Peter Majors
On November 17th, the BBWAA announced the winner of the 2021 National League Cy Young Award, sending the baseball world into a frenzy. Brewers ace Corbin Burnes beat out the Phillies' Zack Wheeler and Dodgers’ Max Scherzer for the award which recognizes the league’s best pitcher.
Both Burnes and Wheeler received 12 first-place votes. However, the Brewer’s ace edged out the Phillies’ in terms of second-place votes, which allowed him to take home the award by a slim 10-point margin
Despite Burnes’ indisputable dominance on the mound, he threw far fewer innings than second-place vote-getter Zack Wheeler. This difference was so large, in fact, that it called into question the proper weight of two competing factors in Cy Young voting: volume and efficiency.
As it almost always does, the debate took to Twitter, where some of the most prominent names in baseball voiced their opinion on the matter. As the Tweet below makes clear, ESPN’s Jeff Passan chose to do so quite bluntly.
As you may have been able to guess, I agree wholeheartedly with the first half of Jeff’s tweet. However, the second half is where our differences in opinion arise. Despite his harsh recommendation, gravel won’t be on my dinner plate anytime soon. Luckily, Jeff's opinion wasn’t the only one out there - as many similarly important figures took the opposite stance. The battle raged on throughout the night and following days, touching every corner of baseball media.
Regardless of any one person's opinion, the voters are ultimately the ones responsible for determining what it truly means to be the “best” pitcher in the league. This awards season, they chose to reward a low-volume high performer. So let’s explore why they did so, as well as the historical precedent for a decision of this nature.
Figure 1 (Green Indicates Best Among Group)
Looking at rate statistics alone, it is clear that Corbin Burnes had the best season of any qualified starting pitcher in the league. Burnes’ sharp 95 mph cutter, which he threw 52 % the time, along with his lethal curveball are what propelled him to such incredible success. His absurd 1.63 FIP was the result of a 30.2 K-BB% and a 0.38 HR/9 (2nd-best among qualified pitchers since 2015). These peripherals suggest that he somehow should’ve allowed fewer runs than he did.
In most years, 28 games started to the tune of a 2.43 ERA would result in a slam dunk league Cy Young Award. However, for Burnes this season, that was not the case. Out of the 112 Cy Young-winning seasons that qualified for an ERA title with at least one game started, Corbin Burnes’ 2021 ranked second to last in terms of innings per start at 5.96.
This was largely due to a combination of two factors: his ineffectiveness the third time around the batting order and the strength of Milwaukees’ bullpen. Burnes’ low mark for Cy Young winners only came ahead of Blake Snell’s 2018 campaign, in which the left-hander averaged 5.81 innings per start over 31 games and 180 ⅔ innings. Knowing what we know now about how the Rays utilize their arms, this is hardly a surprise.
That awards season, first-place votes were split between Houston’s Justin Verlander and the aforementioned Snell. While Verlander sported a 2.52 ERA over 214 innings, Snell owned a 1.89 ERA in 33 ⅓ fewer innings. At first glance, this tradeoff between volume and effectiveness is eerily similar to that of Burnes and Wheeler. However, as the table shows, these discrepancies were considerably less than 2021's dichotomous pairing.
Figure 2
What made 2018’s results so much less controversial than 2021’s was just how much more effective Blake Snell was than his counterpart in terms of run prevention. With a greater difference in ERA, but a slimmer one in IP, voters were tasked with deciding between two less extreme options four years ago.
In both 2018 and 2021, a Cy Young award went to the finalist with the best rate statistics, not to the one with the best combination of innings and rate statistics. While voters are not always tasked with deciding between candidates at either end of the volume and efficiency spectrums, the results of these years stand out because of their recency and proximity to one another.
Do these recent results signal a change in voting style or are they the consequence of some other, unaccounted for, factor? To find out, let’s first explore how innings totals have changed over the decades.
Volume
Below is a line graph depicting the average number of innings thrown by all pitchers who qualified for an ERA title (blue) along with the average number of innings thrown by Cy Young winners who also qualified for an ERA title (green line) by season.
Figure 3
Noticeably excluded is the mean inning total from the 1984 Cy Young winners, Willie Hernandez and Rick Sutcliffe since neither pitcher threw enough innings to qualify for the season’s ERA title. In order to qualify, a pitcher must throw the same number of innings as their team competes in. Thus, in a non-strike or non-Covid interrupted season, a pitcher must throw at least 162 innings to be in the running for an ERA title.
It is important to note that this minimum threshold has increased ever so slightly with time, as a handful of games have been added to each team’s schedule since the advent of the award in 1956. The threshold can also vary from team to team, as in some years, teams do not play in all of their scheduled games. Regardless, the purpose of this minimum remains the same: to put all pitchers on an even playing field for the sake of comparison.
I should also mention that the Cy Young winners not aforementioned who failed to qualify for the ERA title in their respective seasons include: Sparky Lyle ‘77, Bruce Sutter ‘79, Rollie Fingers ‘81, Steve Bedrosian ‘87, Mark Davis ‘89, Dennis Eckersley ‘92, and Eric Gagne ‘03. I decided to exclude non-qualifiers, which are mostly seasons by relief pitchers. For their seasons on the green line in the above graph, the inning total was taken from the pitcher in the other league who did qualify.
Now that those technicalities have been taken care of, the above graph tells us something fascinating about Cy Young winners over time: their inning totals are becoming more and more in line with the league average. While qualified starters are decreasing their innings each year by a rate of -.74 IP per season, Cy Young winners are doing so at a rate of -1.59 IP per season. These two linear trend lines project to converge in 2050 at 159 innings pitched. Unless the parameters surrounding this threshold are altered, there will be extremely few pitchers who meet it in just a few decades.
Below is another line graph, this time detailing the percentile of average innings thrown by the Cy Young winner(s) out of the maximum innings thrown among all pitchers (green) and the standard deviation of innings among all qualified pitchers for each season (blue).
Figure 4
Here, we can observe two important phenomena. First, the standard deviation in innings pitched among those who qualify for an ERA title has decreased dramatically with time. In 1972, it peaked at 49.5, while in 2020, it bottomed out at 5.6 due to the Covid-shortened season. However, this should have been expected given pitchers are throwing fewer innings than they ever have while the parameters to become a qualifier have remained unchanged. This means fewer pitchers are qualifying nowadays, and the ones who do are throwing fewer innings than their predecessors.
Second, since the turn of the 21st century, the percentile of innings pitched among Cy Young winners has decreased, meaning voters have become less concerned with a pitcher’s inning total with respect to the league’s inning leader. The last time the average percentile for Cy Young winners was within the same realm as 2021 was in 1974. That season, Athletics ace Catfish Hunter (318 ⅓, IP, 95th percentile) and Dodgers super-reliever Mike Marshall (208 ⅓ IP, 29th percentile) combined to land in the 62nd percentile.
Given the pitching climate at the time, Marshall’s victory must’ve been an absolute shock to the baseball world. That season, legendary knuckleballer Phil Niekro threw to a 2.38 ERA in 302 ⅓ IP, besting the eventual winner in both categories. Not that it is relevant for today’s voters, but Niekro also won 20 games that season - a marker of immense pitching prowess at the time.
What makes the relief pitcher’s victory different from Burnes’ is that nowadays, pitchers are throwing far fewer innings, and the variance between qualified pitcher’s inning totals is a fraction of what it was half a century ago. Voters simply do not have as many high-volume options to choose from, and as a result, are tending to side with efficiency metrics to make their decisions.
In 1974, the same year Marshall won NL Cy Young, flamethrower Nolan Ryan led the sport with 332 ⅔ IP. While in 2014, David Price led all of baseball with 248 ⅓ IP, and 31 pitchers threw at least 200 innings. To better conceptualize this precipitous decline in innings pitched, here is another graph, this time showing the number of pitchers who threw at least 200 innings (blue) and those who tallied less than that (light green) by season.
Figure 5
From 1961 to 2014, awards voters had at least 30 pitchers to choose from with at least 200 innings for the Cy Young award (barring strike years). However, that all began to change in the mid-2010s as a result of the analytics movement and a greater awareness of the impact high inning totals have on arm health. Since 2015, 86 pitchers have thrown 200 or more innings in a season, which is an average of only 14.3 pitchers per year (excluding 2020). Even though 2021 had a full 162-game slate, only 4 pitchers crossed the 200 inning mark!
How much of this is the result of a shortened slate of games the year prior is tough to say. But given the trends at play, it would not be surprising to see a similar total in 2022.
Efficiency
The other even more important component in Cy Young voting is how efficient a pitcher is at preventing runs. After all, isn’t that the whole goal of pitching? In today’s game, the main two metrics used to evaluate a pitchers’ ability to prevent runs are Earned Run Average (ERA) and Fielding Independent Pitching (FIP).
Similar to FIP, metrics such as xERA, xFIP-, and SIERA utilize the events a pitcher induces to determine what their ERA should be. Unfortunately, the lack of Statcast-esque data prior to 2015 makes these impossible to use for long-term analysis. Thus, I will only be considering ERA and FIP in this section.
To begin, let’s take a look at the percentile of average FIPs (orange) and ERAs (blue) among qualified Cy Young winners with respect to the rest of the league.
Figure 6
As the above graph makes clear, in almost every season ERA has been a critical component in winning the Cy Young Award. However, as the positive slope on the blue linear trend line indicates, it has become increasingly relevant with time.
The more interesting revelation here, though, is the increased relevance of FIP, or rather its various components, over time. By various components, I am referring to strikeouts, unintentional walks, home runs, and hit-by-pitches. FIP was not invented until the early years of the 21st century, so the exact metric was not available during this transition. However, its aforementioned components were surely under consideration to some extent by voters before its inception.
Prior to 1991, there were several seasons where the percentile of average FIP was significantly less than that of ERA, most notably 1958 (Turley), 1976 (Palmer, Jones), and 1990 (Welch, Drabek). These outcomes suggest that earned run prevention, not the outcomes pitchers were in control of, were the favored efficiency metrics for Cy Young winners those seasons.
In the 31 seasons since, there has only been one such occurrence of a 20+ percentile point difference: 2021. This was due to Robbie Ray’s 3.69 FIP, which placed him in the 43rd percentile of qualifiers. Though, given recent voting history, this outlier can hardly be seen as indicative of a greater trend.
If the above graph was separated into two time periods, one ending in 1990 and the other beginning in 1991, below are the segments’ mean and variance statistics broken down into one table.
Figure 7
Here we see the exact change between “old school” and “new school” voting trends with respect to efficiency metrics. As Figure 6 suggests, FIP underwent the greatest change between the two periods. The table above supports this, as between 1991 and 2021 the mean percentile of FIP was practically in line with that of ERA.
What is the most relevant part of the above table is the “ERA - FIP Pctl” row, which describes the mean difference between ERA and FIP percentiles by season. For the first 35 years of voting, ERA was on average .078 percentiles higher than FIP. However, over the subsequent 31 years, that figure fell a full 74% to .020. While it seems unlikely that FIP will ever surpass ERA in its relevance for awards voting, this change is certainly telling.
Now that the efficiency metrics of ERA and FIP have been properly explored, let’s summarize the results of these findings.
Conclusion
If one were to look at each of the charts and tables in this study, one might conclude one of many things. The worst of which would be the following: voters care less about volume when looking at Cy Young candidates and now prefer pitchers with better efficiency metrics. Admittedly, this was my hypothesis heading into this project.
The question posed just before the volume section of this piece asked whether Corbin Burnes and Blake Snell’s Cy Young Awards signal a change among the philosophy of voters or are the result of something else. Based on my research, it seems that the second of these is the far more likely scenario.
In the 66 years of Cy Young voting, Major League Baseball has changed dramatically. Pitchers are making fewer starts per season, are averaging fewer innings per start, and as a result, are throwing fewer innings each season. At the same time, the criteria to become a qualified starting pitcher remained the same: to throw at least the same number of innings as games one’s team plays during the season.
Naturally, this decreased the number of qualified pitchers in the league, making quality choices for the award more and more scarce. As a result, this lowered the standard deviation among innings for qualified pitchers and leveled the playing field for all candidates. Soon after, pitchers with lower inning totals began to win the award. This is likely because having candidates with similar volume reduces the number of factors a voter has to make their decision.
With fewer factors, comes more of a reason to examine efficiency metrics - which is exactly what voters began doing. The most widely accepted run prevention efficiency metric before the turn of the century was ERA. However, the analytics revolution revealed the importance of statistics such as invented FIP, which better isolates which outcomes a pitcher can control. In the '90s, voters relied mostly on ERA, but in the years since have come around to FIP. As a consequence, percentiles of ERA, and to a greater degree FIP, by Cy Young winners rose and began to stabilize in the mid-1990s.
At the same time, the spread between ERA and FIP among winners shrunk dramatically. This means that voters not only selected pitchers who were better at preventing runs, but also selected ones who truly dominated when it came to the three true outcomes: strikeouts, walks, and home runs.
While this project took a macro look at the landscape of Cy Young voting over the years, it can still be applied to recent controversies, such as the 2018 AL and 2021 NL Cy Young results. Looking at Figure 3, we are reminded that inning totals among Cy Young winners have been falling more and more in line with the league average for qualifiers each season. Furthermore, Figure 4 reinforces the idea that a pitcher does not need to be near the league lead in innings to win the award.
In both 2014 (Kershaw, Kluber) and 2018 (Snell, Degrom), the average winner placed in the 75th percentile of qualifiers. Granted, three of these performances were three of the top four seasons by ERA in a non-shortened season of the century. Nevertheless, the precedent for efficiency over volume exists, especially in recent years.
As discussed earlier, Corbin Burnes’ massive outperformance of his peers on efficiency metrics is what earned him his NL Cy Young award. He beat out second-place vote-getter Zack Wheeler in terms of ERA by a substantial .35 points and in terms of FIP by a whopping .96 points. In today’s voting environment, any qualified starter should expect to walk away with the award.
The reality is that innings totals have been, and will continue to, decline around the league. Even if 2021 possessed abnormally low innings totals because pitchers were coming off of a shortened season, the long-term trends are there to support this continual decline.
Whether you like it or not, there will be many more Corbin Burnes-esque Cy Young finishes in the years to come. So voters and fans, in the immortal words of Billy Beane ...
Sources:
Peter Majors is the Founder and President of the Fordham Sports Analytics Society and a junior at Fordham's Gabelli School of Business. He is majoring in Accounting Information Systems, minoring in Computer Science, and plans to attend graduate school for Applied Statistics and Decision-Making. He is currently seeking a Summer 2022 internship in Data Science, Information Systems, or Baseball Analytics.