By Benedict Brady
There is a general feeling of uncertainty among the masses headed into Week 1 of the NFL Season. Older players have retired, young new talent has come into the league through the draft, and teams have been trading assets all offseason. The casual fan has a much better idea of how their season will go after the fourth game than during the preseason. It is also intuitive that games played during the current season have much more informative about a team’s strength than than either the previous season or the time in between. For this reason, one might think that betting lines during the first week of the season are more inaccurate compared to the rest of the season.
Interestingly enough, this is simply not true, as line prices are actually more accurate for Week 1 than the rest of the season. Going back to the start of the 2012 season, we can look at the line established by bookmakers and the spreads that occur from each week. After calculating the actual spread for each game and then subtracting out the “betting line,” we can compare the average absolute error.
We can also test the significance of this with a two sample t-test. The p-value comes out to be .075, which is not less than .05, but the consistency with which the Week 1 projections are roughly 1.5 points more accurate than the rest year by year suggests that this would become significant if more years are considered. (Editors note: going back 1985, lines in Week 1 have been off by 10.8 points on average, while lines for the rest of the weeks have been off by 10.7/ This suggests that Week 1 lines are not more accurate, but they are also not less accurate either, as one might have thought).
While this may seem like a surprising result, there is a fair amount of intuition behind it. Bookmakers have much more time to prepare for Week 1, allowing them to collect more information and better adjust their lines based on the incoming money from bettors. Bookmakers respond to how the markets behave, adjusting their lines, and with a longer time to place bets, the market gets tighter.
This analysis could be taken in a few different directions. One could break it down by week (as HSAC has already done for College Basketball: https://harvardsportsanalysis.wordpress.com/2011/10/17/how-much-does-vegas-learn/), or to even break it down by certain characteristics of teams. Regardless, if you are looking for a decent prediction of how your team will do in opening week, look no further than the NFL Betting Markets.
Note: Betting lines are taken from http://www.donbest.com/.