The Value of Versatility
Takeaways from Fearing’s research on positional flexibility.
In baseball, as with many other elements in life, it is important to be flexible, as even with all the impressive predictive tools we have at our disposal, the “variance gods” of the universe frequently generate unforeseen events. Even with the advent and continued development of biomechanical analysis in Major League Baseball, injuries are difficult to predict (especially on the public side), and all organizations must have multiple contingency plans in place to prepare for all potential scenarios.
In my opinion, one of the most important pieces of analytical research in the past decade or so has been Doug Fearing’s effort to quantify the effects of positional flexibility and organization’s robustness to injury, in his research paper “Process Flexibility in Baseball: The Value of Positional Flexibility”, co-authored with Timothy C.Y. Chan. In the research paper, Fearing and Chan utilize operations management principles to model the effects of positional flexibility as a supply chain and identify which teams are the most robust to potential injuries.
Fearing’s research into positional flexibility has had a considerable effect on the way in which Major League Baseball teams construct the rosters, with many analysts providing their own insights into how to utilize and improve upon the findings of the research paper. Joe Tourville posted an excellent summary of Fearing’s research on his Medium blog last year, along with his takeaways from the research, and I would highly recommend reading Tourville’s summary (or the paper itself) before continuing with this article. This article will discuss, in a conversational format, the findings of Fearing’s research, methods upon which I believe the research could be improved, and, complimenting Tourville’s article, the implications of positional flexibility on roster construction in Major League Baseball today.
Findings:
First off, here is the abstract from the research paper on positional flexibility:
“This paper introduces the formal study of process flexibility to the novel domain of sports analytics. In baseball, positional flexibility is the analogous concept to process flexibility from manufacturing. We study the flexibility of players (plants) on a baseball team who produce innings-played at different positions (products). We develop models and metrics to evaluate expected and worst-case performances under injury risk (capacity uncertainty) with continuous player-position capabilities. Using Major League Baseball data, we quantify the impact of flexibility on team and individual performance, and explore the player chains that arise when injuries occur. We discover that top teams can attribute at least one to two wins per season to flexibility alone, generally due to long subchains in the infield or outfield. The least robust teams to worst-case injury, those whose performance is driven by one or two star players, are over four times as fragile as the most robust teams. We evaluate several aspects of individual flexibility, such as how much value individual players bring to their team in terms of average and worst-case performance. Finally, we demonstrate the generalizability of our framework for player evaluation by quantifying the value of potential free agent additions and uncovering the true “MVP” of a team.”
To evaluate the true impact of positional flexibility, Fearing and Chan developed a series of advanced metrics rooted in operations research and optimization theory. At the foundation of their model is the concept of continuous player-position capability, which estimates the expected run value that player j provides when playing position i. This metric moves beyond binary assumptions (ex. “This player can/cannot play this position), and quantifies performance at each position in terms of runs above replacement per game, incorporating both offensive and defensive contributions. By using this foundation, the model is able to assign optimal playing time values to each player at each position to maximize total team performance, while also accounting for player capacity, which adjusts for injury risk and availability.
From this framework, they introduce two central metrics to evaluate flexibility at the team level: Average Value of Flexibility (AVF) and Robust Value of Flexibility (RVF). AVF captures the expected gains in performance from positional flexibility under normal conditions, while RVF quantifies a team’s resilience to worst-case injury scenarios. Both metrics can be computed in two forms, AVF⁺/RVF⁺ to measure the marginal value of adding flexibility, and AVF⁻/RVF⁻ to assess the share of current performance already attributable to existing flexibility. Another metric, Protection Level (PL), estimates how many injuries a team can sustain before suffering a meaningful drop in performance. These metrics not only provide a nuanced assessment of roster construction and injury preparedness but also allow for player-level evaluations of who contributes most to a team’s flexibility profile.
Using these metrics, Fearing and Chan revealed several important insights about roster construction and team performance. Their analysis showed that teams could attribute between 0.7 and 2.7 wins per season to positional flexibility alone, often the margin between a playoff berth and missing the Postseason entirely. Conducting their analysis on data from the 2012 season, the most-flexibility-optimized teams, the Chicago Cubs and Milwaukee Brewers, benefited from long and interconnected “subchains” of players who could competently shift across the infield or outfield without a major performance drop. In contrast, teams like the Baltimore Orioles, which lacked such flexibility, were far more vulnerable to performance decline when injuries struck. Notably, RVF uncovered that the least resilient teams were over four times as fragile as the most robust ones in worst-case injury scenarios.
Implications:
Upon reading the research paper, I initially tried to re-create Fearing’s methodology, as I believe the research could use an update, considering the metrics created in the paper were utilized to evaluate players and teams from the 2012 Major League Baseball season. Not only would re-creating the methodology allow us to identify the most versatile players in the league since 2013, but with the subsequent advancements in tracking technology, perhaps incorporating Statcast metrics into the framework of the models could improve the performance of the metrics, particularly as it pertains to evaluating defense. I realized pretty quickly that the methods in which Fearing created his versatility framework were quite above my current skill level, and considering that some of the data involved in the research is not publically available (such as detailed injured list records), I decided to shelve this idea for the foreseeable future. Nevertheless, there are a few important takeaways and potential improvements to this research that I have not seen discussed elsewhere that I believe can add additional value to evaluating players through Fearing’s versatility framework.
As mentioned earlier, it seems likely that the inclusion of Statcast data into this methodology would provide more insight into evaluating which players provide more flexibility value than others. Since 2012, new Statcast-based defensive metrics such as catch probability, outs above average, and defensive run value, have been made publicly available, which provide more insight into a player’s defensive performance than “traditional” metrics such as defensive runs saved or ultimate zone rating. In addition, athleticism-based metrics, such as sprint speed, can be used to better understand and predict how well a player’s defensive abilities will transfer to a secondary position. For example, Fearing’s research indicates that during his free agency, part of Albert Pujols’s value derived from his ability to play third base (which he started 4 games in 2011, the first time playing the position since 2002), however, it seems unlikely that this was a correct assumption to make, considering Pujols’s advanced age and declining athleticism. Hypothetically speaking (and of course, I am making this statement with the benefit of hindsight), the inclusion of athleticism-based metrics would likely indicate that Pujols was likely to be a DH long-term, not a player who could cover 3B in a pinch.
The inclusion of Statcast data into this analysis would also improve the predictive ability of offensive production utilized in the versatility framework. Since 2015, numerous metrics to evaluate offensive production have been created using Statcast data such as xwOBA, Barrel%, and Bat Speed, and these metrics would likely capture a player’s “true talent level” better than simply using a hitter’s past production, as measured by the framework’s offensive model.
Being able to attribute ~1-2 wins a season to versatility alone is a significant finding of the paper since only a couple of regular season wins can make the difference between a team that misses the playoffs and a team that just makes the playoffs and makes a World Series run. Since the ability to be versatile is a valuable attribute to any Major League team, there is potentially an inefficiency in WAR, because the metric does not directly take into account how a player fits within a team’s subchain in a given season. A player whose versatility allows for the team to more consistently produce their best possible lineup is perhaps underrated by WAR (which only takes into account a player’s offensive, defensive, and baserunning contributions), and the same can be said for a player whose absence would cause a team’s subchain to fall apart and result in a sharp decline in total production.
Current WAR metrics from FanGraphs (fWAR) and Baseball-Reference (bWAR) do not take into account these versatility factors, and it does make sense as to why they do not incorporate these factors into their calculations. The general public’s utility of WAR is to compare players to each other, across different organizations, and the addition of versatility factors to the metric’s calculation would likely complicate the metric’s interpretability for casual fans who simply use the metric for fantasy baseball and/or argumentative purposes. For teams, however, the use of a versatility-infused WAR (vWAR) would provide immense value, as it would better encapsulate each player’s importance to the team’s player subchain. The use of versatility-infused WAR alongside traditional WAR could also be useful in evaluating players on the free agent market, as contracts are largely based on WAR projections, and a player who has a vWAR greater than their WAR will potentially be undervalued by the market.
One shortcoming of incorporating versatility factors into WAR is that I would imagine it would be very difficult to project each player’s future vWAR. Major League rosters are constantly changing, and it is difficult to project how a roster’s infield and outfield subchains will look three weeks from now, let alone three years from now. However, I believe that putting “a couple of cents on the scale” of a player’s WAR to reflect their versatility ability (for example, add a fraction of WAR to the future projections of José Caballero, Josh Smith, and Maikel Garcia) would properly encapsulate a player’s value to the roster’s chain.
In my opinion, the findings of Fearing’s versatility should also have an impact on player development and draft strategy. Currently, most teams value selecting a player who plays an “up the middle” position with an early selection in the draft. This is because players who play shortstop and/or center field tend to be the most athletic players on the field, and therefore have multiple paths to generate production once they arrive at the Major League level. In contrast, prospects who play corner positions tend to be “riskier” prospects at the top of the draft, because a majority of their value is concentrated in one area, generating offensive production.
Given the value that is provided to a team via a player’s ability to play multiple positions, I believe that placing a focus on teaching players new positions can be a priority in player development. If a prospect does not have a good chance at staying at shortstop, center field, or catcher once they reach the Major League level, then I believe they should focus on attempting to play as many secondary positions as possible, to improve the potential value they can provide upon promotion to the Major League team. Perhaps this is “easier said than done” as prospects have a lot of elements to their game they have to focus on developing at the Minor League level, however, given the demonstrated value positional flexibility provides to the team, I believe that resources should be allocated to developing this skill.
Fearing’s research on positional versatility has made a large impact on how Major League rosters are constructed, and one team that clearly displays an emphasis on positional flexibility is the Los Angeles Dodgers. Fearing was the R&D director for the Dodgers until 2018, and his emphasis on versatility has been exhibited throughout the entire roster, with Los Angeles historically rostering “utility players” such as Enrique Hernández, Chris Taylor, Tommy Edman, and Hyeseong Kim. Outside of these “utility players”, the Dodgers have also historically been comfortably utilizing players such as Max Muncy, Mookie Betts, and Cody Bellinger at multiple positions, further strengthening their infield and outfield subchains. The presence of multiple versatile players on the roster allows for the Dodgers to also have more offense-focused players on their roster who provide little defensive value, such as Teoscar Hernandez, as their roster has enough flexibility to remove these players late in games (when defensive value is more important), and they are always in a position to sign the best player available in free agency as (other than their financial resources), they can reconfigure their lineup to integrate the free agent into their player chain, due to their roster’s flexibility.
The Boston Red Sox are another team that has valued positional flexibility in recent seasons, with their current roster containing players such as Romy Gonzalez, David Hamilton, and Ceddanne Rafaela, who have each played at multiple positions over the past couple seasons. Understanding the value of positional flexibility and roster subchains can also help to better contextualize the effects that the Rafael Devers trade will have on the Red Sox roster over the remainder of the season. While on the surface, a return of Jordan Hicks, Kyle Harrison, James Tibbs, and Jose Bello may seem underwhelming, there is a clear path that exists to replace most of Devers’s lost production when viewing the trade through the process flexibility framework.
Opening up the DH position allows for Rob Refsnyder and Masataka Yoshida to potentially share a platoon at the position, and in the aggregate, both players are capable of replacing a large share of Devers’s offensive production. Starting Refsnyder and Yoshida (both “bad” defenders) at DH will allow for a more capable defender to start in left field (most likely Jarren Duran or Roman Anthony) which will improve the team’s defensive production. Kristian Campbell could also spend some time at DH as well, removing another underwhelming defensive player from the field, and giving a better defensive player (David Hamilton) more playing time at second base. Looking at the trade through this process flexibility framework, it is easy to envision a scenario where the Red Sox will be able to cover at least 80-90% of Devers’s “lost” overall production.
With the absence of robust, up-to-date flexibility metrics this article has waded into speculative territory at times, however, it is apparent that evaluating how a player affects a team’s robustness to injury via their versatility value is an important factor to take into consideration when constructing a Major League roster. Organizations that successfully integrate flexibility considerations into their roster-building process can gain a strategic edge, not just in injury scenarios but also in daily lineup optimization, and I believe that it is imperative that versatility and positional flexibility be viewed as a core roster-building principle. In conclusion, players who are able to adapt to changing needs and play multiple positions are more valuable than their surface-level statistics might suggest, and I am hopeful that with the emergence of new metrics in the future, we will be able to better understand the value of versatility.
Thanks for reading!
Follow @MLBDailyStats_ on X and Adam Salorio on Substack for more in-depth MLB analysis. Photo credits to Jayne Kamin-Oncea, Ezra Shaw, and Orlando Ramirez.
Link to Fearing and Chan’s research:
Link to Tourville’s summary:
https://joetourville.medium.com/engaging-with-fearings-player-flexibility-framework-d8eefa2f51af
Citation:
Chan, T. C. Y., & Fearing, D. (2019). Process flexibility in baseball: The
value of positional flexibility. Management Science, 65(4), 1642–1666.
10.1287/mnsc.2017.3004