Bring The Sharpness Of Vegas Lines Into Focus

Bring The Sharpness Of Vegas Lines Into View

The first question any stat geek should be interested in answering is how close Vegas handicappers are when setting the line on sports. Jon Wheeler at CFBMatrix did just that when he studied the accuracy of Vegas lines in the NFL. Without skipping ahead in this blog post, what would you assume is the average standard error for Vegas lines on every NFL game for the past 8 years? Would you guess 3 points, 4.5 points? Would you think it is even higher, say 7 or even 10 points? Save your answer and keep reading.

The NFL has enjoyed increased exposure over the past two decades, and with that heightened exposure comes greater sports information. Is the heightened exposure a contributing factor in the difficulty of accurately handicapping NFL games? There is a common misperception that the handicappers are too sharp. But are they? Is there real evidence to support the perception? When I first asked myself how large is average standard error is the Vegas line NFL games relative to the final score I figured it would be around 3-5 points. I did some digging and the results were very surprising.

To complete this task I tracked the closing lines for the past 8 seasons using the websiteFootballLocks.com. From 2006-2013 there have been 2048 regular season games played in the NFL. Over that span the average difference between the line and the final outcome of the game was a staggering 10.76 points. Only one season over the past 8 has there been an average below 10 points and that was during this past season when the average was 9.96.

Another assumption made by most NFL bettors is that the handicappers get sharper as the season goes on. The chart below tends to debunk that narrative. The handicappers have better weeks but there is no correlation with the average getting smaller the further you get into a season. What does stand out is that over 136 weeks of NFL games, there was only one week where the handicappers were at an average margin of 5 points, and at no time did they have an average week below that number.

Breaking it down further take a look at the data regarding where the spread is set. From 2006-2013 the lines closed between PK-4.5 points 1025 times (50%). In those games the average difference between the line and the outcome was 10.27. Games with lines between 5-9.5 points occurred 721 times (35%), and had an average difference of 11.57. The line was set at 10 or more points 302 times (15%), with an average difference of 11.44.

Judging by the number of lavish casinos continually being built in Las Vegas, it’s obvious the handicappers are smart, and in no way do I think you can take this information and break the books. The next time you watch a game go down to the final play and hit right on the spread and Al Michaels anoints the handicappers as geniuses, just remember all the other times they are not within two scores.

Breaking it down further take a look at the data regarding where the spread is set. From 2006-2013 the lines closed between PK-4.5 points 1025 times (50%). In those games the average difference between the line and the outcome was 10.27. Games with lines between 5-9.5 points occurred 721 times (35%), and had an average difference of 11.57. The line was set at 10 or more points 302 times (15%), with an average difference of 11.44.

So, how close was your estimate to the reality of 10-12 points in the NFL? Would you further believe that the average standard error in Vegas lines for college football tend to run about 4 points higher at 14-16 points?

More to follow soon.

Jack