NFL Betting Fundamentals

Because NFL teams are constructed according to strict rules—the NFL salary cap, draft structure, standardized schedule, and other forces aimed at maintaining parity—the talent level from top to bottom occupies a narrow band. That means successful teams have superior coaching, schemes, preparation, treatment, technology, process, contingency planning, and so on.

The paragraphs that follow discuss non-football factors that have an impact on the game. Some of the best handicappers and analysts will tell you that winning bets is gaining mastery over these elements of the sport.

Schedule spot (travel, bye week)

When we build a mental model of what an NFL team is, it’s tempting to imagine it as a static list of players, slotted into positions, each with a statistical profile that, when combined with the other players on the team, reveal themselves to be a big deterministic machine. Generalizing is a habit: “This team is good” and “that team is bad,” but it’s a terrible habit for someone trying to predict performance.

Teams are a function of their players, but its important to remember that they function in context. The way they perform on the field is as connected to who they’re playing against, how healthy the teams are, how rested they are, and so on. Yes, the teams you’re watching this week has the same names and jersey numbers as last week, but they aren’t really the same.

Travel

There have been scads of studies devoted to travel effects on pro football teams. A 2013 Stanford study expanded on previous theories about circadian advantages enjoyed by west coast teams playing night games on the east coast due to their body clocks being set to "afternoon" while the east coast team’s body clocks were set to "bedtime." Other studies have tried to determine if the stress of travel has a detrimental effect on teams crossing time zones in either direction.

I twisted the data every way I could for teams going coast to coast in either direction and couldn’t find an angle. That’s probably an indication that the betting market has priced these travel factors into the point spread. Having said that, in the last four years betting against Eastern Time Zone teams playing at any time on the west coast has netted a 34–19 spread record. Not much to hang your hat on.

You can take advantage of the market when it underreacts and when it overreacts. In an ideal world, you find a factor that affects betting outcomes of games that nobody has considered. Bet it for profit until other people start mentioning it. Then when you finally hear other bettors mention it, consider betting the other side.

A Boston College research team led by Kyle Waters took the Stanford study one step further and tried to isolate specific travel distance as it impacts winning percentage. They concluded that NFL teams’ winning percentage drops 3.5 percent for every 1,000 km traveled, with bigger impacts when the team changes time zones and when they play outside. That’s a built-in disadvantage for teams in the west. If your team is in California, Arizona, or Nevada, your team spends easily twice the time in the air than a team in Ohio or Pennsylvania.

BET on early season wanderers

I didn’t want to drag you through that discussion without giving you a system to look for. This system takes both time and space variables into account:

In the NFL regular season AND

It’s the first half of the season (week 9 or less) AND

The home team is based in the Eastern time zone AND

The road team is based in the Pacific time zone

BET the road team’s point spread.

This system comes up a few times a year and has gone 42–23 (64 percent, with a modest Z-score of 2.3). It is possible, after all, that both the Stanford and Boston College research studies are correct (that winning percentages drop with more travel) and for the betting markets to have totally overreacted to this phenomenon in the form of plum point spreads for the visiting team.

Rest

I am a believer that rest is a useful add-on characteristic when evaluating matchups. While I think the oddsmakers build bye weeks into their models, I think it’s helpful to set aside preconceived notions of what player rest actually does for betting outcomes. Rest is very likely a net positive in terms of improving player health, and for most coaches, more time is good for game-planning activities. But there’s also value in routine and focus, which are sometimes at odds with rest.

By far most NFL games are played on the reverse-Genesis schedule: Work on Sunday, rest for six days, work again the next Sunday. But with Monday and Thursday night games interspersed with bye weeks, plus Thanksgiving day and a few Saturday games late in the season, there are a lot of rest possibilities, as you can see in this table.

BET the curse of the well-rested

Look what happens in the latter half of the season when a home team is playing after an unusually long rest period:

During the NFL regular season

It’s week 9 or higher AND

The home team is playing after more than 8 or days of rest AND

The home team lost their last game

BET on the road team’s point spread.

This system actually works with or without a loss in the previous game, but this setup has gone 65–28 ATS since 2012 (69 percent with a Z-score over 3) and went 10–7 in 2019. I like this approach, and I think there’s much more to be discovered when looking at rest.

Rest differential

The NFL takes pains to pit teams against each other that have a similar rest profile so neither team is at a major disadvantage. Obviously, it doesn’t always work that way. This table shows the distribution of rest differential.

Injuries

Injury creates uncertainty around games. Uncertainty means higher variance of possible results. This you probably already knew. The biggest challenge for someone examining the games is not when a player is hurt; it’s deciphering injury information for players who might play.

A Football Outsiders study looked at NFL injury reports and calculated that when a player is listed as Probable, they play 95 percent of the time, and 32 percent of the time when Questionable. But teams, general managers, and coaches have different approaches to reporting injuries. Perhaps not a major factor, but if there’s a player who’s key to your handicap listed as probable, it’s a good idea to understand the team’s history in reporting injuries.

Teams also demonstrate their ability — or lack thereof — to make adjustments to scheme and style in the event a key injury. The 2017 Philadelphia Eagles won the Super Bowl in spite of a rash of injuries, which most people credited to coach Doug Pedersen’s adjustments to play-calling, personnel groupings, and so on to fit the strengths of the remaining healthy players. As a Dallas fan, this is doubly painful because the Cowboys of the last decade have shown the opposite skill: the collapse like a house of cards when key players sit.

The final point on injuries that sharp analysts watch for is that players will play before they’re at 100-percent speed. Going at 80 percent if you’re a precision-route wideout like Amari Cooper can mean the difference between a 9-catch day and a 2-catch day. I like these three approaches with skill position prop bets:

• Fade a yardage total for a skill position player returning from a minor injury that’s kept him on the sidelines for at least two games.
• Watch for big performance dips from skill players that occur without explanation during a win, and fade the yardage total prop the following week.
• Is there such thing as "injury prone"? And are certain injuries subject to re-aggravation more than others? Yardage total prop bets for skill players are usually built on assumptions of full speed, but going under that total on a player with even a 10-percent or 15-percent chance of having to come out of a game can move the bet into positive EV territory.

Weather

Football is an all-weather sport, although the NFL has sensibly started pausing games if they’re in the eye of massive electrical storms. When it comes to whether effects, we’re interested in three basic factors: wind, precipitation, and temperature.

It should go without saying that everyone reacts differently to weather. I can think of some notable examples of skill players who had particular issues with certain conditions. Hall of Famer Troy Aikman, who played high school and college football in Oklahoma and Southern California, was vocal about not being able to comfortably grip a wet ball.

Data scientist Josh Mancuso studied three decades of NFL data and drew a handful of conclusions about quarterback play in bad weather:

• Games played in high wind (20 mph or more) have the greatest negative impact on QB production.
• QB efficiency doesn’t change much in bad weather, but production metrics (touchdown passes, and yards passing) take a hit.

A different Stanford team led by Rory Houghton-Bery drew some fascinating conclusions in a 2017 paper about weather when they tested differential effects for the home and road team.

• As a rule, rain and snow are an advantage to the home team and produces statistically significant increase in point differential toward the home team. The inference they made was that home teams are more used to the bad weather. Makes sense to me.
• Teams playing in a dome see more home-field advantage than a team that plays their home games in the open.
• Run-heavy teams (if there are indeed any left in the NFL) get less of an advantage in bad weather than a pass-heavy offense.
• Wind is the great equalizer; it’s bad news for both teams’ offensive expectations. Bet under the total on a windy day.

Multiple teams have looked at the impact of temperature on a game. If you’ve ever tried to throw or catch a football in bone-chilling cold, you might be surprised to learn that cold weather — by itself — has not historically limited total points in a game. If anything several studies have found that extremely hot weather has more of a negative impact than cold weather.

A numberFire.com article written during the 2017 season is probably the most useful of all. In 256 regular season games, 20 percent of the games were played in moderately windy conditions (7–12 mph) where average total scores dipped by a point, and 15 percent were played in high wind (over 12 mph), where total scores were down over 6 points on average, with the under bet winning 19 times and losing 11.

Wind looks like a promising angle for a fundamental handicapper, but it’s hard to pass final judgment on an angle like this without more precise data. The biggest question is always this: Does that betting record represent wins and losses against the consensus closing total, the opening total, or some total in between? What wind data is being used? Was that wind as measured at the stadium? If you looked up the wind speed for Hot Springs, Arkansas (or a random city of your choice) on May 2, 2010 (or a random date of your choice), you’d probably see a single vector representing an average direction and velocity, but that doesn’t necessarily represent what’s happening at the stadium.