||rmaxEV||vwRC+||vEV||VHH%|
Some of these players, like Max Muncy, have seen big jumps in their overall EV stats. They had high average hit rates as rookies not because they weren’t strong enough to hit the ball hard, but because there were too many balls they missed to squash. A few select players, like Arraez, Kemp, and Narváez, don’t have that kind of power, but they’ve managed to succeed as seasoned veterans by relying on medium-hitting balls. Unfortunately, this is a very difficult tightrope to walk:
Exit velocity correlates with wOBA
|total exit speed||.71||.69||.70||.76|
|Hard exit speed||.64||.68||.70||.73|
|Not strong exit speed||.19||.02||-06||-06|
The first row reminds us that it is better to hit the ball harder. Row two says that even if you’ve already crossed the threshold into a hard-hit area, every last bit of speed makes a difference. It’s the last row that isn’t entirely intuitive.
During a short sample, players who have a high EV on balls that aren’t hit hard tend to perform better. However, when we increase the number of balls in play that are required to be in the sample, the small correlation there turns to negative. Players who make the most of medium-hitting balls don’t fare well in the long run. This type of ball can fall in one place, but it requires luck in hitting the ball, which equalizes over time. Regardless of sample size, the correlation between a player’s average hit rate as a novice and as a warrior is 0.31, much weaker than the average hit rate. If getting to base in this way is a repeatable skill, the only current players who really seem to have it are Arraez and Kemp.
Among this year’s rookie group, Cowan appears to be the best candidate to join this club, along with the Baltimore club Terenne Favra. The Vavra EV had an undamaged 83.5 mph and a WRC+ 97, though with a severely damaged rating of only 23.6%. However, if Kwan and Vavra do not increase their overall strength, they will always be more dependent than most players on the luck of the hit ball. Kwan’s wOBACON was 37 points higher than his team’s xwOBACON, the 17th highest differential in the league, so he may be due for a visit from a monster of regression. As for Robles and Perdomo, it may be time to leave them in the cruel hands of fate.
I’d like to believe I’m wrong here, and that players have a better chance of increasing their EV and damage rate severely than the numbers indicate so far. We only have eight years of exit velocity data, and other patterns may emerge after a few more seasons. This is also an era when players have more tools than ever before to reconfigure their offensive profiles. Today’s players receive more instruction at the big-league level and have access to outside coaches and hitting labs, so there are more resources for the player to rework their swing and release more power. For now, though, there doesn’t seem to be a great answer to Victor Robles’ problem.
Thanks to Mike Petrillo, who encouraged me to research this topic with a text that started, “I’ve got a story idea for you!” And he almost certainly didn’t expect me to run with Victor Robles Trouble as the title.