NFL Advanced Metrics for Betting: EPA, DVOA and Data Sources That Matter
Available in US
US sports betting revenue hit $16.96 billion in 2025, a figure that reflects not just how much money is wagered but how sophisticated the market has become. The days of picking NFL winners based on team records and gut feeling are over for anyone who wants a genuine edge. Advanced metrics – statistical models that measure team and player performance beyond traditional box scores – have become the analytical language of sharp NFL betting. And they’re freely available to anyone willing to learn them.
I resisted advanced metrics for longer than I should have. Points scored, yards gained, win-loss record – those felt like enough. Then I started tracking my bets against EPA-based predictions and discovered that the advanced model outperformed my intuition by a meaningful margin every single season. The numbers don’t care about narrative. That’s their value.
EPA Explained: Expected Points Added
Expected Points Added is the metric that changed how I approach every NFL bet. The concept is elegant: assign a point value to every game situation based on down, distance, and field position, then measure how much each play changes that expected value.
At the start of a drive, say first-and-10 at the offence’s own 25-yard line, the expected points for that possession might be roughly 1.2. If the first play gains 15 yards to the 40, the expected points rise to maybe 2.1. That play added 0.9 expected points – its EPA is +0.9. A play that loses yards produces a negative EPA. Over hundreds of plays, a team’s cumulative EPA tells you how much value their offence or defence is generating above or below average.
The power of EPA over traditional stats like total yards is that it accounts for context. A five-yard gain on third-and-three (converting a first down) has a very different value from a five-yard gain on third-and-eight (falling short). Yards don’t distinguish between these. EPA does. This context sensitivity makes EPA a much stronger predictor of future performance than raw yardage or even points scored, because points include noise like field position after turnovers and defensive touchdowns.
For betting, EPA per play is the most useful form. Offensive EPA per play tells you how efficient a team’s attack is on a play-by-play basis. Defensive EPA per play (a negative number means the defence is reducing opponents’ expected points) tells you how effectively a team limits offensive value. The gap between a team’s offensive and defensive EPA per play gives you a composite efficiency rating that correlates strongly with spread outcomes.
Free EPA data is available through nflfastR (an open-source R package) and rbsdm.com, which publishes updated EPA tables throughout the season. You don’t need coding skills to access the data – rbsdm.com presents it in readable tables that update weekly.
DVOA Explained: Defence-Adjusted Value Over Average
The global sports betting market stands at $111.9 billion, and DVOA – developed by Football Outsiders – is one of the metrics that professional bettors use to find value within that ocean of money.
DVOA measures a team’s efficiency by comparing its performance on every play to the league average, then adjusting for the strength of the opponents faced. A team with an offensive DVOA of 15% performed 15% above average after adjusting for the quality of defences they played against. A defensive DVOA of -10% means the defence held opponents to 10% below average efficiency, again adjusted for opponent strength.
The opponent adjustment is what makes DVOA distinct from EPA. A team that racks up huge EPA numbers against weak opponents will have its DVOA tempered downward. A team that posts modest EPA against elite defences will see its DVOA boosted. This adjustment makes DVOA particularly useful for predicting outcomes in matchups where the strength-of-schedule disparity between the two teams is significant – exactly the kind of game where casual analysis tends to go wrong.
Football Outsiders publishes DVOA ratings weekly, broken down by offence, defence, and special teams. They also split by pass and rush, letting you identify specific matchup advantages. If a team’s passing offence DVOA is elite but its rushing offence DVOA is below average, and it’s facing a defence that’s weak against the pass but strong against the run, you have a clear matchup advantage to exploit in both spread and totals markets.
Applying Metrics to Your NFL Bets
Metrics without application are academic exercises. Here’s the practical workflow I use every week to turn EPA and DVOA into betting decisions.
Step one: pull each team’s offensive and defensive EPA per play for the last five weeks. Season-long averages are useful but can be dragged by early-season noise (new coaching schemes, injuries, soft schedules). A five-week rolling average captures current form while maintaining a reasonable sample.
Step two: check DVOA for the same period to confirm whether EPA performance is supported by opponent-adjusted data. A team whose EPA looks strong but whose DVOA is middling may be feasting on weak opponents. The reverse – modest EPA but strong DVOA – suggests a team performing well against quality competition, which the market may be undervaluing.
Step three: compare your derived prediction to the bookmaker’s spread. If your EPA/DVOA analysis suggests a three-point margin and the spread is 6.5, there’s a potential value bet on the underdog. If your analysis aligns with the spread, there’s no edge and you pass.
Limitations matter. Advanced metrics are descriptive, not prescriptive – they tell you what happened, not what will happen. Injuries, weather, and coaching adjustments can override what the numbers suggest. A team’s EPA profile changes materially when its starting quarterback is out. Always cross-reference your metrics analysis with situational factors before committing money.
One mistake I made early on: trusting metrics over too-small samples. EPA and DVOA become reliable around Week 6 of the NFL season. Before that, the data is noisy and overfit. Using early-season advanced stats to make confident bets is false precision. For a broader overview of how metrics fit into a complete approach, the NFL betting strategy guide integrates these tools with bankroll management and line shopping.
A practical tip for UK punters new to advanced metrics: start with one metric, not four. Pick EPA per play, learn where to find it, and spend one full season using it as a cross-reference against your existing analysis. Once you’re comfortable interpreting EPA and seeing how it correlates with spread outcomes, add DVOA. Layering metrics gradually builds genuine understanding, while attempting to use every available metric simultaneously leads to analysis paralysis and no bets placed at all.
Where can I find free NFL advanced stats?
rbsdm.com publishes EPA data in accessible tables updated weekly. Football Outsiders (footballoutsiders.com) publishes DVOA ratings. Pro Football Reference offers a wide range of traditional and advanced statistics. The nflfastR project provides play-by-play data for those comfortable with R or Python.
Do advanced metrics work better for spread or totals betting?
Advanced metrics are useful for both but tend to have slightly more predictive power for spreads, where team efficiency directly determines margin of victory. For totals, pace and play volume matter alongside efficiency, so metrics like plays per game and time of possession should supplement EPA and DVOA analysis.
Prepared by the American Football bet editorial staff.
