And then there were 40.

It’s September, and that means that rosters have expanded beyond the usual constraint of 25. For the next month, teams could (in theory, anyway) have 40 players enter a game for them. It also happens to be a time when the minor-league season is winding down. For the teams who are looking forward to looking back on this year, it’s a time for bringing up some of those hotshot prospects and taking a look at them against major-league pitching, because they have nothing better to do. But, hey the teams that are fighting for a spot in the postseason can also call out 40 names if they want. Now everyone can have 10 relievers (including a third lefty), and a sixth starter, two guys who have no defensive value but can hit, two guys who can’t hit but have plenty of defensive value, a third catcher, and a guy whose only talent is that he’s fast. For five months of the year, a baseball roster must be carefully managed, and often, a team faces a tough decision about whether to carry an extra pitcher or an extra position player. In September, those decisions are mostly gone.

September also happens to be the month of pennant races. The contenders and pretenders have largely sorted themselves out. But for five months, they played under the tyranny of roster restraints, while now they play under these strange mutant rules. And if you happen to be a contender who just happens to run into a string of games where the other teams are mostly playing their “B” team in preparation for two years from now, is that fair to the other team who faces a bunch of good teams, augmented by extra resources from the farm?

On the flip side, if call-ups are getting regular playing time, they are often stepping into a vacuum. The Braves are giving hot prospect and “shortstop of the future” Dansby Swanson extended playing time this September (yes, I know he came up in August), but he’s taking time away from… Chase d’Arnaud? Gordon Beckham? Maybe they would have kept Erick Aybar around? The “extra” third utility infielder types weren’t good enough to break onto the roster in July, and while they might be seeing some playing time in September, they’re likely not all that different than the bench players and marginal relievers who were already up. Maybe teams are rostering three lefties in the pen instead of two, but they basically now just have a good LOOGY and two awful ones, rather than just one awful one.

There will be a few hot takes over the next few weeks about how the September roster expansion somehow smears the integrity of the pennant races. Sure, those pretender teams might not have been good anyway, but in May, they were probably at least giving it ye olde college try. Now… who the heck is that at second base anyway? And the contenders are playing a different type of baseball than they used to. And the games (and oh, the box scores!) are getting out of hand.

Are September call-ups really a problem that needs to be “fixed” in the pennant race? In the strict sense, sure the game is different in September, and eventually, the question becomes one of aesthetics, but let’s see if we can get some actual numbers to with the hot takes.

Warning! Gory Mathematical Details Ahead!
First, it’s worth reviewing evidence that I found last year that while the percentage of plate appearances that are taken up by “fringy” players (defined then as players who accumulated fewer than 250 PA or batters faced, as the case may be, that season), the results that we see from those players are (slightly) better in September than they are the rest of the year. But, let’s slice and dice this a little more.

For all of these analyses, I looked at data from 2011-2015. I looked for batters who took their first at-bat in September of that year (excluding pitchers batting). These are the “pure” September call-ups. I also looked for the subgroup who played in September, but not in July or August, although they played in April, May, or June. This group represents players who were (apparently) on the fringes of a major-league roster earlier in the year. They fell off that roster, but they were worth bringing back up once rosters expanded. These are the “fringe call-ups.” Both of these groups (particularly the “fringe” group) could be snagging some injured players who got hurt earlier in the year and just made it back for September, but we’ll live with that. The “pure” and “fringe” groups are by definition mutually exclusive, but as a backstop, I also coded for hitters who simply had fewer than 100 PA during that season, whenever that happened to be.

Now, since we’re going to be asking how much these guys are affecting the pennant races, we need some idea of whether the game being played has actual playoff implications for one of the teams (maybe both!). To that end, I coded a game as meaningful for a team if, on the day of the game, they had not yet clinched a playoff spot, nor had all the playoff spots available to them been spoken for. On top of that, the team had to be within three games either way of a playoff spot. Therefore, a team that was still mathematically in the race, but was four or more games out was not playing a “meaningful” game. Nor was a team that had not yet clinched the division but was four or more games up on their nearest competitor. (This definition will annoy some of you, but we’ll live with that. Obviously, a game might feature zero, one, or two teams for whom the game has “meaning.”

During the month of September (and early October if the season spilled over), 3.1 percent of plate appearances were taken by “pure” call-ups and 2.1 percent by “fringe” call ups. So, roughly one in 20 plate appearances (not even two per game?) were being staffed by guys who weren’t around in July and August.

Now, of course, that doesn’t tell the whole story. In news that should surprise no one, teams who were playing a “meaningful” game were much less likely (less than half as likely, in fact) to throw a call up into the mix. For those teams, only 1.6 percent of their plate appearances were given to “pure” call-ups and 0.8 percent were given to fringe players, compared to teams not playing a meaningful game at 3.6 and 2.7 percent respectively. Teams that are out of it have more room to experiment.

In theory, there’s also a certain honor code among managers that they try not to “throw” games against a contending team by playing the scrubs. Play your “A” lineup until you’re going up against a team that’s also dead in the water. It turns out that there’s some evidence for this, but not a lot. When teams play an opponent who is playing a meaningful game, 2.8 percent of their plate appearances are taken by “pure” call up hitters and 1.9 percent are taken by “fringe” call-ups. When an opponent with nothing to play for comes to town the numbers go up, but only slightly (3.2 percent “pure”, 2.3 percent “fringe”).

Now, there are players who are fringy who cycle on and off the roster all year. Recall that the two groups that I have made are composed entirely (by definition) of players who didn’t appear in July or August. So, here, let’s switch to “players who logged fewer than 100 PA” so we can see over the course of a year how often “fringy” players come to bat. This is going to be a different list of players, so the numbers will be different, but this can give us some idea of how different September is from other months.


Percentage of PA’s by Fringe Players (< 100 PA, season)











Everything After


I suppose that there are two ways to spin this. One is that fringe players only account for 10 percent of the plate appearances in September/October (which isn’t a lot), but that’s still roughly 50-75 percent more than we see the rest of the year. We know that the fringy players who get the call in September are at the very least not any worse than the fringy guys who play the rest of the year, but they are below average players.

Could this really throw a pennant race? I suppose that it’s possible that two teams fighting for the same playoff spot could end up facing off against two very different sets of teams. One might have a bunch of teams lined up who are playing the scrubs while another is playing their regular nine all the time.

I looked at games in September, 2015 and the teams who contended down the stretch to see what percentage of the time they faced a September call-up guy in the batter’s box. It was a pretty big spread. The Toronto Blue Jays, as they fought for what eventually became the AL East title, saw fewer than 1 percent of the opposing hitters fall into either our “pure” or “fringy” call-up categories. On the other hand, the Rangers, Astros, and Twins (in early September, they were actually making a run!) all checked in around 5 percent of the batters whom they faced being scrub guys. And then there were the Pirates. The Pirates didn’t play a lot of “meaningful” games last year. Toward the end of September, it was obvious that they were going to be the Wild Card, but in the games that were “meaningful” as we have described them, their opponents sent a call-up to the plate 13.6 percent of the time! There’s no conspiracy here. No one put out a memo that everyone should take it easy on the Rangers and stick it to the Blue Jays. That’s just the way that the schedule happened to fall.

Let’s say that our maximum swing within the course of one season is about five percentage points on this measure. One team might have the “advantage” of facing a September call-up rather than a “real” hitter an extra five percent of the time, compared to its rival. We assume that most of these September call-ups are replacement level (or maybe a little below?) but whom are they actually replacing? Are teams dropping the 33-year-old stopgap guy who’s already barely above replacement level himself and giving the 21-year-old kid a three-week run? Are they giving super-stars extra days off that they otherwise wouldn’t have gotten, but for the expanded rosters? There’s going to be a little from column A and a little from column B (and C and D and E). We don’t have a way to know this exactly. We’ll—for a moment—assume that the call-ups are replacing players worth half a win on average.

Let’s pretend for a moment that September call-up rules were in effect all season long, and that for some reason, teams played like they do in September all year. If we re-assigned five percent of a team’s plate appearances from players worth roughly half a win above replacement each to players at replacement level (on an annual basis), then we’re talking about losing .025 wins over the course of a whole year. If we assume that our call-ups are replacing league-average regulars (roughly 2 win players over the course of a season), then we’re talking about bleeding away a tenth of a win. And that’s the value lost if a team were to play an entire season that way.

So, given what we know about how much September call-ups actually play and even if we assume that there is some unfairness in which teams get the “advantage” that comes with seeing Triple-A players fill in for big leaguers, the advantage is that some teams have to play a team that we might expect to win 81 games over the course of the year and another gets to play one that we would expect to win 80.9.

Could that be the difference between winning the pennant and not? In theory, yes, but the chances of that being the tipping point are rather small.

We can re-run the same set of analyses on pitchers and find that (again, 2011-2015) 5.0 percent of batters faced in September have been staffed by “pure” call-ups, and 2.5 by fringy pitchers. We find the same sorts of patterns that we found with batters appear again. Teams that are no longer playing “meaningful” games (5.9 percent of batters faced are “pure” call-ups and 2.8 percent are “fringy” call-ups) are more likely than teams who are in the thick of it (2.6 percent and 1.8 percent) to give the kids a run-out. As with batters, we still don’t see evidence of much movement based on whether a team’s opponent is fighting for a playoff spot (when they’re in the race, 4.9 percent “pure” and 2.2 percent “fringy”; when they’re not, 5.0 and 2.7).

When we look at pitchers who faced fewer than 100 batters during the course of a season, we see that they are more likely to be on the mound during a random September plate appearance than in other months. September really is different for pitching use too.


Percentage of BF by Fringe Players (< 100 BF, season)











Everything After


Similar to the results we saw for batters, there was some difference between teams in how often they got to face off against a call-up. In 2015, most teams who were in the thick of a pennant race ended up facing a call-up between 5 and 10 percent of the time. It’s possible that some teams got a bit of a benefit from that, but again, the benefit that the team that the team who sees 10 percent call-ups gets is actually fairly minimal.

In the same way that we overstate the value of one player’s ability to make a difference in a team’s record, we overestimate that player being replaced by a hotshot prospect a few times over the course of one month.

Now Batting, Number 78…
Yes, September is a different brand of baseball. Yes, things “count” a bit more in September. And yes, for some reason, there are a lot of extra bodies in the dugout in September as teams make their big push for the playoffs. And yes, games which feature a seventh inning in which five different guys pitch are annoying. (Though, let’s not let that one get out of hand either. The average April game in 2011-2015 was 180 minutes long. The average September game was 184 minutes long.)

There have been plenty of suggestions that the roster rules are too liberal and that the abundance of warm bodies somehow distorts the game on an aesthetic level. I will leave the aesthetic argument alone (along with the accompanying proposal that perhaps teams might carry 40 players on the roster, but must designate 25 of them to be “eligible” before a game.) But in terms of one team being given an unfair advantage in September because they just happen to run into a stretch of teams who are playing all of their call-ups while another team fights against the regular troops, yes that can happen and yes it makes a difference. But it’s a tiny difference when you do the math. There will be those who will say that any difference is an unfair one, and I can see that argument, but there are so many other things that can turn the tide and which have a much more potent effect.

Leave the call-ups alone. It’s a way for struggling teams to get some of their top prospects a little more developmental time. They aren’t really ruining anyone’s pennant race.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
I would like to see more Sept-type baseball year round but with one big change. Mid-inning substitutions just suck regardless of the month. Baseball players should be playing baseball. I don't pay to go to a game to watch how well the manager micromanages the bench/bullpen - and then watch a cold player warm up or initiate some psych-up batting routine. Even though yeah you can't just ban substitutions.

But you could create a sub rule that mostly affects the ump and the manager - rather than pace/nature of play . Eg - create a window for 'costless' subs. Teams can only make those changes at the top of an inning - submitting all batting/fielding/pitching changes then. I wouldn't even mind if players can be resubbed in later that game - as long as they play complete innings when they do play. But every substitution during an inning renders that player ineligible to play for the remainder of that series.

Course it would make scoring the game almost impossible but that's a small price to pay
Thanks for laying out a range of possible changes in win percentage, it helps to be able to say "sure, maybe it happens, but it's pretty minor". My impression from watching games throughout the year is that when call-ups play during a game falls into one of two categories:
A: Replacing an injured/ineffective starter (SP or position player), in which case they are effectively a replacement player, and any value greater than replacement they provide is a bonus
B: Garbage time. Plenty of appearances (by RP or position players) come in lopsided or extra-inning games.
From a "ruining the pennant race" perspective, I'm not going to worry much about Joey Votto getting replaced in the 6th inning when the Reds are down by 5.