Semi-Charmed Kind of Schedule
The cost to making team schedules pretty

On June 8th, the Baltimore Orioles hosted the Mariners for a four game series at Camden Yards. Then, after a quick palate cleanser against the Padres, the Orioles flew 2,300 miles to Seattle to play the Mariners. If you follow any team closely you've seen this pattern: play a team at home, blink twice, and you're playing them again in their park like it’s the second leg of a Champions League knockout round. This isn't rare. The 2026 Orioles do it six times. The Rays do it five times. Every team but one does it at least once (more on that exception later).
It turns out this quirk has a name, and the name comes with a body count.
The Firing of 2004
For 24 years, the MLB schedule was made by hand by Henry and Holly Stephenson, a married couple who put together every season's calendar with pencils, paper, and a level of pattern recognition that borders on savantry (ESPN made a great 30 for 30 short about them). In 2004, MLB put the contract up for bid and gave it to the Sports Scheduling Group, a small firm of operations-research PhDs. The reason, per MLB's own SVP of scheduling at the time, was specifically that the computers did a better job avoiding "semi-repeaters", the league's term for the same two teams playing back-to-back series at home and then away (ESPN, 2004).
Read that again. The husband-and-wife team lost their job of two decades, in part, over the exact scheduling pattern I just described the 2026 Orioles doing six times.
So what happened? The 2023 balanced schedule happened. Once every team owes every same-league, non-division opponent exactly one home series and one road series (plus one series against each interleague club, flipping venues annually), the cheapest way for the optimizer to satisfy all those obligations is to pair the legs into tidy little two-week "swap windows", you host me, I host you, everybody's travel books balance. The semi-repeater are not there by accident, it’s an intentional mechanism.
How often does the rematch come?
To see this, I measured every "venue flip" in the 2026 schedule: any time a team's series against an opponent is followed by the next series against that same opponent at the opposite park, how many days pass in between?

Across all 435 opponent pairs in the 2026 schedule, there are 320 venue flips, and the tallest bar in the whole histogram is the 0–6 day bucket. 19% of all venue flips are semi-repeaters (62 of 320, within 14 days). From there the distribution rolls through a broad second mound at 4–6 weeks, the "paired but looser" windows, thins out through the summer-length gaps, then spikes again at the far right: pairs used as season bookends, meeting in April and not again until 120+ days later. The thing MLB switched vendors to eliminate in 2004 is now the single most common way the schedule handles a rematch.
Why is that spike there? Because the schedule is built out of what I've been calling swap windows: two teams interlock their calendars so that your homestand is my road trip, then we flip. Here's a real one from May, the Orioles and Rays paying off both of their series debts to each other inside a single twelve-day block:
How the MLB schedule works
The balanced schedule owes most opponent pairs exactly one home series and one road series. The cheapest way for the optimizer to pay both debts is to interlock two teams' calendars: your homestand is my road trip, then we flip. That interlocking block is where semi-repeaters come from.

The optimizer loves this shape for three reasons: homestands stay contiguous and sellable, the pair's home-and-road ledger closes locally instead of haunting the other 24 weeks of the puzzle, and the blocks stack the Rays' road trip in that diagram runs through the Yankees, whose own block interlocks with someone else's. The season is a quilt of these windows, and hold that thought because it's about to matter.
Why Not Just Spread Everything Out? (The Soccer Question)
The obvious rebuttal: fine, pair up the legs, but why not make every block eight weeks wide instead of one? Play the first leg in April, flip it in June, everybody's rematch breathes.
Here's the fun part: that schedule exists, and half the sporting world uses it. It's called a mirrored schedule, play every opponent once in the first half of the season, then run the whole sequence again with venues flipped. The Premier League, La Liga, and most big soccer leagues are built exactly this way, which is why you never hear a soccer fan complain about semi-repeaters: every pair's two legs sit roughly half a season apart by construction.
And here's the really fun part: the people who build MLB's schedule wrote the academic textbook on why baseball can't have it. In 2001 - three years before winning the MLB contract - Easton, Nemhauser, and Trick published the Traveling Tournament Problem, which became the canonical research benchmark for scheduling a round-robin while minimizing travel. Nemhauser and Trick are Sports Scheduling Group co-founders. The problem's standard formulation literally includes a "no-repeater" constraint, and the mirrored variant is studied as its own, harder problem, because mirroring is an added constraint, and added constraints can only ever make the travel-optimal answer worse or equal. The schedule-makers invented the homework problem before taking the job.
Why doesn't the soccer solution port to baseball? Three structural reasons. First, MLB isn't a uniform round-robin: division pairs meet in four series (already well-spread, the semi-repeaters aren't there), and most interleague pairs meet in one series at one venue, so there's no second leg to separate at all. The mirror would rigidly pin the entire calendar to fix only the two-leg intraleague pairs. Second, the flip leg is the other team's travel: pin the Orioles' Seattle trip to week 5 and you've pinned a Mariners East Coast trip to week 13, and Seattle's trip needs to bundle Baltimore with other eastern stops that each have their own mirror pins at different weeks. Uniform separation turns the flexible quilt into a rigid lattice. Third, baseball's calendar is structurally lopsided in ways a mirror can't absorb: cold-weather teams front-load road games in April, shared markets need one team home while the other's away, the Red Sox must be home on Patriots' Day, and the league deliberately back-loads division races into September. A mirror demands symmetry. The season doesn't have any.
Now, before we grab pitchforks and march on the Sports Scheduling Group's office in Pittsburgh, I wanted to answer the obvious follow-up: fine, the schedule is ugly and the pretty version is constrained, but what would it cost to force it anyway?
The Price of a Pretty Schedule
Taking the Orioles and their six semi-repeaters (Yankees in May, Rays twice, Blue Jays twice, and that Mariners sandwich) and ran a simulation: keep every home date locked in place (ticket sales are sacred), and shuffle the road series among their existing calendar slots until no home/road pair against the same opponent sits within 14 days. Then count the frequent flyer miles.
The actual 2026 Orioles travel 29,028 air miles across 39 flights. I sampled 20,000 random shuffle plans, and here's the first surprise: only about 4.5% of them even produced a fully de-clustered schedule. The constraint space is tight. Among the valid plans:
Orioles Actual Travel Schedule

Average “De-clustered” Travel Schedule

The luckiest valid plan came in at 26,384 miles, 2,644 miles less than the real schedule
The median plan cost +5,583 miles (+19%), roughly two extra cross-country round trips
The unluckiest cost +12,095 miles (+42%), and features a Baltimore-to-LA-to-New-York sequence that would make the players' union file a grievance mid-flight
That first bullet deserves a double take. There exists a version of the Orioles schedule that is both prettier AND cheaper for the Orioles. The algorithm left miles on the table! Except, and here's the entire point of this article, the algorithm was never trying to minimize the Orioles' travel. It's balancing 30 teams at once, and every one of these shuffles moves a dozen series onto ten other teams' calendars, each of which cascades into their homestands, their travel, their semi-repeaters. As the late MLB scheduling executive Katy Feeney once put it: "It's not a two-week schedule. It's a 26-week, 30-team schedule."
Okay, But What If Everyone Got a Pretty Schedule?
Naturally the next step was to run this experiment for all 30 teams. One important honesty note on the method: each team's fix is simulated unilaterally, I'm pretending their opponents graciously absorb the moves, when in reality the fixes conflict with each other (when the Orioles de-cluster their Rays pair, they're moving the exact series the Rays would want to move differently). So treat these as lower bounds on the real chaos.

The league-wide results:
Average cost of a semi-repeater-free schedule: +3,435 miles per team
Illustrative league total if everyone de-clustered: +103,040 miles, about four extra laps around the Earth, distributed across baseball
The Rangers pay the most (+7,150 median), with the Braves, Giants, Angels, Orioles, and Phillies all north of +5,500
Meanwhile the Reds (−40) and Guardians (−263) could de-cluster for free, their repeaters are with nearby Central rivals, so the clustering isn't even saving them gas money
Now let’s convert those miles to time: at a charter's ~500 mph cruise, the average team's +3,435 miles is about seven extra hours in the air per season. Teams fly roughly 40 times a year, and de-clustering doesn't add flights, it only lengthens them, so the tax works out to about ten extra minutes per flight. Ten minutes! Players lose more time than that to tarmac delays. If the punchline of this whole exercise were "a pretty schedule costs each team one Ted Lasso cold open per flight," this would be a very anticlimactic article.
The Miles Were Never the Problem
So if the fuel is cheap, why doesn't MLB just do it? Because the miles are the visible cost, and the schedule's real currency is sequences. Three things my odometer didn't count:
Feasibility is the scarce resource. Remember that only ~4.5% of random shuffles produced a valid de-clustered schedule for one team, and my model wasn't even enforcing the real rules. To see why, let's actually follow one fix. Take the Orioles' May Yankees semi-repeater and apply the cheapest possible move: trade the dates of the Bronx trip (May 1–4) with the July trip to Cincinnati (July 3–5).

Domino one falls before anyone boards a plane: the Bronx series is four games and the Cincinnati slot holds three, and the balanced schedule requires Orioles–Yankees to be exactly a 13-game split (six in the Bronx, seven at Camden), so one game is homeless immediately. Domino two: the Yankees were already hosting the Twins on July 3–5, so now they're double-booked on Fourth of July weekend, and the Twins series has to slide into the May 1–4 hole Baltimore just vacated. Domino three: the Twins had their own series scheduled May 1–4, so that opponent gets orphaned, and its only natural landing spot is July 3–5, where it's already booked against somebody else. The chain doesn't close. It propagates, and every hop has to dodge the collective bargaining agreement's actual scheduling rules: the cap of 20 consecutive dates without an open day (24 in rainout emergencies), the getaway-day rule that forces early start times whenever either team faces a 2½-hour-plus flight before playing the next day, and the flat prohibition on scheduling a game before 5 p.m. when a team played a night game in another city the evening before (rules the union fought for after years of teams checking into hotels at 5 a.m.) Oh, and every relocated home date is a date some regional network already sold; the Yankees losing a July 4th weekend home series is a phone call nobody at MLB wants to make. The real feasibility rate, with the CBA and TV layered on, is dramatically lower than my toy version.
The miles don't arrive as minutes, they arrive as bad trips. The median de-clustered plan isn't 40 flights that are each slightly longer; it's a few specific ugly sequences, like my unluckiest Orioles plan's Baltimore-to-LA-to-New-York whiplash. Those are the timezone-crossing, sleep-wrecking itineraries the union spent two CBAs legislating against, and there's real research suggesting eastward travel measurably degrades performance. Two extra cross-country round trips isn't a comfort cost, it's a competitive one.
Every fix is someone else's problem. The Reds can de-cluster for free, but doing it uglifies the Rays' calendar, because the series the Reds want to move is the same one the Rays would move differently. Thirty teams, one calendar, every clause touching every other clause. It's a treaty, not a to-do list.
The Immaculate Calendar Rankings
And finally, the standings you didn't know you needed: every team ranked by how "pretty" their 2026 schedule already is - fewest semi-repeaters - with ties broken by how cheaply the remaining ugliness could be fixed.
Rank | Team | Semi-Repeaters | Cost to Fix (median) |
|---|---|---|---|
1 | Pirates | 0 | already perfect |
2 | Nationals | 1 | +496 mi |
3 | Marlins | 2 | +1,715 mi |
4 | Reds | 3 | −40 mi |
5 | Guardians | 3 | −263 mi |
6 | Astros | 3 | +2,572 mi |
7 | Padres | 3 | +2,378 mi |
8 | Brewers | 3 | +2,487 mi |
9 | Red Sox | 3 | +3,098 mi |
10 | Dodgers | 3 | +3,262 mi |
11 | Diamondbacks | 3 | +4,551 mi |
12 | Cubs | 4 | +1,175 mi |
13 | Tigers | 4 | +1,867 mi |
14 | Royals | 4 | +1,953 mi |
15 | Athletics | 4 | +3,618 mi |
16 | Mets | 4 | +3,678 mi |
17 | Twins | 5 | +1,630 mi |
18 | Blue Jays | 5 | +3,390 mi |
19 | Cardinals | 5 | +4,648 mi |
20 | Mariners | 5 | +4,767 mi |
21 | Rays | 5 | +4,884 mi |
22 | Yankees | 5 | +4,854 mi |
23 | White Sox | 5 | +5,370 mi |
24 | Angels | 5 | +5,859 mi |
25 | Rockies | 6 | +3,755 mi |
26 | Orioles | 6 | +5,650 mi |
27 | Phillies | 6 | +5,544 mi |
28 | Braves | 6 | +6,766 mi |
29 | Giants | 6 | +6,226 mi |
30 | Rangers | 6 | +7,150 mi |
Yes: the Pittsburgh Pirates have the only semi-repeater-free schedule in Major League Baseball. Zero. They're also the league's lightest travelers at 27,369 miles. The algorithm has a favorite child, and it plays at PNC Park. Whether being the schedule computer's golden boy translates to actual wins is a question for a franchise that has spent this century answering "no" to questions like that.
So What's the Verdict?
This isn't a defense of the robot schedule. The robot schedule IS ugly, and the fans grumbling about it aren't wrong, there's a whole genre of columns pining for the mom-and-pop era. But the surprise of running the numbers isn't that a pretty schedule is expensive. It's that a pretty schedule is cheap (ten minutes a flight) and still nearly impossible, because the cost was never fuel. It's that the calendar is a 30-way treaty where every semi-repeater is quietly holding up a start-time rule here, an off day there, a broadcast window somewhere else. The schedule isn't ugly by accident, and it isn't even ugly because beauty is expensive. It's ugly because 29 other teams signed for it.
Methodology notes: Travel is modeled as venue-to-venue flights with returns to the home city between road trips (this slightly undercounts off-day detours). Series were built from the MLB Stats API schedule feed; games ≤2 days apart against the same opponent at the same venue are treated as one series, which handles most rainout makeups correctly. Each team's de-clustering is simulated unilaterally, so league totals are illustrative, a true simultaneous re-solve is the actual hard problem, and it's why the Sports Scheduling Group gets paid.