The gap between replacement- and non-replacement level players isn’t as wide as you’d expect.
Each of the 17 Sundays of the NFL season begins with me turning on the TV and scrolling the guide. I check the local affiliates as though flipping the cards of a tarot deck to learn what fate the universe’s football gods have dealt me. Since I live half a country away from my preferred team, it rarely appears as an option.
I start with the early games. I sit on the edge of my sofa, mentally begging for something worth watching. The first game I see: Browns vs. Dolphins. Ugh, okay, maybe the other matchup is better. The second game: Bills vs. Jets. I slump back. My shoulders touch the back of the couch, but most of my torso lays flat along the seat.
Maybe the late game will be better. I advance the guide to find 49ers vs.Cardinals awaiting me. Well, at least the Sunday Night Football game has to be worth my while, right? I hit the right arrow button on the remote a few more times. Oh, hey! Bears vs. Packers at least has Aaron Rodgers! Oh, wait. He’s injured.
Now consider a random Tuesday in August. I open the At-Bat app to peruse the pitching matchups for the day. It’s not a full schedule, but I’m just looking for a game or two to maybe check in on in addition to my preferred team’s game. Here’s the problem: The starting pitchers are a conga line of Buck Farmers and Bronson Arroyos and Jason Hammels. Maybe I’ll re-watch Grey’s Anatomy instead.
Major league baseball has 30 teams. The NFL has 32 teams. By Week 10 of the NFL season, quarterbacks such as Brandon Weeden, Brian Hoyer, Mike Glennon, Blaine Gabbert, Nathan Peterman, Brett Hundley, Tom Savage, Jay Cutler and Matt Moore have all seen playing time. Honestly, that’s not even the full list of embarrassing quarterbacks to play at the highest level, but I started to doze off while typing.
What happens when baseball follows in the footsteps of the NFL and expands from 30 teams to 32 teams? Assuming the new teams will following conventional 25-man roster construction, this means 50 new replacement level (or worse) players: 10 starting pitchers, 14 relievers, and 26 position players.
The inevitable wave of injuries will force teams to dip into an already partially drained pool of talent. How will this new group of big league “stars” compare to the current average? The percentage increase or decrease between roughly-replacement level players (those who logged between -1.0 and 1.0 WAR over the full season) with at least 200 plate appearances and the major league average for the last two seasons for several offensive categories is shown below.
In 2016, there were 133 replacement-level players, and in 2017 there were 124 replacement-level players. As expected, the replacement-level players averaged worse numbers across the board, but that said, the drop in performance falls between four and 10 percent, so while worse, the replacement-level players are not hugely worse, at least with respect to offense.
A similar comparison for pitchers, done separately for starters and relievers (the WAR interval for relievers was narrowed to -0.5 to 0.5, since they tend to accumulate a smaller overall WAR than starters and position players), shows the percentage dip in performance is even more significant with respect to pitching categories. The replacement-level numbers were averaged across 56 starters and 112 relievers in 2016 and 61 starters and 124 relievers in 2017. All starters pitched at least 60 innings and all relievers pitched at least 30.
Again, the charts show replacement-level players are below average in all categories, as they should. However, when it comes to starting pitchers, the percentage increase or decrease is significantly larger, with seven of the categories fully eclipsing the maximum percent change for the position player categories and maxing out with a nearly 40 percent increase in home runs per nine innings and xFIP.
But maybe the drop in numbers wouldn’t be quite as bad as the comparisons suggest. After all, the fresh wave of replacement-level hitters has to meet up with the new batch of replacement-level pitchers standing on the shoreline, so wouldn’t it all just average out?
Answering this question is difficult to approximate with existing data. One approach is to add two teams worth of replacement-level players and recalculate the major league averages, but the added numbers would still represent how these players did against the current level of major league talent, not how they would do facing a marginally lower level of talent. Such an experiment also wouldn’t account for the inflated numbers of the above replacement-level players.
Rather than wander too far down the path of speculation, drinking liquids from vaguely labeled bottles to expand and shrink the league at will, it’s worth a trip to the past to see what happened following the two most recent instances of MLB expansion. In 1993, the National League added the Florida Marlins and the Colorado Rockies.
Tracking several offensive categories through the years leading up to and following the expansion year reveals a gentle spike in 1993 across all but one category (walk rate), which merely holds constant. Looking at a window of several years makes it clear that the spike in numbers was not part of general trend leading up to expansion, and seeing the numbers maintained for several years following expansion suggests the spike was not a temporary side effect.
The pitching metrics behave similarly, showing a sharp increase in 1993 that is maintained in subsequent years. Worth noting is the increase in strikeout rate, which might be the category where we see the impact of the replacement-level hitters peeking through the branches of the surging forest of offense sprouting throughout the majors.
The ‘93 data point supports the suspicion that the influx of bad (relative to the league) pitchers is worse than the influx of bad hitters, at least in terms of outcomes on the field. However, 1998 is going to blow into town and inject some fog into the crystal ball. The same graphs centered around the ‘98 expansion show no discernible spikes, hills, or valleys, for that matter.
The numbers mostly hover within a narrow window, which casts some doubt on the theory that expansion will disrupt the game’s competitive balance. Historical context offers a potential explanation. As you may recall, major league baseball had a bit of a steroid issue in the late ‘90s and early 2000s. There’s no way to know exactly the impact PED usage might have had on the numbers, just that it most definitely did. Inflated and skewed numbers could be obscuring the effects of expansion, so go grab a shovel because it’s time to dig a little deeper.
The charts below depict the same percent-change calculations between replacement-level players and the league average as previously shown for the 2016 and 2017 seasons as a means of determining which of the two most recent expansions is the most comparable data point to a near future expansion. For the ‘93 expansion, the two previous seasons were used. The replacement-level players comprised 124 position players, 46 starting pitchers and 83 relief pitchers in 1991, and 125 position players, 44 starting pitchers, and 86 relief pitchers in 1992.
Recall that in recent seasons, replacement-level position players were roughly four to 10 percent worse than average. In ‘91 and ‘92, the replacement-level players are approximately three to 15 percent worse, but with far more categories at the high end of the range than the low end, indicating a larger talent gap among hitters.
With respect to pitching, the percent differences are very similar to those posted in 2016 and 2017, with the exception of walks. Replacement-level pitchers in ‘91 and ‘92 gave up a lot of walks relative to their peers. Since the replacement-level hitters were further below average leading up to ‘93 than in the present day, one might expect an even larger uptick in offense today than in 1993.
The ‘98 expansion requires consideration of 1996 and 1997. The 1996 season contributed 130 position players, 48 starting pitchers and 101 relief pitchers at the replacement level, while the 1997 season threw in 117 position players, 50 starters and 102 relievers. The hitters logged numbers similarly sad (in some cases slightly more sad) relative to league average as those posted prior to ‘93. The replacement-level pitchers prior to ‘98 allowed a way bigger margin of walks relative to the average than their present day compatriots, but beyond that, they matched them in the other categories, just like in ‘91 and ‘92.
The fact that replacement-level players were similarly bad when compared to league-average numbers leading up to both the 1993 and 1998 expansions suggests expansion should have had a similar effect on the overall talent and numbers posted across the league. One interpretation is that these comparisons contribute evidence to the possibility that PEDs and/or other lurking variables obscured the effects of expansion when looking at the league-average numbers over the nine-year window shown previously.
Additionally, since the disparity between the performance drop of replacement-level pitchers and replacement-level hitters was smaller leading up to ‘93 and ‘98 than it is presently, perhaps the surge in offensive numbers observed in 1993 would be even more pronounced in the event of a near-term expansion. Even still, there are other factors and changes to the game to consider.
Injuries are now more common in baseball than tea at a tea party. Though it’s almost certainly true that sometimes the DL is used for strategic purposes, for this conversation, the outcome is the primary concern. Namely, an above-replacement-level player is removed from the roster, and an at-or-below-replacement-level player is added in his place. In a study on injury trends in major league baseball published by the American Journal of Orthopedics, the authors tallied the number of players on the DL, the total number of days players spent on the DL, and the length of the average DL stint for each season from 1998 to 2015. The study’s results are shown below.
Since the most recent expansion in 1998, the total number of DL days per season has increased by roughly 10,000. In today’s game, we find ourselves blessed with the joy of watching an additional 10,000 days over the course of a 182-day season during which a sub-par player occupies the roster spot of a major leaguer. After expansion, the cream of the replacement-level crop already will have been scraped from the minor leagues to fill out the additional teams, so when a player slices his hand open cutting carrots, trips over his dog, or wrecks his dirt bike, the player called up to fill in won’t even be in the top 50 of what the minor leagues currently have to offer.
But thus far, all of this injury talk speaks only to the overall watering down of talent. The study previously cited claims that from 1989 to 1999 pitcher injuries accounted for 48.7 percent of all injuries, creating a fairly even ratio relative to position players, with talent being evenly thinned from the herd. However, if one were to make baseball player cocktails today, position players would be mimosas heavy on the orange juice, while pitchers are bloody marys made by showing a photo of vodka to a glass of V8.
To demonstrate, consider the following chart from a Hardball Times piece written by Jeff Zimmerman and published in December of 2016. The graph shows the chunks of the overall DL days occupied by pitchers and hitters for the seasons from 2002 to 2016. Note the blue chunk is quite large, significantly larger than the red chunk. Consequently, more of the major league-caliber players being replaced by lower-caliber players are pitchers than position players.
The unbalanced ratio of pitcher injuries to hitter injuries persisted in 2017. Data compiled from Roster Resource’s Disabled List Tracker reveals a total of 32,182 DL days, where 20,151 were occupied by pitchers, approximately 63 percent. It would be nice to approximate the impact of the extra 8,000-ish DL days needed for pitchers, but the number of variables in play render the task intractable, at least with any kind of precision.
Such an approximation requires tracking down which pitching and hitting performances were in place of an injured player (a determination that is often ambiguous, and tedious to the point of impossibility) and comparing those performances to either projections of what the injured player would have done or league average. By the time all the guessing and heavy-handed averaging goes in, the final calculation is not likely to offer any more insight than the existing conclusion that the imbalance is likely to lead to a notable increase in offense, piling on the existing evidence previously discussed.
To summarize the mountain of data we’ve flung ourselves down in pursuit of the White Rabbit that controls competitive balance: Replacement-level hitters are bad; replacement-level pitchers are worse, and because of injuries, even more replacement-level pitchers are required than hitters. Meanwhile, if enough replacement-level pitchers flood the majors, a spike in offense could appear as if from nowhere, not unlike the Chesire Cat, who is here to remind us how arbitrary sports fandom really is.
With a heap of evidence showing expansion comes with a decent chance of disruption to competitive balance skewed toward offense, the eternal question hovers overhead as though written by a skywriter: So what? Offense is exciting! Chicks dig the long ball! Or at least that’s what the MLB line at Victoria’s Secret claims, so it must be true.
On the lengthy list of things that are difficult to quantify is how much of the excitement tied to offense stems from its current state as a somewhat rare commodity. There’s no way to know the exact point at which increased offense loses its glittering sheen, especially since that point is likely different for everyone, but at some point our honeymoon with dingers must end. If the scoring patterns in games become more homogeneous, the honeymoon is likely to be cut short by a tropical storm of baserunners. As it stands, a randomly selected game from the MLB.tv menu has a strong possibility of being a 1-0 pitching duel, a 12-2 blowout, or 6-5 nail-biter.
Instead, imagine baseball taking the same path as the fifth season of most sitcoms. Every episode follows the same formula, and your every waking moment binge-watch becomes more of a chore than a treat (Frasier and Niles may as well be relievers with control issues for all the jams they’ve worked out of). Or perhaps the trajectory matches up more closely with the hour-long drama that became too dramatic, deadening your senses to the death of beloved characters (I’m looking at you Shonda Rhimes). So sure, chicks dig the long ball now, but wash that tank top too many times, and the sequins declaring the sentiment will gradually disappear.
Now imagine watching the same episode of the same show every night, but the episode is slowly getting longer, as well. Baseball-Reference lists the average length of game at 3:00, 3:04, and 3:08 for the 2015, 2016, and 2017 seasons, respectively. Multiplying the average game time by the total number of games for the season, then dividing the product by the total number of plate appearances on the season reveals the average length of a plate appearance (perhaps inflated a little, since the estimate also includes commercial breaks, but since the breaks are evenly distributed to all PA, it won’t matter much to the overall point) at 2 minutes 22 seconds, 2 minutes 25 seconds, and 2 minutes 27 seconds. If expansion triggers an uptick in offensive numbers, then an increase in plate appearances will follow. An increase in plate appearances translates to longer games. But how much longer?
To get a rough (emphasis on the word rough) estimate of how much longer the average game would be, consider the added innings tacked onto each season and shouldered by replacement level (or lower) pitchers. Adding two teams would add 162 games to the major league schedule, or at least 2,916 innings pitched. Over the last two seasons, replacement-level pitchers allowed about 1.50 hitters to reach base per inning, which is up from the league average of 1.39 base runners per inning. Extend these numbers to the added 2,916 innings, and that means an extra 321 hitters will reach base over the course of a season, allowing at least another 321 plate appearances.
However, this estimate does not account for the replacement-level hitters introduced to the league at the same time. Looking at the rates with which replacement-level hitters make outs relative to the league average reveals a difference of only two percent. Assuming replacement-level hitters receive an equal number of plate appearances relative to other hitters (a generous assumption, since they would likely be hitting at the end of the order) allots them three percent of all plate appearances, implying about 10 of the 321 extra plate appearances might not happen with the same likelihood, since the replacement-level hitters are slightly worse than those used to calculate the original estimate. However, since replacement level hitters are only two percent more likely to make outs on average, only 0.2 of the 10 plate appearances given to replacement-level hitters are likely to be an out rather than a hit, which is hardly worth subtracting.
When the added time for each of the 321 plate appearances is added to the calculated average game time over the last two seasons (3:06), it increases by only about a minute. Not the end of the world. That said, the scenario under consideration here is best-case. How often are errors that extend innings followed by a rally consisting of multiple plate appearances? What impact do extended innings have on pitcher health? More injuries mean more downgrades to the major league rosters.
Major League Baseball’s rage against the pace-of-play machine, which currently is running slower than dial-up internet, is well-documented. Without trying to read the minds of MLB executives, it’s probably safe to assume they fear games clocking in much beyond three hours may cause fans to spend their precious entertainment hours on other endeavors. Additionally, given the uprising of streaming on-demand television options, sports are one of the few entertainment options that is still best viewed live.
Unfortunately, most people have to work for a living, requiring them to go to sleep at night. Staying up late into the night to finish watching a game is not an option. If too often fans are unable to finish games, a fading interest feels only natural. Looking at the usage numbers for streaming services offers insight into when people choose to consume their viewable entertainment, specifically when they are given the freedom to watch whenever they like, rather than being chained to a programming schedule.
A May 2017 Variety article summarized a study done by Netflix on users’ viewing preferences. The study found that prime time viewing is from 5 p.m. to 10 p.m., as one might expect. Given that a majority of baseball games start at 7 p.m., that leaves three hours before viewers start tuning out. The average length of a game is already north of three hours, so baseball is already in the danger zone. Added offense that elongates games by any amount of time runs counter to the pace-of-play cause.
Up until this point, all conclusions have been drawn on the assumption the landscape of baseball will remain basically the same. That is, the talent pool will maintain roughly its same size and overall level of quality. The effects of expansion shrink if a surge in talent waits on the horizon. Future talent hails mostly from youth baseball programs. The trends in youth sports don’t paint the most optimistic picture. A study by Engage Sports shows an overall downward trend in youth sports with the following sport-by-sport breakdown.
Baseball, like many other sports, has trended downward over the years from 2009 to 2014. The decrease is not as large as the hit taken by other popular sports, but given that not all sports are experiencing a decrease in participation, one explanation is that the upward-trending sports–such as hockey, rugby, and lacrosse–are luring kids away from baseball and other sports.
The Associated Press published some more recent numbers in March of 2016 indicating a three percent increase in baseball participation among kids from ages 6 to 14, but the increase did not continue into the 15-to-18 age bracket. The increased interest in younger kids provides some promise, but not if those kids aren’t continuing to play in high school.
Finally, the following chart from a study by The Aspen Institute corroborates the general decline in youth sports from 2008 to 2013, and more specifically, baseball. Failing a spike in baseball’s international popularity, a luxurious, heart-shaped pool of talent is not coming to save us.
At first blush, expansion promises an injection of all things fresh and new. New cities and stadiums, fresh faces in the crowds and the dugouts, and maybe even a journey into a different country. But anyone who has tried to select a perfectly ripe avocado knows freshness is fleeting. Over-exposure to offense could turn baseball into a pile of brown goo because we couldn’t use it all in time, or else leave us like Michael Scott with a vat of guacamole to eat all alone because no one came to our party.
References & Resources
- Unless otherwise noted, all data were gathered from FanGraphs.
- S. Conte, CL Camp & JS Dines, The American Journal of Orthopedics, “Injury Trends in Major League Baseball Over 18 Seasons: 1998-2015”
- Jeff Zimmerman, The Hardball Times, “2016 Disabled List Information and the Future of Injury Analysis”
- Roster Resource, “Disabled List Tracker”
- Baseball-Reference, “Major League Baseball Miscellaneous Year-by-Year Averages and Totals”
- Todd Spangler, Variety, “Netflix Data Reveals Viewers’ Distinct Genre Preferences at Different Times of Day”
- Paul Langhorst, Engage Sports Blog, Youth Sports Matters, “Youth Sports Participation Statistics and Trends”
- Associated Press, Boston Globe, “Which youth sport is growing fastest? You might be surprised”
- Alice Lee, The Aspen Institute, “7 Charts that Show the State of Youth Sports in the US and Why it Matters”