Mirage Hotel and Casino

The Mirage Hotel and Casino appears as a working example of how a prediction market can collect scattered judgments without asking participants to reach agreement. Its sportsbook takes bets across a wide range of sports, while the changing point spread turns bettors’ decisions into a public forecast: in the week before the 2001 Super Bowl, early wagers on the Baltimore Ravens prompted a shift in the line, and the final line generally proved difficult to beat.

The sportsbook and its operating logic

Robert Walker, the Mirage’s sports book director, fields thousands of bets each week, from professional football to Ivy League basketball. For each game, the sportsbook posts a line, or point spread, showing which team is favored and by how many points. A bettor backing a favorite must see that team win by more than the spread; a bettor backing the underdog wins if it loses by less than the spread or wins outright. In other sports, wagers may instead be expressed as odds.

Walker’s task is not to pick winners. In principle, that judgment belongs to the gamblers. His job is to keep roughly equal amounts of money wagered on both sides, so that the sportsbook avoids taking an outsized position on a single result. The house’s profit comes from the vigorish, or vig: in a point-spread bet, a bettor puts up $11 to win $10. If one bettor stakes $11 on each side, the house takes in $22, pays the winner $21, and retains $1. That small margin pays the bookie’s bills only if the wagers remain reasonably balanced.

How the line becomes a collective forecast

To draw wagers to both sides, Walker adjusts the spread in response to betting activity. Before the 2001 Super Bowl, the Mirage opened with the Baltimore Ravens favored by two and a half points. Soon after, it took a couple of early $3,000 bets on Baltimore. Though not a large sum in itself, the concentrated action was enough for Walker to raise the spread to three. If bettors were eager to take Baltimore at the original number, he reasoned, the line might be wrong; changing it could attract bets on the other side.

The sportsbook’s computerized database could, in principle, move the line automatically whenever the money wagered became significantly unbalanced. But the opening number matters: a badly set line can leave the bookie exposed to a flood of unfavorable bets before later adjustments take effect. Once posted, the line also becomes more than the bookmaker’s own estimate. It shifts in response to what bettors are willing to risk, much as stock prices respond to investor demand, and thus comes to represent their collective judgment of the likely outcome.

Forecasting performance and limits

The market’s final line does not foretell exact scores particularly well; point spreads only weakly predict the final scores of most NFL games. Yet consistently beating the spread is difficult even for well-informed gamblers. Favorites cover the spread in about half of games, while underdogs cover it in the other half, and the evidence offers no obvious, persistent error such as home teams regularly winning more often than predicted. The Mirage’s final line is described as the most reliable available forecast of NFL outcomes roughly three-quarters of the time.

Evidence from other sports supports the broader point while marking its limits. In horse racing, final odds reliably order horses by their likelihood of finishing first and provide reasonably good estimates of winning probabilities: a horse at three-to-one odds wins about a quarter of the time. Accuracy is weaker in smaller, less liquid markets, such as hockey, golf, and small-college basketball, where only a few bets can sharply move the odds. Those are also settings in which professional gamblers can find greater opportunities to make money. In horse racing, bettors additionally favor long shots somewhat more than they should and favorites somewhat less, a pattern associated with risk-seeking behavior, especially among people who have been losing.

The Mirage sportsbook therefore illustrates both the strength and the conditions of market aggregation. Its odds are not a crystal ball, and the crowd can display identifiable biases. But where enough bets enter a market, the final line can absorb information dispersed among many participants and provide a forecast that individual gamblers struggle to outperform.