The art and science of sports scheduling

Creating the schedule of a professional baseball league is an extraordinary challenge. In theory, there are more possible schedules than atoms in the universe.

I am a member of the Johns Hopkins Scheduling Group, an independent study project under the direction of Hopkins professors, Donniell Fishkind and Anton Dahbura. Our group’s schedules have been or will be used by four baseball minor leagues: New York-Penn League (Short Season-A), South Atlantic League (Low-A), Southern League (AA), and International League (AAA). Here, I will provide the readers of this newspaper a glimpse into the creation of a baseball schedule.

The most important thing to remember as you read this column is that professional sports are a business. As such, the maximization of revenue and minimization of costs are hugely important in our work. For example, all the leagues my group has worked with strongly discourage a team being away on both sides of an off day, since this requires paying for an extra night of hotel rooms.

Leagues also require that teams get an equal number of weekend home dates, since these games draw the best and generate the most money for the teams. Some leagues have extended this requirement to Thursdays, as Thirsty Thursday has become more popular around the country.

The necessity for competitive balance differs as you progress down the baseball ladder. In Major League Baseball (MLB), competitive balance is of the utmost importance — no league executive wants to hear a call from an owner that their team missed the playoffs because an opponent had an easier schedule. However, even at this high level, money sometimes trumps this need.

For example, MLB teams have “geographic interleague rivals,” a nearby team whom they play every year. For instance, the Orioles play the Washington Nationals at least four times a year, whereas their division rivals in Tampa play the Miami Marlins every year.

The Nationals have been considerably better than the Marlins for the past few years, but the MLB feels the need for high TV ratings produced from these local games trumps the need for a perfectly equitable schedule. In lower leagues, competitive balance is even less of a consideration.

The Low-A league I scheduled had no issue with extreme scheduling discrepancies (for example, one team played another team in its division 25 times, but a different divisional opponent just eight times), because player development directors may be more concerned with individual players improving than a team succeeding.

In all levels, the considerations of players have some impact on scheduling parameters. For example, minor league teams in leagues that travel by bus are not allowed to travel 500 miles without an off day.

This prevents the unenviable scenario of finishing a game at 10 p.m., hopping on a bus for a 10 hour drive, getting a few hours of sleep and playing another game, especially when the livelihood of the players depends on their performances.

In the minors, teams can go no more than 30 days without an off day, as negotiated by the players’ union, while majorleaguers can go no more than 20 days without an off day.

Until we came along, the scheduling process in the minor leagues was done by hand. The MLB switched to an approach similar to ours in 2005.

The hand-scheduling process was very tedious and time-consuming, with the creation of a single schedule taking many weeks. You can imagine the challenges this posed for the league: What if a team wanted to host a concert in its stadium? What if a team’s new stadium was behind schedule? What if the League wanted to compare multiple feasible schedules?

With the computer-based approach used here at Hopkins, new schedules can be produced in just hours, allowing the league to obtain answers to all of these questions.

Additionally, our schedules are considerably better at satisfying all league requirements than those that are produced by hand.

Despite our advantages, however, we do face some challenges. Leagues may have constraints that are not obviously quantifiable.

For example, what does it mean, mathematically, if teams are supposed to play each other in good balance throughout the season? It is not clear, so part of our task is to quantify the qualitative.

Our computer programs also stretch the computational limits of modern computers. Without the tricks developed by the group, our computer code would take so long to run that it would not be of much help to anyone.

So, scheduling is an incredibly difficult task. Each league gives us many scheduling requests, but unsurprisingly not all requests can be completely honored.

Teams will see draft schedules and submit more requests but it is not possible to satisfy everyone’s wishes.

Eventually, the league, not the individual teams, decides it is satisfied with a schedule and officially adopts it. As MLB Senior Vice President of Scheduling Katy Feeney says, “If we have made everyone equally unhappy, we have done a good job.”