Wednesday, July 10, 2013

NFL Point Spreads and Winning Percent

As we near the opening of NFL training camps I'll be posting NFL items in addition to baseball. I collected data a couple months ago using the Pro Football Reference Play Index feature that allows me to slice and dice team records by point spread. For you degenerate gamblers, I'm probably 8-10 years behind you, and I claim no expertise in gambling or interpreting its arcane language--I merely present the data I found because I think it's interesting.

This table shows team records by point spread for the 2000-2011 seasons:
Explaining the columns left to right:
UD>14.5--the team was an underdog by 14.5 points or more
14>UD>7.5--underdog by between 7.5 and 14 points
7>UD>3.5--underdog by between 3.5 and 7 points
Btwn -3 and 3--the spread was between being an underdog by 3 and favored by 3
3.5<Fav<7--favored by between 3.5 and 7 points
7.5<Fav<14--favored by between 7.5 and 14 points
Fav>14.5--favored by 14.5 points or more

Every game is a mirror image in that for every team that is favored by 14.5 points, the opponent is the underdog by 14.5 points. One way to interpret the first and last columns is to recognize that extreme point spreads of 14.5 or more occur in only about 5 of the 256 games a year. 

This table does NOT show how well teams perform against the spread--betting against the spread is a totally different proposition altogether:
There's a reason Doug Buffone often states "He who gambles lives in shambles," and it's sitting right there in black, white and blue. Never argue with professionals in any area of life, and gambling is the same--this shows just how amazingly effective the line setters were in the aggregate over the 2000-2011 period. Of course, there are losses in there, but as with every other gambling aspect, the house wins in the long run. Regardless of point spread, the casino maintains the advantage--it's a small one, but small percentages still bring in big bucks. By the way, this is the gambling revenue breakdown for 2012 in Las Vegas (full document viewable here, p12):

Sports betting is approximately 1.4% of Las Vegas casino betting revenue, but anytime 1.4% of ANYTHING amounts to close to $150 million, that's worth my attention.
Any spread less than 7 points is where the action is, representing 81.9% of games (if you add the numbers up, you'll note that not all games are represented--in this time span there were 31 games with spreads of 0 points), with fully 52.9% of the games with point spreads of 3 or less. This is where success or failure lies in that these are (relatively) evenly-matched teams, be it great vs. great, good vs. good or bad vs. bad. This table splits that category between being an underdog by 3 or less and being favored by 3 or less: 

As you think back to the better teams of the past 10 years or so and those that seemed to underperform, some explanation is in this chart. Consider the Vikings, for the most part a decent team between 2000-2011 and yet only 14-27 in games where they were underdogs by 3 points or less and only 16-15 when they were favored. It would be expected that teams in this range would be around .500, because the thinking at the time of the game was that the teams were equal. Since I have the data:
Underdog--.429 winning percent at home
Favored--.544 winning at home
Essentially a wash, implying that home field advantage, while real, isn't as huge a factor as is commonly thought.

























The last table breaks down games with points spreads of 3 or less by coach and goes back to 1978, the first year in which point spread data is available on Pro Football Reference (minimum 50 games):

Pretty interesting--these are arguably the best coaches in recent memory, and an argument can be made that one definition of a good coach would be one that wins games against evenly-matched opponents. There are some oddities in this data, due to sample size as much as anything--for example, Don Shula had a .605 winning percent when his teams were underdogs by 3 or less points, but below .500 when favored by 3 or less. Same for Bill Cowher. Bill Belichick, widely considered the best coach in the NFL today, is merely pedestrian against opponents of equal ability.
























There's a reason for this, and I'll investigate it in a post I've already written but won't post for another week or so. I can tell you the title:
The Ballad of the Bumslayer
Expectations are that the records between evenly matched teams will be around 50/50, and why not--by definition, they're roughly equivalent. If the NFL were like the other sports and had best-of-seven playoffs, the better team would win at a very high rate, but that's not the way it works. If the BEST a team can do is go a smidge above .500 against other good teams, they better be REALLY GOOD against those teams they should dominate. In other words, they SHOULD be bumslayers, and my next post on this issue will study that precise phenomenon. Be on the lookout for it.


No comments :

Post a Comment