- Joined
- Jul 18, 1998
- Messages
- 25,252
- Reaction score
- 54,786
Online
Interesting article I meant to post last month
================================
As the football season unspools in the early fall and NFL enthusiasts debate the strengths and weaknesses of their favorite teams, power rankings occasionally take center stage. They don’t mean anything — this isn’t college football, where humans still select the playoff teams — but with the postseason months away, these rankings offer a framework for understanding how the league’s 32 teams are sorting themselves out. They’re also, to put it kindly, not very rigorous.
This scenario played out when I did a radio spot with Brian Mitchell and J.P. Finlay last week (and every Wednesday at 11 a.m. on 106.7 the Fan, tune in!), discussing where the Washington Commanders should rank two weeks into the season.
At the time, many of the most popular power rankings listed the 2-0 Commanders as an above-average team, while my own ratings had them 25th, inside the league’s bottom third. As you can imagine, Washington fans were upset and dumbfounded at such a disconnect. Fast forward a week and, after a 37-3 home loss to the Buffalo Bills, the Commanders have dropped to a consensus ranking of just 21st.
Time for a victory lap? Maybe. At the very least, it’s a useful starting point to illustrate how there are two distinct approaches to sorting NFL teams: subjective and objective power rankings. Understanding the distinction is crucial to grasping the nuances of evaluating teams’ talent and potential — and to not being fooled by early results.
Subjective NFL power rankings are determined by the opinions and judgments of individuals, or possibly a group of experts. These rankings are often based on personal observations, plus individual interpretations of team performance and other factors. Some even defend rankings based on a team’s intangibles (remember Skintangibles?), which is odd, because by definition, an intangible is impossible to qualify. Most obviously, subjective rankings largely mimic the standings, with teams ranked from the most to least wins rather than based on the true strength of performance.
To illustrate this point, let’s look at a few subjective power rankings — we can use NFL.com, ESPN, The Score and USA Today — heading into Week 4. As you would expect, the most recent subjective rankings have the three remaining undefeated teams (the San Francisco 49ers, Miami Dolphins and Philadelphia Eagles) at or near the top, and the three winless teams (the Carolina Panthers, Denver Broncos and Chicago Bears) at the bottom. The rest of the squads are slotted in between, mostly by record.
There are some deviations, but as you can see, there is a close link between a team’s win-loss record and where it is ranked. In mathematical terms (using linear regression), 70 percent of the consensus ranking appears to be influenced by how many games a team has won. That is why I and many other analysts discount these type of lists: They are a regurgitation of the standings, with little added context. Look at some of the best known subjective rankings from last season, and you will see how closely they mimic the standings week after week.
Objective NFL power rankings, on the other hand, are generated using statistical analysis and data-driven algorithms. These rankings rely on quantifiable metrics, and aim to provide an unbiased and standardized assessment of the 32 teams based on measurable criteria, providing a more consistent and transparent view of team strength. They are often used by sports analysts and gamblers to make predictions and assess team performance without the influence of opinions — and they have almost nothing to do with win-loss record.
Those records can be influenced by factors that are not likely to repeat, such as severe weather, penalty calls and turnovers. A tipped pass that becomes an interception or a fumble recovery in a scrum can be unpredictable, even if they lead to a win.
Some examples of the objective method can be found in ESPN’s NFL Football Power Index, Kevin Cole’s Substack “Unexpected Points,” Aaron Schatz’s Defense-adjusted Value Over Average and Nfelo. These ratings use mathematical algorithms to calculate which teams are better than others. As a result, just 47 percent of these rankings are influenced by how many games a team wins; the balance is based on less-traditional metrics such as expected points added, success rate, net yards per play and points per drive, although every version will have its own formula...............
================================
As the football season unspools in the early fall and NFL enthusiasts debate the strengths and weaknesses of their favorite teams, power rankings occasionally take center stage. They don’t mean anything — this isn’t college football, where humans still select the playoff teams — but with the postseason months away, these rankings offer a framework for understanding how the league’s 32 teams are sorting themselves out. They’re also, to put it kindly, not very rigorous.
This scenario played out when I did a radio spot with Brian Mitchell and J.P. Finlay last week (and every Wednesday at 11 a.m. on 106.7 the Fan, tune in!), discussing where the Washington Commanders should rank two weeks into the season.
At the time, many of the most popular power rankings listed the 2-0 Commanders as an above-average team, while my own ratings had them 25th, inside the league’s bottom third. As you can imagine, Washington fans were upset and dumbfounded at such a disconnect. Fast forward a week and, after a 37-3 home loss to the Buffalo Bills, the Commanders have dropped to a consensus ranking of just 21st.
Time for a victory lap? Maybe. At the very least, it’s a useful starting point to illustrate how there are two distinct approaches to sorting NFL teams: subjective and objective power rankings. Understanding the distinction is crucial to grasping the nuances of evaluating teams’ talent and potential — and to not being fooled by early results.
Subjective NFL power rankings are determined by the opinions and judgments of individuals, or possibly a group of experts. These rankings are often based on personal observations, plus individual interpretations of team performance and other factors. Some even defend rankings based on a team’s intangibles (remember Skintangibles?), which is odd, because by definition, an intangible is impossible to qualify. Most obviously, subjective rankings largely mimic the standings, with teams ranked from the most to least wins rather than based on the true strength of performance.
To illustrate this point, let’s look at a few subjective power rankings — we can use NFL.com, ESPN, The Score and USA Today — heading into Week 4. As you would expect, the most recent subjective rankings have the three remaining undefeated teams (the San Francisco 49ers, Miami Dolphins and Philadelphia Eagles) at or near the top, and the three winless teams (the Carolina Panthers, Denver Broncos and Chicago Bears) at the bottom. The rest of the squads are slotted in between, mostly by record.
There are some deviations, but as you can see, there is a close link between a team’s win-loss record and where it is ranked. In mathematical terms (using linear regression), 70 percent of the consensus ranking appears to be influenced by how many games a team has won. That is why I and many other analysts discount these type of lists: They are a regurgitation of the standings, with little added context. Look at some of the best known subjective rankings from last season, and you will see how closely they mimic the standings week after week.
Objective NFL power rankings, on the other hand, are generated using statistical analysis and data-driven algorithms. These rankings rely on quantifiable metrics, and aim to provide an unbiased and standardized assessment of the 32 teams based on measurable criteria, providing a more consistent and transparent view of team strength. They are often used by sports analysts and gamblers to make predictions and assess team performance without the influence of opinions — and they have almost nothing to do with win-loss record.
Those records can be influenced by factors that are not likely to repeat, such as severe weather, penalty calls and turnovers. A tipped pass that becomes an interception or a fumble recovery in a scrum can be unpredictable, even if they lead to a win.
Some examples of the objective method can be found in ESPN’s NFL Football Power Index, Kevin Cole’s Substack “Unexpected Points,” Aaron Schatz’s Defense-adjusted Value Over Average and Nfelo. These ratings use mathematical algorithms to calculate which teams are better than others. As a result, just 47 percent of these rankings are influenced by how many games a team wins; the balance is based on less-traditional metrics such as expected points added, success rate, net yards per play and points per drive, although every version will have its own formula...............
Hmmmm... they're being pretty generous these days.