Sports betting is big business. In 2018 the United States Supreme Court ruled that states could choose to legalize sports betting in the case Murphy v. National Collegiate Athletic Association. Since then more and more states have legalized and embraced sports betting. Amid the wave of acceptance for gambling on sports, sports leagues and media corporations have also increasingly looked to sportsbooks to use their odds in their analytics. If a major casino in Las Vegas is willing to commit money to the likelihood that a certain team will win the Super Bowl, World Series or Stanley Cup, perhaps we should pay attention to their predictions. At least that has been the reigning philosophy for the leagues and for the media. Programs like Fox NFL Sunday and networks like ESPN regularly integrate betting odds into their coverage. ESPN opened an office in Las Vegas in 2020, and the ESPN program “Bettor Days” may be more dedicated to the gambling than the sports side of sports betting.
I have always been a bit skeptical of the Vegas oddsmakers. There is something cold and unyielding in all of their odds and analysis. I wanted to see how accurate their predictions are when compared with those from major sports commentators. Using the results from the 2020-2021 NHL season I created a method of comparing preseason power rankings from ESPN and NBCSports with preseason Stanley Cup odds from BetMGM, DraftKings, and Fan Duel. All the odds and predictions were taken from the few days immediately preceding the first game of the NHL season. In a nutshell, each set of predictions are assigned an “inaccuracy score” based on how well they projected where each of the thirty-one NHL teams finished. The higher the score the worse the predictions were. Links to the articles I got my data from are included at the bottom of the page. Neither I nor Ten Thousand Sports endorse or are affiliated with any sportsbook or other sports media site. I outline how I collected my data in the “Methods” section that follows the results. Simply skip it if you have no interest in how I calculated the inaccuracy scores
Results
There is a small bias in my method that favors the ESPN and NBCSports experts over the sportsbooks. The bias and the reason for it is described in the method section of this article. Despite the bias, all three of the sportsbooks made more accurate predictions than their media counterparts. The ESPN score from the projections done by the ESPN editorial staff came pretty close to the BetMGM score although their predictions were still 10.5% more inaccurate than the DraftKings projection. The only significant outlier was the NBCSports projections done by NBCSports Boston Bruins beat writer Nick Goss. His projections were the most inaccurate by a significant margin. They were a whopping 24.2% more inaccurate than DraftKings’ projection.
This is too small of a sample to judge the difference between sportsbooks and media experts as a whole, but it is a useful case study. The fact that all the sportsbooks performed better than the media experts despite the possibility of a bias aiding the media experts points to the idea that there is something the sportsbooks are doing differently. Their approach to analytics in this case garnered them better results. This is unsurprising since the sportsbooks are the ones with money on the line.
I would like to believe that there is some sort of sports prediction magic that comes from a person eyeing up a team, having experience in a sport, or just having a gut feeling. These things can certainly help, but they are not enough to beat Vegas. There is a reason why mainstream sports media and the leagues are increasingly embracing sports betting. Money is a big part of why, but the oddsmakers also know what they are doing.
Methods (intro)
Even though the Stanley Cup odds put out by a sportsbook are dedicated solely to predicting who wins the Stanley Cup, they can also serve as a good predictor of where each team finishes in the NHL. For example, in 2020-2021 preseason projections the Detroit Red Wings consistently had the worst odds to win the Stanley Cup. This is a good indicator that they will probably finish somewhere near the bottom of the league. In fact, the Red Wings finished twenty-seventh in the NHL by points. Likewise, teams with good odds are usually the best Stanley Cup contenders. The Montreal Canadiens were an exception to the rule. The two sources that I used that rated Montreal the highest projected them to finish thirteenth (by Fan Duel and Draft Kings). ESPN and NBCSports expected a twentieth and twenty-first place finish for Montreal, respectively. They finished second (league runner-up / lost in the Stanley Cup Finals).
My method is simple, I took the odds or power rankings from each sportsbook and sports expert and compared them with the actual results from the NHL season. The actual results were determined where the teams finished in the playoffs with teams losing in the same round being ranked by regular season points and the NHL’s official playoff tiebreaker rules. Teams that did not make the playoffs are ranked by regular season points then the playoff tiebreaker rules. I tallied up how many spots a team finished out of their predicted spot. So if a sportsbook or expert projects the Red Wings to be the thirty-first best team in the NHL (which also means the worst) but actually finished as the twenty-seventh best team I would add four to the inaccuracy score of the sportsbook or expert. The total tally is a number that represents the total number of inaccuracies in a prediction. I also call this tally the “inaccuracy score”. The higher the score is, the worse the predictions were.
Methods (the pro expert/anti-sportsbook bias)
The biggest issue with using this method is that oddsmakers are not intentionally making power rankings; they are making the odds that a team will win the Stanley Cup. This means that multiple teams can have the same odds of winning the Stanley Cup. If we translate the odds to a projection of how good each team will be this means some teams will be equally good. For example BetMGM gave five different teams the eighth best odds to win the Stanley Cup. I cannot list them as the ninth, tenth, eleventh or twelfth best team because they all have the same odds. In other words my method expects those five teams to finish in the same place. Multiple teams finishing in the same place is impossible and it damages the inaccuracy score of the sportsbook’s projections.
However, projecting two or more teams to finish in the same place can help your inaccuracy score. If team X, team Y, and team Z are predicted to finish the same place does not matter as much what order they finish in. The score is the same whether X is better than Y and Z or if Z is better than X and Y. Having too many teams ranked the same can cause strange results. If a sportsbook said every NHL team had the same odds of winning the Stanley Cup that prediction would only have an inaccuracy score of 230. The result would always be 230 regardless of if the Red Wings were the best team or the worst. Keep in mind that the best score inaccuracy score was 161 and the worst was 200.
To compensate, whenever a sportsbook gives two or more teams the same odds I translated both of those teams rankings into the possible highest rank. In other words BetMGM’s five teams that were tied for the eighth best odds are all ranked as the eighth best team on BetMGM’s projections. There is no ninth, tenth, elevent, or twelfth best team. Since most of the teams that shared odds on any of the sportsbooks’ projections actually finished below their projected spot, using the highest possible projected spot for each of them will negatively skew the inaccuracy score for the sportsbooks. This is the bias in favor of the media experts I mention above. I kept this bias to offset the benefits of sportsbooks giving teams the same odds as highlighted earlier.