Advanced Football
Analytics Guide
This guide is designed to help you understand the advanced analytics terms and metrics used by Sports Snaps.
Getting Started
Consider this a starting point into the world of advanced football analytics. This guide is not comprehensive, there are many other terms and metrics to learn about.
How to use this guide
This guide is not meant to be read in any particular order. It is in a glossary format, so you can jump to the section you want to learn about.
Glossary
Expected Points Added (EPA)
"That play may have well cost the team points."
While a play may not result in points, it is so significant that it is considered to have had an impact on the final score.
What it represents
When analyzing gameplay impact, expected points added (EPA) per play is a way to measure which plays were pivotal moments in a game, and which weren't.
Technically, EPA measures how many points can be expected to be added to the score of the possessing team based on field position and the game situation.
Why it matters
Not all plays in a football game are equal. Some plays don't matter much to either team's success, while some plays determine who wins or loses.
While a play may not seem important on the surface, like a 2 yard loss on a 1st down, it can have a surprising impact on the final score if we look closely at the game situation. EPA helps quantify this impact.
EPA assigns a value to the impact/importance of a play.
Think of it as an answer to a hypothetical question:
"Instead of points being scored only when a team scores a touchdown or field goal, what if teams were awarded fractions of points after each play?"
How it's measured
Let's start with a (simplified) real-world example:
- A team starts a drive on its own 25-yard line.
- Historically, thousands of drives have started on their own 25-yard line.
- Some drives end up scoring touchdowns or field goals, some do not.
- If we average all of these historical drives, we know that these drives score an average of 1.1 points per drive. Thus, having the ball on the 25-yard line is worth 1.1 expected points.
- On the next play, the offensive team gains 35 yards on a pass play.
- The team now has a 1st and 10 on their opponents' 40-yard line.
- Historically, teams that have a 1st and 10 on their opponents' 40-yard line score 3.3 points per drive on average.
- Therefore, the impact of the 35-yard pass play is 3.3 - 1.1 = 2.2.
- The EPA of the 35-yard pass play is 2.2.
The EPA of a play is the difference between the expected points before and after the play.
While the previous example uses basic calculations, statisticians calculate expected points using historical play-by-play data and assign an expected points value to each yardline on the football field while considering:
- Field position
- Down and distance
- Game time remaining
What is considered a "good" EPA per play?
- Most EPA values range roughly from -8 to +8, but there are some outliers (think of a pick-6 in the end zone on the last play of the game).
- The higher the EPA value, the better the impact of the play for the offense.
- The lower the EPA value, the better the impact of the play for the defense.
How to use it in analysis
EPA helps answer some practical questions about a game and team's performance:
- Which plays were the most/least impactful?
- On average, how impactful was one offense/defense compared to the other?
- Did the average EPA per play of an offense/defense change over the course of a game or season?
- How extreme were the good/bad plays for an offense/defense?
- Which players were involved in the most/least impactful plays?
- This can go on and on...which is what makes it a great metric!
More information
Credits
The concept was first introduced in 1971 by Virgil Carter, a graduate student who also happened to be an NFL quarterback for the Chicago Bears.
Carter and his professor at Northwestern University, Robert Machol, published their findings based on the play-by-play data from the first half of the 1969 season in a research paper titled "Operations Research on Football," which analyzed expected points based on down-and-distance situations.
Success Rate
"Staying ahead of the chains"
Refers to an offense consistently being in favorable down-and-distance scenarios
What it represents
On offense: The offense's ability to consistently gain the yards needed to keep drives moving downfield against a defense.
On defense: The inverse. The defense's ability to put offenses in more difficult down-and-distance scenarios.
Why it matters
Success rate (SR) evaluates how effectively an offense performed against a defense. Instead of using yards-per…carry (or pass), it takes into account the down-and-distance of each play.
By the numbers
Dating back to the 2018 season, teams with higher success rate than their opponents win 66.9% of games (95% CI: 64.7-68.9%), which is significantly above random chance (p < 0.001).
How it's measured
Success rate = (Number of successful plays) / (Total number of plays)
A “successful” play for an offense is determined by how many yards were gained in a play for its given down.
The chart below illustrates the number of yards required on each down to consider the play successful.
What makes a play “Successful”
| 1st Down | 2nd Down | 3rd Down | 4th Down | |
|---|---|---|---|---|
| Success = gaining this percentage of the remaining yards needed for a first down | 50% | 70% | 100% | 100% |
| Example: | 1st and 10 A successful play gains 5 yards or more (50% x 10 yards = 5 yards) | 2nd and 4 A successful play gains 3 yards or more (70% x 4 yards = 2.8 yards) | 3rd and 2 A successful play gains 2 yards or more (100% x 2 yards = 2 yards) | 4th and 1 A successful play gains 1 yard or more (100% x 1 yard = 1 yard) |
What is considered a “good” success rate?
At Sports Snaps, we compare the success rate of a team against the success rates for home and away teams in the last 3 seasons.
This bullet chart features colored steps and text labels, which identify the percentage of historical success rates that were better than (or worse than) the given measurement.
This example has a SR of 0.47. It is higher than 75 percent of the total success rates measured (SR of 0.45), which is great and therefore the bar is green.
Defensive success rate
For every play that an offense was not successful, by defaut the defense was successful.
Defensive success rate = (Total number of defensive successful plays) / (Total number of plays)
How to use it in analysis
Success rate helps answer some practical questions about a team's performance:
- Did the offense move the ball downfield consistently?
- How often did the offense gain enough yards to keep drives moving downfield?
More information
Credits
Early conceptual roots: Bill Walsh, the legendary Stanford and San Francisco 49ers coach, introduced the idea that successful offenses stay "on schedule" and convert plays earlier in a down sequence, rather than relying solely on third-down conversions.
Quantification and refinement:
- Football Outsiders: developed success rate as a true statistical measure of efficiency.
- Bill Connelly: adapted and developed the success rate metric for college football.
Explosiveness
Expected Points Added (EPA) per successful play
"The offense is generating chunk plays."
A play that results in a significant gain of yardage is a “chunk play”.
What it represents
An explosive play is a play that efficiently puts the offense in a better position to score points, usually gaining 10-20 yards or more.
How it's measured
This metric has two parts:
- Expected points added (EPA)
- Successful plays
Let's start by understanding its parts.
Understanding the components
"That play may have well cost the team points."
While a play may not result in points, it is so significant that it is considered to have had an impact on the final score.
What it represents
When analyzing gameplay impact, expected points added (EPA) per play is a way to measure which plays were pivotal moments in a game, and which weren't.
Technically, EPA measures how many points can be expected to be added to the score of the possessing team based on field position and the game situation.
Why it matters
Not all plays in a football game are equal. Some plays don't matter much to either team's success, while some plays determine who wins or loses.
While a play may not seem important on the surface, like a 2 yard loss on a 1st down, it can have a surprising impact on the final score if we look closely at the game situation. EPA helps quantify this impact.
EPA assigns a value to the impact/importance of a play.
Think of it as an answer to a hypothetical question:
"Instead of points being scored only when a team scores a touchdown or field goal, what if teams were awarded fractions of points after each play?"
How it's measured
Let's start with a (simplified) real-world example:
- A team starts a drive on its own 25-yard line.
- Historically, thousands of drives have started on their own 25-yard line.
- Some drives end up scoring touchdowns or field goals, some do not.
- If we average all of these historical drives, we know that these drives score an average of 1.1 points per drive. Thus, having the ball on the 25-yard line is worth 1.1 expected points.
- On the next play, the offensive team gains 35 yards on a pass play.
- The team now has a 1st and 10 on their opponents' 40-yard line.
- Historically, teams that have a 1st and 10 on their opponents' 40-yard line score 3.3 points per drive on average.
- Therefore, the impact of the 35-yard pass play is 3.3 - 1.1 = 2.2.
- The EPA of the 35-yard pass play is 2.2.
The EPA of a play is the difference between the expected points before and after the play.
While the previous example uses basic calculations, statisticians calculate expected points using historical play-by-play data and assign an expected points value to each yardline on the football field while considering:
- Field position
- Down and distance
- Game time remaining
What is considered a "good" EPA per play?
- Most EPA values range roughly from -8 to +8, but there are some outliers (think of a pick-6 in the end zone on the last play of the game).
- The higher the EPA value, the better the impact of the play for the offense.
- The lower the EPA value, the better the impact of the play for the defense.
How to use it in analysis
EPA helps answer some practical questions about a game and team's performance:
- Which plays were the most/least impactful?
- On average, how impactful was one offense/defense compared to the other?
- Did the average EPA per play of an offense/defense change over the course of a game or season?
- How extreme were the good/bad plays for an offense/defense?
- Which players were involved in the most/least impactful plays?
- This can go on and on...which is what makes it a great metric!
More information
Credits
The concept was first introduced in 1971 by Virgil Carter, a graduate student who also happened to be an NFL quarterback for the Chicago Bears.
Carter and his professor at Northwestern University, Robert Machol, published their findings based on the play-by-play data from the first half of the 1969 season in a research paper titled "Operations Research on Football," which analyzed expected points based on down-and-distance situations.
A “successful” play for an offense is determined by how many yards were gained in a play for its given down.
The chart below illustrates the number of yards required on each down to consider the play successful.
What makes a play “Successful”
| 1st Down | 2nd Down | 3rd Down | 4th Down | |
|---|---|---|---|---|
| Success = gaining this percentage of the remaining yards needed for a first down | 50% | 70% | 100% | 100% |
| Example: | 1st and 10 A successful play gains 5 yards or more (50% x 10 yards = 5 yards) | 2nd and 4 A successful play gains 3 yards or more (70% x 4 yards = 2.8 yards) | 3rd and 2 A successful play gains 2 yards or more (100% x 2 yards = 2 yards) | 4th and 1 A successful play gains 1 yard or more (100% x 1 yard = 1 yard) |
Measuring explosiveness - Putting it all together
To recap, we now understand:
- How to calculate a play's EPA: the difference between the expected points before and after the play.
- What makes a play "successful": the play gains the minimum amount of yards for its down.
To measure how explosive an offense performed, we measure the average EPA of its successful plays.
Wait, why aren't we using yards per play again?
Hopefully after understanding the concept of EPA, we can appreciate that not all yards on the field are equal. A 5-yard gain on 4th and 3 in the red zone is much more valuable than a 5-yard gain on 3rd and 15 at midfield.
While both scenarios yield 5 yards per play, their impact on the game's result varies widely. EPA provides more context for the value of those yards.
Why it matters
Remember, an explosive play is an offensive play that dramatically increases the offense's chances of scoring points.
EPA per play helps us answer a practical question:
How good were a team's good plays... to what extreme?
EPA per successful play is a metric that helps answer this question.
By the numbers
Dating back to the 2018 season, teams with higher EPA per successful play win 60.8% of games (95% CI: 58.6-63.0%), which is significantly above random chance (p < 0.001).
Credits
Bill Connelly is the pioneer of measuring explosive plays with the IsoPPP metric, and has highlighted the importance of explosive plays to the outcomes of college football games.
Scoring Efficiency
Points per scoring threat (PPST)
"This is a bend but don't break defense."
A term used when a defense allows opponents to get near their end zone, but does not allow them to score points.
What it represents
How efficiently offenses actually score points when they get near their opponent's end zone.
Why it matters
7 points are better than 3…or 0 points.
Put simply, can teams actually score points when they're within reaching distance?
By the numbers
Dating back to the 2018 season, teams with higher PPST than their opponents won 66.9% of games (95% CI: 64.8-68.9%), which is significantly above random chance (p < 0.001).
How it's measured
- A. Count how many times a team has possession within 40 yards of their opponent's end zone (“scoring threats”).
- B. Add how many points they scored during those “threats”
Points per scoring threat (PPST) = (b)Total points scored during scoring threats / (a)Total number of scoring threats
What is considered “good”
Let's consider this question from both perspectives:
- On offense: 8 points per threat would mean they scored a touchdown and a 2-point conversion every time they approached the end zone. An ideal scenario.
- On defense: they hopefully “bend but don't break”. Meaning offenses may get near their end zone, but are kept off the scoreboard with 0 points.
Realistically, teams perform somewhere between 0 and 8.
Benchmarking
At Sports Snaps, we calculate this metric for three seasons' worth of games and use that sample to compare performances.
This bullet chart features colored steps and text labels, which identify the percentage of historical scoring efficiency rates that were better than (or worse than) the given measurement (the bar).
The example above has a PPST of 1.5, putting it below the 10th percentile (2.0).
This means that more than 90 percent of the PPSTs measured were more than 2.0. The team's 1.5 performance is not good, and therefore the bar is red.
Credits
Bill Connelly quantified the importance of measuring Points Per Trip inside 40 ("Scoring Opportunities") as one of his five factors for determining college football results.
Havoc Rate
"The defense wants to create negative plays."
Refers to a defense making plays to disrupt the offense's ability to move the ball down the field.
What it represents
Havoc rate represents how often a defensive unit actively creates disruptive plays that significantly reduce the opposition's ability to gain yards or score points, like a sack or interception.
Why it matters
Limiting the opposition's ability to score points is half of the battle in outscoring the other team.
While defensive success rate measures the overall efficiency of a defense from a high-level...
Havoc rate accounts for the disruptive plays a defense made that put the opposition in a disadvantageous position.
By the numbers
Dating back to the 2018 season, teams whose defense generated higher havoc rates than their opponents win 72.9% of games (95% CI: 70.9-74.8%), which is significantly above random chance (p < 0.001).
How it's measured
Havoc rate is the percentage of defensive plays that created “havoc”:
- Tackles for a loss of yards, including sacks
- Forced fumbles
- Pass break-ups
- Interceptions
Havoc Rate = (Havoc plays / Total defensive plays)
What is considered “good”?
At Sports Snaps, we calculate this metric for three seasons' worth of games and use that sample to compare performances.
This bullet chart features colored steps and text labels, which identify the percentage of historical havoc rates that were better than (or worse than) the given measurement (the bar).
The example above has a Havoc Rate of 0.17 (17%), placing it fractionally above the 75th percentile (0.17).
This means that it is higher than 75 percent of the havoc rates measured, which is great, and therefore the bar is green.
How to use it in analysis
Havoc rate is a useful metric to evaluate how often defenses make disruptive plays.
However, it should be evaluated alongside other defensive metrics, like defensive success rate and scoring efficiency, to paint a more accurate picture of the unit's overall performance.
Credits
Bill Connelly invented and quantified the importance of measuring havoc rate for college football games.
Field Position Advantage
Average drive starting distance from the end zone
"Winning the field position battle"
Refers to a team consistently starting its offensive drives closer to the opponent's end zone and/or forcing the opponent to start its drives farther from the end zone.
What it represents
The number of yards a team must travel to score points on an average possession , compared to their opponents.
Why it matters
On offense, the closer a team starts its possession from the end zone, the more likely it is to score points.
On defense, the farther away the opposing offense starts its possession from the end zone, the less likely they are to score points.
By the numbers
Dating back to the 2018 season, teams whose defense had a longer average distance to defend than their opponents win 68.3% of games (95% CI: 66.2-70.4%), which is significantly above random chance (p < 0.001).
How it's measured
By taking the average yard distance from the end zone a team starts its possessions.
What is considered “good”?
At Sports Snaps, we calculate this metric for three seasons' worth of games and use that sample to compare performances.
This bullet chart features colored steps and text labels, which identify the percentage of historical average field positions that were better than (or worse than) the given measurement (the bar).
The example above has an average distance to score of 67.1 for the away team (left) and 76.8 for the home team (right).
The away team started its drives on average 67 yards away from their target end zone, which is just under the 25th percentile, 67.9. This means that 75% of the average starting field positions measured were farther than 67.9 yards.
Since needing to travel a shorter distance to score points is an advantage, the smaller the number, the better, and the bar is therefore green.
How to use it in analysis
Comparing the averages of the away team and home team, we can say that the away team “won the field position battle” by nearly 10 yards [76.8 - 67.1 = 9.7 yards]. 10 yards may not seem like a big deal, but we should consider that this is only the average.
Let's assume each team in this game had 10 possessions. This means that the away team had an advantage of 10 yards on each possession: 10 yards x 10 possessions = 100 total yards.
The away team had to travel a full length of the field less than the home team to try and score points, a massive advantage!
Credits
Bill Connelly quantified the importance of measuring field position advantage as one of his five factors for determining college football results.
Penalty EPA
Total EPA of Dead-ball Penalties
"The team's lack of discipline is costing them."
Refers to a team drawing so many penalties that they are significantly hurting their chances of winning.
What it represents
Penalties lead to a change in field position and can even result in a loss of a down. How significant were a team's penalties to the game's outcome? This metric provides some clarity.
Why it matters
Considering the full context of a team's penalties, including the number of penalties, where on the field, and when the penalty occurred, they can have a large impact on the game's outcome.
How it's measured
In a game, penalties award yards to the team that suffered the penalty. As we learned with expected points added (EPA), not all yards are equal. Instead of tallying a team's total penalty yards, we:
Sum the EPA of the penalty plays to determine the impact those penalties had on the final score.
Note: Currently, only dead-ball penalties are included in this metric due to data limitations. We hope to account for all penalties in future releases. Dead-ball penalties occur before a play starts.
What is considered “good”?
Penalties against a team are always negative. Ideally, a team does not commit any penalties in a game, and this metric's value is 0.
Realistically, in just about every game, teams commit penalties.
At Sports Snaps, we calculate this metric for three seasons' worth of games and use that sample to compare performances.
This bullet chart features colored steps and text labels, which identify the percentage of historical average field positions that were better than (or worse than) the given measurement (the bar).
The example above has a total penalty EPA of -9.6, placing it below the 10th percentile (-7.2).
This means that it is worse than 90 percent of the total game penalty EPAs measured, which is absolutely terrible, and therefore the bar is deep red.
How to use it in analysis
Penalty EPA is a useful metric to evaluate how how impactful penalties were to the game's outcome.
A fair comparison would be to look at the total penalty EPA of a team against the total penalty EPA of the opponent to determine the marginal impact of its penalties.