Oriole Park at Camden Yards

From 2007-2011, fewer than 2 million baseball fans had annually passed through the gates on Eutaw Street. Since, the team has tacked on more fans every year, with over 2.4 million returning to the Yards in 2014. Bringing fans back to the stadium has been important to the Orioles, both as a source of revenue and as visible support for the team and its players, and it’s been important to the City. It wasn’t long ago that the predominant response I got when I watched Orioles games was, “you still follow them? But they suck.” This might have had something to do with going to a university packed with Yankee and Red Sox fans, but the tides have certainly turned in favor of the O’s. Excitement about prospects is rampant, and people gather in watering holes to watch big games. Even the out-of-state transplants begrudgingly agree that the rising popularity of the Orioles is good for Baltimore and good for baseball. But what brings people out to Camden Yards? And more importantly, what keeps them there? To find out, I put together a regression that predicts season-long attendance using a handful of independent variables. These are mostly things that the team itself has some semblance of control over, and therefore represent levers that the front office could pull to keep the ticket scanners busy. Here are my variables and why I think they’re relevant:

(You can discuss this on the BSL Board here.)

  • Winning Percentage: people like to see a winner
  • All-Star Starter on Roster: people like to watch players they know
  • Pennant Winner: people like to see a winner, especially a team deemed definitely capable of contention
  • Wild-Card Team: people like to see a winner, especially a team with post-season aspirations
  • AL East Team: the Yankees and Red Sox have huge fan bases (including those not living in the New York or Boston areas) that are willing to travel to less expensive stadiums, and AL East teams are hospitable enough to let them shack up in their cities for 9 games a year. This variable ended up not being included in the final regression.
  • Average High July Temperature: the hotter it is, the fewer people are willing to sit for what feels like a meaningless game in the summer
  • Price of Beer: people like to have a beer at the game, and cheap beer is a reason to grab a seat
  • Price of Hot Dog: people like to have a hot dog at the game, and an expensive dinner at the park could turn fans away, especially if they have kids
  • Price of Parking: if people go to the game, they have to leave their cars somewhere
  • Price of Ticket: cheaper tickets encourage attendance
  • Metropolitan Median Household Income: people have to be able to afford to consider a baseball game at all, regardless of food, drink, and ticket price
  • Metropolitan Population: the more people that live in an area, the more have the potential to pass through the gates and be counted as attendees.

I’ve included median household income as a variable because leisure time is expensive. People with more money have a greater ability to take a trip to the ballpark because the opportunity cost of doing so is relatively low: they aren’t missing valuable and necessary work hours by being at the game, for example. Household income could also have a parabolic effect on attendance, though. As someone’s time becomes more and more valuable, they may not be able to afford to miss work to watch a baseball game. You’d probably have to find families whose incomes is in the millions of dollars per minute to find someone too busy to watch baseball, though. Presidents go to games, and if the President of the United States can take in a game, so can you. I’ve collected this data for 2013 and 2014, and adjusted all of the 2013 dollar figures for inflation. My regression treats every team – all 60 of them between 2013 and 2014 – as equal data points without a distinction between years. Had this been done over a longer period of time, I would include the year and maybe the month so that broader economic variables would be represented briefly. Perhaps attendance dipped in New York in 2007 when things began to crash. For our purposes, 60 points and no time series data is sufficient. As a result, this will be a very general regression. Perhaps Orioles fans are more affected by the price of hot dogs than fans of other teams – we won’t be accounting for that. I considered including total stadium capacity as well, but decided against it. Since we’re looking at season-long attendance, stadium capacity would only be relevant if a team were to sell out and have full attendance at every single game. Since that doesn’t happen, stadium capacity isn’t a limiting factor to annual attendance, and I left it out. I also considered promotional nights, thinking that some really great ones could boost attendance by 50,000 across the course of a season. That’s a big number! But who am I to say which nights give the biggest boost to Orioles attendance versus Rays attendance? I don’t make a special point to go to the bucket hat game or the fedora game; apparently many people do. But do the same promotions draw the same crowds in Milwaukee? Or San Francisco? I thought giveaways were better left out with the assumption that each team boosts its attendance through giveaways that its fans would like most, and each team probably boosts its attendance about as much as each other team. Here’s how my independent variables were correlated with one another and with attendance: attendance_correl 

Correlation between a dummy variable like whether a team won a pennant and attendance is a little difficult to put much stock into. Winning percentage was, not surprisingly, the most highly correlated continuous variable with attendance. Ticket price was more positively correlated with attendance, which was unexpected. I imagine that has something to do with the Red Sox and Yankees being the two most expensive tickets in town and less with people clamoring to spend money at the stadium. The average high temperature in July in each team’s home city is slightly negatively correlated with attendance, which was expected. Also interesting to note is the correlations that pop up between variables: teams with a higher winning percentage are most likely to win a pennant or wild card and to have a player voted to start in the All Star Game. The stadiums with the most expensive tickets don’t cut you a deal when you walk in the door; expensive seats also mean expensive beer. In fact, the hit to your wallet starts outside the gates. More expensive tickets also means more expensive parking. And how do you know if parking will be expensive? If you’re going to see a team in a densely populated area, where space is at a premium. On a second pass, I added previous year’s attendance to this chart. It’s clearly the largest determinant in next season’s attendance. A visualization of the relationship between winning percentage and season-long attendance shows a positive relationship, but not a very strong clustering of data points: attendance_winpct

The line shown represents a linear regression between winning percentage and attendance and a 95% confidence interval shaded around it. Lots of teams end up with many more or far fewer people than is expected by winning percentage alone. The Rays are not alone! One of the more challenging relationships to come to grips with is between ticket price and attendance. My first assumption was that higher ticket prices would reduce attendance; that doesn’t seem to necessarily be the case. attendance_ticketprice

I’m going to adjust my assumptions and assume that ticket price has less to do with attendance than I originally thought. Ticket prices might be high because attendance – or demand – is so great for those seats. It might not matter how expensive tickets for the Yankees are, for example. Their fans are still going to show up regardless. It’s more than likely that both of these effects are in play and that fans can both be turned off by high ticket prices and so excited about their team that they drive ticket prices higher. In such an event, the most important variable would be last year’s attendance. That is, some measure of fans’ historical willingness to attend games is necessary to predict their future willingness to attend games:

attendance_prevyear

This is especially true if a team won a pennant in the previous season:attendance_pennant

I suspect that this effect is seen in season ticket sales and early-season attendance. The effect shouldn’t last long because it’s usually apparent pretty quickly if a team will be competitive in the current season. It only takes an offseason and a few games to lose players and recognize that the team isn’t as good as it used to be. The final regression is long, but the coefficient variables can be seen easily here. These represent the change in attendance caused by a unit of change in the variable:

attendance_coefs

Note that the size of the coefficient is also dependent on the size of the data in each variable. Winning percentage is all small decimals between .400 and .700; ticket prices are from $15 to $55, and previous year attendance is in the millions. Winning and the number of fans that watched games last year have the greatest effects on this year’s attendance. To show the values of the smaller coefficients better, I’ve zoomed in on this table:

attendance_coefs_zoom

Going to the playoffs is important, and so is fielding a popular player. Not charging too much for food or beer is nice too, and so is cooler weather. So what does this mean for the 2015 Orioles? Can they keep raising the bar? To answer that question, I’ve made a couple of assumptions. I held prices, median household income, and temperature constant (inflation is probably too low to mean much over one year, and so is the temperature difference due to global climate change) and put in a couple of win estimates. These range from 79 wins, PECOTA’s estimate, 88 wins, about what I usually see on the BSL forums and what I feel is a likely scenario, and 96 wins, a complete re-do of last season. I assumed a 96-win team would either take a pennant or a wild card and that an 88-win team could take a wild card or nothing, which created a range of attendance values the team could draw:

attendance_projected

The better the team is, the more fans it’ll draw. Winning a pennant would take this team into the 2.6 to 2.65 million attendees range. Realistically, the only way the Oriole will lost attendees is if the team tanks and runs into its first sub-.500 season since 2011. It’ll hold attendance steady by winning 88 games and staying home in October, and any better than that is going to raise attendance for the fifth consecutive year.

Patrick Dougherty
Patrick Dougherty

Patrick was the co-founder of Observational Studies, a blog which focused on the analysis and economics of professional sports. The native of Carroll County graduated with a Bachelor’s degree in Economics from Loyola University Maryland. Patrick works at a regional economic development and marketing firm in Baltimore, and in his free time plays lacrosse.

X