Pillars of Consistency and Paragons of Change

There will either be a truncated 2020 MLB season or no MLB season at all. While this fact is due to a phenomenon drastically more serious and important than baseball, it gives way to a host of baseball-related considerations. One cannot help but speculate when and where the season may take place, for instance, or how any particular team’s odds of success are adjusted in light of the shortened season. In the case of individual players’ performances, even assuming an adequate “build-up” period that may offer a crude proxy for spring training, there are obvious repercussions as well.

One particular consideration is how the relative “streakiness” of any given player might impact their performance in a sub-162 game season. While generally consistent performers across months might not see their rate-based statistics particularly impacted in a shortened season, those streakier hitters’ tendencies might be exposed, for better or worse.

Of course, for MLB and its players, month-to-month streakiness likely isn’t an overwhelming consideration given all the other variables that will change in 2020 (and beyond). Still, this line of thought did more broadly pique my curiosity regarding players’ relative levels of consistency. As a result, this post steps back from month-to-month variance in player performance, and instead attempts to evaluate the year-to-year streakiness of players in the last decade.

In order to address this question, players who have had 500+ plate appearances in at least five seasons from 2010-2019 have been included. The decision to include players with 5+ eligible seasons wasn’t quite an arbitrary one. It was motivated by the fact that by including only 5+ player-seasons of data, there was essentially no correlation between number of qualifying seasons (anywhere from 5 to 10) and the variance in offensive value. Altogether, 112 players had at least 5 qualifying seasons from 2010-2019. Just a single player, Adam Jones, qualified in all seasons from 2010-2019.

Figure 1 below offers a cumulative view of those players being examined; it visualizes a slight relationship between wRC+ (the offensive metrics of choice from which to evaluate performance) and the wRC+ standard deviation (the metric of choice here to evaluate fluctuations in performance). That relationship is slight though: the R-Squared value indicates that average wRC+ standard deviation explains just .03 (3%) of wRC+.

As aforementioned, wRC+ standard deviation was employed to rank player streakiness across seasons. Below are the top 15 players in highest wRC+ standard deviation, who are represented by the 15 points highest along the y-axis in Figure 1.

There are a couple key takeaways from this leaderboard. A common thread among these players is that they have all achieved considerable success, otherwise their careers wouldn’t feature the relatively massive amount of variance between seasons without coming to an unwilling end. In fact, the lowest maximum wRC+ season above is that of DJ LaMahieu, coming in at 136, or 36% better than league average production. The fact that Chris Davis sits atop this chart signifies the truly massive swings in his production across seasons. His max-min wRC+ of 124 is far and away the largest total in this dataset. This chart can in part capture the enigmatic greatness of Bryce Harper. It also signifies the inevitable decline of all-time greats like Albert Pujols and Miguel Cabrera all the while highlighting the rise of MVPs Christian Yelich and José Altuve.

Below are those players on the other end of the spectrum. You’ll notice that, while they offered consistent production, the average wRC+ figures in this top 15 aren’t so robust across the board as in the case of the high-variance players.

Who would have guessed? Kevin Pillar, a pillar (forgive the pun) of consistency. The average wRC+ across the two tables above indicate a clear selection bias; the fact that the players’ statistics listed immediately above do not vary as much is at least in part a function of less exceptional, if not inconsistent, production. 8 of 15 players listed above have average wRC+ figures between 90 and 110.

There are some obvious exceptions though, led by one Mike Trout. Trout, whose wRC+ has vacillated between just 23 points between each of his 8 qualifying seasons, sits fifth on this leaderboard for consistency while producing 75% more than the league-average player offensively. This chart also rightfully highlights the sustained proficiency of players like Carlos Santana and Brett Gardner, who have been fixtures of this past decade (given their 9 qualifying seasons), yet have just a single all-star appearance apiece.

Playing 5+ qualifying seasons at all is a significant achievement in and of itself. At the end of the day, figures and leaderboards like those above are nice if only for scratching the surface of these players’ achievements and the cadence with which they made them.