It seems like any time the margin of victory of a sporting event falls close to the closing line, mainstream sports media personalities declare something along the lines of, “Wow, it ended up as a three-point game and the spread was three, Vegas knew!”
While it’s fun for content and garnering Retweets, the idea that oddsmakers have some magical model or information not available to the rest of us is simply not true. Why don’t we ever hear about the times that three-point underdogs win by 21 points? Because it doesn’t make for good content.
So, outside of cherry-picked examples, how good are oddsmakers at actually predicting the outcomes of sporting events, specifically in the sharpest betting markets such as the NFL?
Since the start of the 2003 season, the average NFL margin of victory has fallen 10.3 points from Pinnacle’s closing line. The median is also a full eight points off.
The chart below summarizes the difference between NFL margins of victory and closing lines, since 2003:
It doesn’t take long to realize that oddsmakers don’t know much more than the rest of us when attempting to predict NFL results. Only 21.3% of games in our sample finished within three points of the closing spread.
In fact, only 50.1% finished within eight points, meaning only half of all games played fell within one possession of the closing spread. Forty-one percent also missed by more than 10 points, showing just how far off many NFL games land.
Even if closing lines were more predictive, the NFL market, and not “Vegas”, would be responsible. Oddsmakers simply provide an opening number, and at that point the market takes over.
Sportsbooks don’t have an advantage because they’re smarter, their profits come from an ability to charge bettors -110 to make wagers. Assuming an efficient market moves lines to 50/50 probabilities, charging a tax to bet keeps sportsbooks in business — not an ability to predict the future.