Three days, seven rounds and 257 picks later, the culmination of months of planning and detail for 32 NFL franchises has come and gone. All that’s left now is to sort through the results. Last year, discussed in detail why parsing the noise for signal in draft grading can be so difficult, mostly because the draft media industrial complex doesn’t focus on what’s most important and predictive.

Draft content is mostly about who will be the best players in the NFL, so that’s also what drives most draft grades. The problem is that all the research around the draft shows that giving kudos or demerits based on how teams drafted versus your expectation is a fool’s errand. There’s only a little correlation between a loss in value when teams reach for a player versus consensus, and nearly no correlation if they draft a player later that we think they’ll go before the draft. As I explained in my draft grading philosophy primer:

It isn’t that prospect evaluation doesn’t matter, it’s that NFL teams pour so many resources and expertise into the task that relative success and failure will be narrow, especially over a longer timeline. You shouldn’t take away from the data that no teams are really good at prospect evaluations. Instead it’s that all teams are really good, and likely much better than outside evaluators.

While we can’t rely too heavily on the specific prospects the team chose at their draft slots, there are larger macro forces that teams can leverage to, on average, better their draft outcomes. The two primary drivers of quantifiable draft value are:

• Trade value gained/lost

• Positional value gained/lost

There are extended bodies of research on the value teams can gain trading back in the NFL, and in this analysis I quantify that value through looking at the equivalent value in NFL salary of players similar to those drafted. Rookie contracts are set at artificially low amounts, so the difference in player value over contract value is what the team gains with each draft selection, or surplus value. Different positions are valued much differently in NFL contract markets, so hitting on an All-Pro level running back in the draft gives less surplus value at a particular draft selection than landing an elite edge rusher. When teams are faced with the choice of wear to allocate their draft capital, especially in early rounds, accounting of positional value will skew the odds of success in their favor.

Looking at all the NFC teams in terms of value added ($M) in trades (average expected surplus of picks traded away versus those gained) and what I call selection value (a combination of positional value and a much smaller contribution from taking players after/before their consensus value), you can see that a couple big trades skews the results, and how powerful the impacts of those trades can be.

Digging further into the two components of selection value, we see that positional value has a bigger relative impact, more than three times the standard deviation of value gained by drafting players versus consensus rankings.

The reach/steal paradigm is going to be the primary driver of most draft grades you see from media analysts this week, yet it’s least impactful. Research shows that steals are much less likely to produce benefit than reaches cause harm, so you should be especially skeptical of draft grades that lean heavily into teams who drafted the most players far after their consensus expectations.

Let’s get to our analytical grading, going through each NFL division here. Again, if you want more details why I believe this is the best method for analyzing drafts in a predictive manner, check out my draft grading primer.

It was a draft of extremes for the Philadelphia Eagles, losing nearly $20 million of surplus value in trades, but gaining it all back and then some in positional value. Analytical formulas are never going to favor trading away multiple third-round picks for the rights to pay free-agent market value to a veteran player. On the other hand, Jonathan Greenard will fill a valuable role in the Eagles pass rush next season, and his contract costs are likely weighted towards future years.

The Washington Commanders didn’t show much concern with positional value in the draft, taking off-ball linebacker Sonny Styles at seventh overall. Styles wasn’t seen as a reach, ranking fifth on the consensus big board.

Read more

Three days, seven rounds and 257 picks later, the culmination of months of planning and detail for 32 NFL franchises has come and gone. All that’s left now is to sort through the results. In the past, I discussed in detail why parsing the noise for signal in draft grading can be so difficult, mostly because the draft media industrial complex doesn’t focus on what’s most important and predictive.

Draft content is mostly about who will be the best players in the NFL, so that’s also what drives most draft grades. The problem is that all the research around the draft shows that giving kudos or demerits based on how teams drafted versus your expectation is a fool’s errand. There’s only a little correlation between a loss in value when teams reach for a player versus consensus, and nearly no correlation if they draft a player later that we think they’ll go before the draft. As I explained in my draft grading philosophy primer:

It isn’t that prospect evaluation doesn’t matter, it’s that NFL teams pour so many resources and expertise into the task that relative success and failure will be narrow, especially over a longer timeline. You shouldn’t take away from the data that no teams are really good at prospect evaluations. Instead it’s that all teams are really good, and likely much better than outside evaluators.

While we can’t rely too heavily on the specific prospects the team chose at their draft slots, there are larger macro forces that teams can leverage to, on average, better their draft outcomes. The two primary drivers of quantifiable draft value are:

• Trade value gained/lost

• Positional value gained/lost

There are extended bodies of research on the value teams can gain trading back in the NFL, and in this analysis I quantify that value through looking at the equivalent value in NFL salary of players similar to those drafted. Rookie contracts are set at artificially low amounts, so the difference in player value over contract value is what the team gains with each draft selection, or surplus value. Different positions are valued much differently in NFL contract markets, so hitting on an All-Pro level running back in the draft gives less surplus value at a particular draft selection than landing an elite edge rusher. When teams are faced with the choice of wear to allocate their draft capital, especially in early rounds, accounting of positional value will skew the odds of success in their favor.

Looking at all the AFC teams in terms of value added ($M) in trades (average expected surplus of picks traded away versus those gained) and what I call selection value (a combination of positional value and a much smaller contribution from taking players after/before their consensus value), you can see that big, early trades skew the results, and how powerful the impacts of those trades can be.

Digging further into the two components of selection value, we see that positional value has a bigger relative impact, more than three times the standard deviation of value gained by drafting players versus consensus rankings.

The reach/steal paradigm is going to be the primary driver of most draft grades you see from media analysts this week, yet it’s least impactful. Research shows that steals are much less likely to produce benefit than reaches cause harm, so you should be especially skeptical of draft grades that lean heavily into teams who drafted the most players far after their consensus expectations.

Let’s get to our analytical grading, going through each AFC division here. Again, if you want more details why I believe this is the best method for analyzing drafts in a predictive manner, check out my draft grading primer.

Through a series of trades in the earlier portions of the draft, the Buffalo Bills were able to accumulate strong surplus value, while also gaining a bit being disciplined according to big board rankings. The Bills finished with the second most total surplus value in the draft, adding numerous players at premium non-quarterback positions, including an edge rusher, cornerback, offensive tackle and wide receiver with their first four selections.

There was nothing spectacular about the 2026 NFL draft for the New York Jets, but they added relative value simply by avoiding trading up (for the most part), drafting the right positions, and, most impactfully, staying disciplined according to the consensus big board. The Jets lost some positional value by taking tight end Kenyon Sadiq, but that was more than offset with taking edge rusher David Bailey with second overall pick, as edge rushers have established themselves as clearly the second most valuable position to quarterbacks in the NFL.

Read more

I’m going to apply the same surplus value methodology to the Day 2 pick as I did for the first round. Come Monday, I’ll have comprehensive team-by-team numbers for total surplus value gained by aggregating trade value, positional value and reach/steal value.

SECOND ROUND VALUES

The “Value” number is my estimate for salary cap equivalent for the player based on positional value and historical trends for prospects taken at his draft slot. Contract is the annual amount due, and surplus value is the difference that the team gains acquiring the player through the draft.

Read more

We’ve finally gotten to and finished the first round of the NFL draft. There will be a ton of content hitting the streets in the next 12 hours to tell you the feeling and opinions on which teams made the best use of their draft capital.

In this post, I eschew the vibes-based analysis that dominates NFL media and judge the first round on explicit metrics that matter. I leveraged the excellent weighted mock draft data from Grinding the Mocks and consensus media Big Board data aggregated from Jack Lichtenstein on his beautifully formatted website. In addition, I estimated the NFL value and subtracted the contractual commitment of each pick to determine the player’s surplus value, or value you’d have to pay in free agency over what the rookie contract demands. I went through this exercise earlier this month building a more useful Big Board using Sports Info Solutions draft grading and positional contract data from Over the Cap.

Let’s get to the data. The first section below has a table with a row for every pick. It has the pick number, the mock draft-based estimated draft position, the consensus big board ranking, and the estimated values of the prospect, his contract amount and the surplus value to the team. The “NFL values” for the picks are based on my research for draft curve value by pick and position.

I also have a plot showing the differences better than the actual pick numbers and the consensus opinions on mocks and big boards. Let’s get to the data

TOP-12 PICKS OF THE ROUND

The “Value” number is my estimate for salary cap equivalent for the player based on positional value and historical trends for prospects taken at his draft slot. Contract is the annual amount due, and surplus value is the difference that the team gains acquiring the player through the draft.

The expected No. 1 pick by mock drafts was Fernando Mendoza, and he goes first overall. His big board consensus ranking was a bit lower at fourth, and that assessment is likely inflated by assumptions of where he would go in the draft. For what it’s worth, Sports Info Solutions had Mendoza ranked 25th on its position-agnostic big board. There’s no higher expected NFL value in the draft than for a quarterback taken No. 1 overall, which offsets the higher price in contract costs to give Mendoza an estimated average surplus value of a little under $18 million a season over his four-year rookie contract.

The New York Jets followed next with the selection of David Bailey, who represents the second most valuable position in the NFL, according to what league general managers are willing to shell out in veteran costs on the free market1 . Because of the exponential increase in contract costs near the top of the draft, Bailey’s second-highest projected “NFL Value” only translates into the seventh highest surplus value of the top-12 picks. What isn’t represented in those numbers is the range of outcomes that truly represent the value of the picks, not only the mean estimates. Bailey has a greater likelihood of being a truly elite player, and therefore exceedingly valuable asset, than players drafted below him at the same position. That said, the degree to which early taken players are assumed in NFL circles to have higher upside is probably overstated.

The running back discussion will never die, and the selection of Jeremyiah Love at third overall by an Arizona Cardinals team that isn’t projected to be very good (Cardinals over/under win total is currently 4.5 and skewed to the downside at DraftKings) will keep the most worn-out debate in the mix this season. Using expected value numbers derived through positional veteran contracts and pick-adjusted hit rates, this selection appears to be a wash - the Cardinals took one of the most valuable assets in the NFL (an early draft pick) and turned it into a pumpkin. However, veteran contracts are less representative of NFL value for running backs than at any other position, with their value peaking much earlier, often playing at an elite level immediately as rookies.

NFL teams are smart to low-ball veteran running backs in contract talks, not knowing if their useful NFL lives will last much beyond their mid-20s. That’s not an issue for rookie running back contracts, but you also can’t make the mistake to assume top-10 running back selection will automatically turn into an elite player. Yes, running backs are more likely to “hit” in terms of productive early years than at other positions, but you can’t ignore that busts exist. From 2010 to 2025, nine running backs were selected in the top-10 picks, which includes some relative disappointments, like Trent Richardson (third overall), Leonard Fournette (fourth) and CJ Spiller (ninth).

Carnell Tate projects as a solid value selection at fourth overall, with wide receiver contracts skyrocketing in recent years. The hedge on the increasing value of wide receivers is that elite ones are more likely to be found later in the draft than at positions like edge and offensive tackle.

The valuations of picks of Arvell Reese and Sonny Styles take a big hit due to positions, with off-ball linebackers generally projecting as “tier 3” values, below several other positions.The market for veteran linebackers has been stagnant recently, with elite players commanding far less than at edge and interior defensive line, not to mention cornerback. These players will likely generate surplus value, though the New York Giants and Washington Commanders may have been better served to address the position at market rates in free agency, and then use their early draft picks to fill other needs.

REST OF THE FIRST ROUND

Read more

Subscribe now

Every March and April I get progressively confused, angry and dismissive of NFL draft media "Big Boards”, rankings and pointless thought-exercises. Most of the draft analysts producing these boards and rankings do a fantastic job with their core competency: evaluating college prospects. I wouldn’t dare to attempt digging through hours of tape and formulating how good of a player each prospect will be in the NFL.

Perhaps I could refine the grading process by identifying and systematically countering various biases, or figure out a way to quantify traits for easier comparison and memory retention. But I would only be making changes to the results on the edges. Draft evaluators are very good at rank ordering players within their position, as indicated by the strong correlations between prospect draft positions and NFL success.

What frustrates me isn’t the individual prospect grading, but the process of aggregating all the non-specialist positions into one list, either titled as a “Big Board” or simply “Prospect Rankings”. It’s fine to say that Jeremiyah Love is more likely to be an elite running back than Fernando Mendoza an elite quarterback, but does that mean you’d follow some rankings and draft the former first, knowing the vast difference in value their respective positions bring to their teams? I’d hope not. But the rankings and Big Boards out there usually aren’t explicit in their application. If you rank a player in the top-5, but wouldn’t draft them at that position, what exactly is the point of your inter-positional rankings?

There are some obvious answers. People like to see a list of all the players together much more than position-specific rankings, so that’s what you give them. Less cynically, some draft analysts in the media might actually believe that you should be position-agnostic when building a draft board, and then draft accordingly. Yes, it’s an antiquated way to think, pushing against ever-increasing mounds of evidence, but a decision-maker like Dave Gettleman was running an NFL front office just four years ago. We haven’t made it completely out of the analytical dark ages, though there are signs of an shift with the hiring of younger general managers across the NFL.

Rather than continuing to complain on social media about what others are doing, I decided to build my own Big Board for the 2026 NFL draft, leveraging the outstanding prospect grades from Sports Info Solutions (SIS), which you can find on their website. What I’m doing here is combining the best of the prospect-grading world, with analytical concepts of NFL value and surplus value based on position, which has been a big focus of my own research.

I ranked the top-100 players by projected NFL value, which I define as the equivalent salary you’d expect for a second contract, based on the number grading from SIS, which comes with a description for each grade cohort. I then weigh the projected NFL value in annual per year contract (APY) against the designated contract costs for each draft slot for the first four years of a player’s career. No matter the contract cost, teams want to draft players with the highest projected NFL value. But the costs decline significantly by draft position, greatly affecting the cost-including values to NFL teams. The player selected 11th makes roughly half of the No. 1 pick in the draft.

I determined the NFL values by position using a mix of art and science. The science part was looking at the expected percentage of cap value curves for each position, illustrated below. The art part is using the descriptive definitions of the SIS grading (e.g. “Solid Starter”) and marrying them with the current contract values by position at OverTheCap.com.

I like how the analysis played out using these combined values, but there are still ways beyond the scope of what I have done to further improve the process. Some ideas I have that would likely help, but require a lot more time and complication:

• Turning the NFL value projections into ranges of outcomes, instead of a single NFL value number. For some draft slots and positions we care more about the ceiling than the floor, and every prospect represents a distribution of potential.

• Aggregating and scaling multiple grading scales to get a “wisdom of crowds” NFL projection , rather than rely solely on SIS.

• Projecting based on my value metric of NFL Plus/Minus and translating back to dollar-based value, rather looking backwards on what players have been paid.

I’ll work on incorporating this stuff in the future, but I think we get 90% of the way towards making a superior Big Board in the rankings detailed below. Big board descriptions are likely overconfident in how good these players will actually be, as we know lots of highly rated early draft picks bust and come nowhere near their projected value. For that reason, the projections for “NFL Value” and “Surplus Value” are likely too high across the board, but that doesn’t affect the relative rankings.

Each table includes: the updated, position-adjusted player rankings (“Adj Rank”), the traditional rankings based on grades (“Trad Rank”), the player names and positions, the SIS grade and its description, the contract value you would expect for a player at that position of that grade (“NFL Value”), the designated annual contract amount for the draft selection (“Contract”), and the annual surplus value the drafting team would get from selecting the player (“Surplus”). The last three value numbers are all based in millions of dollars.

THE “ANALYTICAL” TOP-10 PROSPECTS OF 2026

The important thing to remember comparing the “Trad Rank” to my “Adj Rank” is that the former, and the corresponding grades, are only the views of one big board. There is great info out there aggregate consensus big board information, and I’ll note when those consensus opinions differ significantly from those of SIS.

Fernando Mendoza ranks all the way down at 25th overall when sorting based purely on SIS’s position-independent grades, though he appears to be a lock to go No. 1 in the NFL draft. SIS is relatively low on Mendoza with many big boards having him at No. 1, equal to his expected draft position. I think SIS’s ranking is fair, when explicitly ignoring positional value.

My analytical adjustments in this analysis agree with his likely draft slot, as even a “Lower-End Starting QB” is worth around $30 million per season in today’s NFL, a significant cost saving from the No. 1 pick’s rookie annual contract cost of $13.6 million. Mendoza’s median, or expected, outcome isn’t the only consideration, with every prospect in reality having a range of outcomes, his value calculation likely goes up. Sometimes quarterbacks we expect to only be mid-level starters, or even worse, end up elite players, by far the most valuable commodities in the NFL, commanding upwards of $60 million per season in today’s NFL.

Read more

** Numbers updated through April 20th

Five years ago, I developed the Improvement Index — a way to quantify the value gained and lost by each team in the offseason.

The index is built on the rigorously researched NFL Plus/Minus metric, which translates player values into intuitive, concrete and position-agnostic currency of points added or lost. The index is the point differential gain or loss each team during the offseason, beginning with the first signing of released players.

The model behind the Improvement Index projects the numbers of snaps and per-play efficiency for each player in every facet of offensive and defensive play: passing, rushing, receiving, blocking, pass rush, run defense and coverage. These models were trained on years of historical data, going back to 2006.

The model incorporates player- and team-level assumptions based on each player’s prior performance, team coaching tendencies, forecasted opponents, and championship odds. These features help the model estimate the likely allocation of snaps among each position group and the likely points-added/lost per snap for each player based on that usage and historical trends.

Some teams will move on the improvement index even if they haven’t added or dropped an influential player, as the total point differential gains must net to zero, i.e. one team’s point differential gains must be offset by others’ losses, and vice versa. Also, the team losing a player will not have the opposite point-differential gain/loss as the team gaining a player, as different quality of players will be gaining and losing snaps on those teams.

Each day, the Improvement Index plot below will be updated, along with a table of all the players joining a new team, with estimates for points-added above a replacement-level player at their positions. It’s important to remember that the points above replacement level number will not match the team gains, as the latter will mostly be based on snaps shifting between better than replacement-level players.

Here is the movement in the index, generated by the confirmed transactions, through 9 a.m. Monday April 20th. Released players will not be added until they’re officially added the NFL transaction list, or the confirmation is solid from multiple NFL insiders, as teams sometimes use the threat of release to renegotiate a contract or find trade partners. Trades will be added when multiple NFL insiders report them, along with signings.

The index includes the effects of relative draft capital for each team, going by the expected points added by draft slot. Most of the advantages are netted out, but some teams do have materially higher/lower expected value in the 2026 NFL draft.

The full list of player movement and value added is available on the Unexpected Points subscriber Google sheet (for paid subscribers). Below the main plot of value shifts by team, there are tables for each division with the exact numbers by division.

Read more

Over the last several years, I released the results of my updated historical modeling on the NFL Scouting Combine since 2006, which quantified the effects on NFL value (according to my Plus/Minus metric) and draft position for all measurements and drills.

In this analysis, I’m going to apply those value models to the defensive player participants at the 2026 NFL Combine. Below I’ll show an updated illustrations for the model results, i.e. the attribute effects of each drill and measurement on early-career Plus/Minus (first four seasons) and NFL draft position.

I’ll then give a table with all the combine participants at each position, and their percentile scores according to the model for draft value and actual NFL value. This will help separate how combine performances will be viewed by NFL teams and how they could project to actual NFL value, very often different things entirely. Prospects with low draft scores but high NFL value scores could be relative discounts in the draft.

I also imputed times and jump measurements for the based on weight-based assumptions if the prospect didn’t participate in the sprints, with a penalty for non-participation. Those prospects who didn’t participate in any of the sprints, jumps or agility drills will not have a score in the model. The “PFF Rank” column represents pre-combine positional prospect rankings.

RUNNING BACKS

I have to be explicit here on what “NFL value” means for running backs: it’s based on my NFL Plus/Minus calculation, not an accumulation of stats that propels running backs to accolades and fantasy football MVPs. Sometimes, those values align, like in the case of Christian McCaffrey. But they don’t have to. Three cone is the one drill with meaningful effect of NFL value, having the strongest length to receiving value in the NFL. Size and speed matter a ton for moving running backs up NFL draft boards, but not as much according to my value model.

I’ve sorted the table below by draft model score (“Draft Percentile”), as all-but-one prospect skipped the three-cone drill, and those who did participate didn’t perform well enough on the drill to make up for their relatively lower results in other drills and measurements.

The winner of the combine by draft a NFL value projections was Mike Washingtion Jr., PFF’s 10th ranked player going into the combine (160th overall, a mid-5th round pick). Washington is big (223 pounds), incredibly fast (4.33 forty) and his jumps were impressive when adjusting for size. There’s reason to suspect that he would have performed relatively worse on the agility drills if he had run them, but it’s encouraging that he increased his final-season college receiving totals to 28 receptions, 12% of team receptions and tied for the third highest total on the Arkansas roster.

Jeremiyah Love was a big smaller and slower than Washington, but running a 4.36 forty at 212 pounds was still a top-notch result. Love didn’t participate in the jumps or agility drills. Love has shown receiving skills on the field, totaling 55 receptions for 509 yards over the last two seasons. Love is head-and-shoulders over the rest of the class on PFF’s big board, ranking fourth overall (up from sixth pre-combine), with the next-closest running back his teammate Jadarian Price at 59th.

Speaking of Price, his results weren’t great or poor, running a sub-4.5 forty at a little over 200 pounds. I’d like to see more receiving production from a back of his size (10 total receptions over the last two seasons) and to have agility times.

The relative loser of the combine was Emmett Johnson, whose 4.56 forty time at 202 pounds looked mediocre in comparison to the blazing runs we saw from other prospects. His three-cone time of 7.32 seconds scores below the 10th percentile for running backs who have run the drill over the last 15 years.

WIDE RECEIVERS

Read more

In this analysis, I’m going to apply those value models to the defensive player participants at the 2026 NFL Combine. Below I’ll show an updated illustrations for the model results, i.e. the attribute effects of each drill and measurement on early-career Plus/Minus (first four seasons) and NFL draft position.

I’ll then give a table with all the combine participants at each position, and their percentile scores according to the model for draft value and actual NFL value. This will help separate how combine performances will be viewed by NFL teams and how they could project to actual NFL value, very often different things entirely. Prospects with low draft scores but high NFL value scores could be relative discounts in the draft.

I also imputed times and jump measurements for the based on weight-based assumptions if the prospect didn’t participate in the sprints, with a penalty for non-participation. Those prospects who didn’t participate in any of the sprints, jumps or agility drills will not have a score in the model.

DEFENSIVE TACKLES

Starting with defensive tackles, we can see that they gain draft value at larger sizes, but NFL value - mostly via pass rushing - as small, agile player. Unfortunately, only six participated in the three-cone drill, but I also inputted times for those who participated in enough of the other drills to do so, with a discount to that assumed time for non-participation.

The winner of the combine by my model was DeMonte Capehart. He’s was the 16th ranked defensive interior prospect by PFF’s big board going into the combine (“PFF Rank”, note that PFF rankings are being updated with results), so his strong performance could lead to a significant rise in draft-stock. The second ranked prospect for PFF Caleb Banks posted great numbers for his draft stock (98th percentile), though my model values bigger interior defenders somewhat less than the market.

Only two of the top-9 prospects going into the combine participated in enough drills to make an assessment. While he didn’t participate in the drills and therefore doesn’t have a model score, PFF’s No. 1 prospect heading into the combine Peter Woods made negative headlines with extremely short arm length. My model doesn’t put a lot into arm length for predicting NFL success, but it matter for draft position. PFF has already dropped Woods from 13th overall on their big board pre-combine to 29th.

EDGE RUSHERS

Read more

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2025-2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

NE vs SEA (-4.5)

I’m sure the film gurus will be able to craft a story of the Seattle Seahawks employing such a brilliant defensive scheme to completely shut down the New England Patriots offense. I’m going to lean more towards a combination of a good scheme and a Patriots offense that wasn’t ready for the competition.

Let’s not forget that the Seahawks defense at home allowed Matthew Stafford and the Rams offense to post 93rd percentile EPA efficiency two weeks ago. The Seahawks defense isn’t unsolvable, but the solution was hidden so deep that the Patriots couldn’t find an answer, the best outcome for great defense play and coaching.

From my perspective, the Patriots played and coached a little scared, concerned about the Seahawks pass rush and running the ball on six of nine first downs in the first half. Perhaps they should have been concerned with Drake Maye taking three sacks on his first nine dropbacks, but that worry doesn’t go away by setting up obvious third down passing situations.

Read more

In this post, I’m going to detail the methodology and results applying my NFL Plus/Minus metric to defenders this season, which includes many of the top candidates for NFL honors, including Defensive Player of the Year and All-Pro selections, plus some names you might not expect.

My NFL Plus/Minus methodology will evolve as time goes on, and the results will also change. Even so, the first runs of results over the last two years point to a few potential gaps in our current understanding of where defender value comes from, and give us some ideas on how to apply historical on/off splits to make assumptions about the broader values of different position groups.

Earlier this week I detailed the NFL Plus/Minus methodology and results for all non-quarterback offensive players (showing that Puka Nacua may have been slightly more valuable than betting favorite Jaxon Smith-Njigba), and in this post we turn to defenders. It’s more difficult to find relevant, objective metrics to judge defenders than we have for offensive players, specifically offensive skill position players. Therefore, there’s a higher reliance on PFF grading in this analysis, especially for run defense, and coverage to a lesser degree.

METHODOLOGY

Using participation data and on- and off-the-field splits have a rich history in other sports for building player valuation metrics. No matter how much we try, the human eye and brain cannot notice the contribution to every player on every play while calculating the exact impact of those actions on the game in a comparable metric. By studying the differences in team success — in this case, using expected points added — when a player is on and off the field, we can capture the entire effect in useful numbers for comparison.

The issue with using on-off splits in football is the time for all players on the field is limited in a 17-game season, and the substitution patterns for different positions don’t allow for much comparison. Without injury, cornerbacks can play every snap of the game, whereas certain interior defenders are rotated consistently. The solution I came up with was to look at the on-off splits of groups of similar players rather than individual players.

I used the nflreadR participation data going back to 2016 — along with other traditional and advanced stats — to cluster similar players for each position in each facet of the game and then compiled the on-off splits of those groups. In doing so, we increase the total samples of measurement and reduce the noise of single-player numbers.

For the NFL Plus/Minus results, this exercise of clustering by role is applied to every position in three facets of defensive play: pass rush, run defense and coverage. The results below detail the points prevented (negative is good) numbers for the top-10 defenders in every position group and then the top-12 overall. I’ll post more commentary after the results and how I’m thinking about them.

Read more

I (re)introduced my favored methodology for measuring player value two years ago, and then applied it to all the non-quarterbacks in the NFL the last three seasons. Tyreek Hill was the most valuable offensive non-quarterback in my 2022 and 2023 seasonal analyses, though Justin Jefferson and Christian McCaffrey won the respective Offensive Player of the Year (OPOY) awards in those seasons. Jefferson and McCaffrey did finish third and second, respectively, in the seasons they won the awards, so my methodology isn’t far detached from voter sentiment, instead emphasizing different aspects of player value.

In 2024, my analysis was further off from the consensus, with Ja’Marr Chase my leader in points added, but Saquon Barkley the runaway winner of the OPOY award. Chase was second in the voting tabulations among non-quarterbacks, but he didn’t receive a single first-place vote. Barkley was relatively low in my value analysis, only ranking third among running backs, behind the ultra-efficiency Jahmyr Gibbs and Derrick Henry. The OPOY is not explicitly designed to give the most weight to “value”, but instead more of the most outstanding player, perhaps explaining the differences in rankings.

In current betting markets, Jaxon Smith-Njigba is the heavy favorite to win the OPOY award. As you’ll see below, I don’t have him ranked first in value, but it’s really close.

METHODOLOGY

I explain the methodology for valuing one facet, or phase, of player contributions in the NFL Plus/Minus introductory post: receiver route running. I don’t want to restate everything from that post, so please look back for clarification on the process. In this analysis, I’m applying that same methodology to the four aspects of non-quarterback value contribution for offensive players (rushing, route running, pass blocking and run blocking).

Before we get into the results of the analysis for every non-quarterback offensive position, let’s walk through a quick, specific example of how clustering works, using the route facet for wide receivers, highlighting some 2025 player results.

Without getting bogged down in the details, the graph above divides all wide receiver seasons from 2016 to 2025 with at least 50 routes into five different, color-coded clusters. For each receiver season, I calculated a number of efficiency and volume statistics and settled upon six for best differentiation receiver types: routes per game, slot rate, yards per route run, touchdown as a percentage of routes, first-down percentage, deep rate (percentage of 20+ yard targets) and yards per reception. I translated these seven features into principal components to minimize multicollinearity and make for easier visualization. The technique I used to form groups of similar receiver seasons is called k-means clustering. With this clustering technique, you choose the number of clusters, or groups, to form.

The dashed arrows extending out from the center of the plot show the directional force of the different features and the length of the arrows indicates the importance of each feature on the cluster assignment. The volume of routes and route-based efficiency for yards, first downs and touchdowns all pull in the same direction to the lower-right. Touchdowns are the least impactful feature, which makes sense as they are the noisiest stat of the group. Higher slot rate is directionally indicated by player seasons at the bottom of the plot, with yards per reception and deep rate mostly opposing features. It makes sense that slot receivers are less likely to go deep and have higher yards per reception, and vice versa for receivers who mostly line up out wide.

From these clusters, you can determine the cluster-averages for team offensive EPA when the receivers are on- or off-the-field, then apply them across the entire cluster, thereby reducing noise. Remember that this is just a simplified example of one clustering sample. In the total analysis, I repeat this exercise 100 times, with randomly assigned numbers of clusters, for each of the facets (route running, rushing, run blocking, pass blocking) for every position group. I then average up all the cluster value estimates for each player to get an overall number. There are also additional steps to add stability to the results, like application of the results through a regression, that bring us to the final numbers.

THE RESULTS

With the nerd stuff out of the way, let’s dive into the results for each position, before bringing them all together to see the most valuable non-quarterbacks on offense in 2025. First, let’s start with wide receivers.

Read more

In this post, I’m going to walk through how my team-strength model views every team offensively and defensively for their 2025 performance going into the Super Bowl, plus details on which teams really belong in the playoffs and exactly how strong they are versus each other (points-based model).

TEAM OFFENSIVE AND DEFENSIVE STRENGTH

*Downloadable power ratings/rankings available in the paid subscriber Google sheet.

My anecdotal glances through Twitter sentiment (something I’m trying to do less and less recently) revealed a lean towards the Seattle Seahawks as the better team, with a material contingent seeing a blowout win over the New England Patriots as the likely outcome. Betting markets agree directionally, with the Seahawks currently 4.5-point favorites on a neutral field in Santa Clara. I’m sure someone has investigated if the Seahawks’ familiarity with the location as a divisional rival of the San Francisco 49ers gives them an edge, but that’s not part of my analysis.

Read more

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2025-2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

DEN vs NE (-3.5)

I guess you could call this a “gritty” win for the New England Patriots over the Denver Broncos is a second-half snowstorm, with a healthy dose of poor offensive play on both sides. The Patriots offense was better (less bad) in about every advanced metric, but only marginally so.

Like many tight contests, the differences in final scores came down to a few plays, specifically a failed 4th & 1 try for the Broncos and a Patriots takeaway that set up their only touchdown. The Patriots also narrowly converted their own 4th & 1 sneak, which ended up somewhat moot when their drive quickly stalled and they kicked a short field goal. Both teams missed two field goals, though the combined loss for the Patriots was -4.8 EPA to -5.4 for the Broncos.

Read more

In this post, I’m going to walk through how my team-strength model views every team offensively and defensively for their 2025 performance going into the Conference Championships, plus details on which teams really belong in the playoffs and exactly how strong they are versus each other (points-based model).

TEAM OFFENSIVE AND DEFENSIVE STRENGTH

*Downloadable power ratings/rankings available in the paid subscriber Google sheet.

The Conference Championships include the three best performing teams by adjusted efficiency through the 2025 regular and postseasons, then there’s the Denver Broncos. Before losing their starting quarterback, the Broncos were a half-tier below the remaining competition, with a strong defense, but only slightly better than average offense.

The primary question hanging over the AFC Championship matchup with the New England Patriots is how to quantify the downgrade from Bo Nix to Jarrett Stidham. Nix was roughly an average quarterback in EPA efficiency this season, slightly worse using my adjustments. The primary negative adjustment for Nix this year was his reliance on receiver yards-after-catch, a trait that should be more transferable to a backup than making accurate throws down the field.

Read more

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2025-2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

NE (-3) vs HOU

The adjusted score had this as a fairly even contest, while the scoreboard was tilted fairly heavily in favor of the New England Patriots from halftime on. Both offenses struggled, but the Houston Texans made the higher impact mistakes, with their five turnovers outweighing three from the Patriots by 15.9 expected points in terms of value lost.

The contrast is best illustrated by one of the Texans’ giveaway being a pick-6 on 1st & 10 during the middle of the second quarter that cost them 8.0 expected points, Drake Maye’s one interception was a hail mary at the end of the first half that virtually no effect on EPA of win probability (-0.2, -0.5%).

Read more

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2025-2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

DEN (-1.5) vs BUF

I’m not surprised to see that the adjusted scores had the Buffalo Bills as the better team in their loss to the Denver Broncos. Normally if a team has a -4 turnover differential and was in the game through overtime, they were considerably better by more predictive measures, like offensive success rate - the Bills were at 51% to the Broncos at 44%.

But games aren’t played on spreadsheets - or so I hear - and turnovers have an enormous impact on results, with the Broncos’ advantage translating into +13.6 EPA. That numbers is actually less than you’d expect from +4 in turnover differential, as two of the Bills’ turnovers had relatively low impacts on EPA: Josh Allen’s fumble at the end of the first half (-2.0 EPA) and the controversial almost-catch-turned-interception throw to Brandin Cooks in overtime (-1.2 EPA).

I’m ambivalent about whether the refs, and their abbreviated review process, made the right call, but the most compelling rationale I saw for the interception is that the play likely would have been ruled incomplete if cooks lost the ball while rolling over, meaning Cooks didn’t have complete control and it was still live when Ja’Quan McMillian ripped it away.

Read more

In this post, I’m going to walk through how my team-strength model views every team offensively and defensively for their 2025 performance going into the Divisional Round, plus details on which teams really belong in the playoffs and exactly how strong they are versus each other (points-based model).

TEAM OFFENSIVE AND DEFENSIVE STRENGTH

*Downloadable power ratings/rankings available in the paid subscriber Google sheet.

The eight teams remaining in the divisional round all fall into the top two tiers demarcated above. The Los Angeles Rams are still the best performing team through the 2025 season and now into the playoffs. The gap between them and the crowd continues to shrink, with their Wild Card win over the Carolina Panthers registering as a close contest by the official scoreboard and by my adjusted scores. That narrow win dropped the Rams’ points-based team strength number by around half-a-point, the details you can find on the Unexpected Points paid subscriber Google sheet.

The Rams must have been thrilled to see the San Francisco 49ers beat the Philadelphia Eagles, making their Divisional round matchup a trip to the Chicago Bears, not playing at the Seattle Seahawks. The Bears have been moving up by my numbers, but starting from a lower baseline as they won a lot of games during the regular season that were very close and my numbers saw them as lucky to win.

Their miraculous comeback against the Green Bay Packers falls into that category, with my adjusted scores showing the Packers as fundamentally four points better. My team strength number for the Bears went slightly up, as the adjusted performance against the Packers was better than expected. My numbers would make the Rams a 4.5-5 point favorite in Chicago, a bit wider than the market currently at 3.5.

The New England Patriots basically held steady week-over-week, with their defensively dominant win over the Los Angeles Chargers. While the score differential was an impressive 13 points, my model would prefer a better offensive performance, and my ranking on the Chargers was fairly low going into the game, meaning a home win against them isn’t viewed as a strong accomplishment.

Read more

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2025-2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

PIT vs HOU (-3)

It was a close game early that eventually widened into a blowout down the stretch. The theme for the first two-and-a-half quarters of the game was the Pittsburgh Steelers’ inability to convert third downs (ended the game 2-for-13, -13.9 EPA) offset by C.J. Stroud’s generous nature (two fumble losses and an interception, -14.6 EPA). That left the score 7-6 with the Houston Texans having a slight lead. Then all hell broke loose for the Steelers in the fourth quarter, with two of their offensive drives ending in Texans defensive touchdowns, and the Texans offense finally kicking into gear with 10 points on their final two possessions.

Subscribe now

Outside of third down conversion issues, the Steelers offense wasn’t horrible in the first half, but they couldn’t translate generous field positions into points, and the total breakdown in the second half left them with horrendous metrics, including a 29% success rate and below 1st percentile EPA efficiency. Despite all the losses to turnovers, the Texans offense was able to post slightly positive EPA, and their rushing attack ground down the Steelers defense in the second half, with their running backs Woody Marks and Nick Chubb finishing with 160 yards on 29 carries (+5.1 EPA).

The Texans offense was able to offset big losses with explosive plays, with Christian Kirk the star. Kirk registered huge plays on third downs, including a 46-yard reception by Christian Kirk with 15 yards to go (+4.8 EPA), and two more big gains of 36 and 33 yards (+3.6 and +2.8 EPA). The Texans generated +13.8 EPA on Kirk’s nine targets, while their 23 targets to other receivers totaled -1.9 EPA.

It was a disaster game for Aaron Rodgers, in what you’d be reasonable to assume was the last game of his career. I’m sure the sentiment on Rodgers will now totally underplay just how awesome he was throughout his career, as we often see at the end, before a rebound in perception a few years later. By my numbers for quarterback value, Rodgers ranks sixth all-time (1947 to present, including AFL and AAFC) among quarterbacks.

Subscribe now

There were two Stroud’s on Monday night, one horribly inadequate protecting the ball, the other willing to stretch the ball down the field and make big plays on high-leverage downs. They evened out to a perfectly flat EPA performance, which is relatively poor considering that quarterback dropbacks average slightly positive EPA.

The Texans travel to face the New England Patriots, currently 3-point underdogs on the road.

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2025-2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

JAX (-1.5) vs BUF

The Buffalo Bills win an extremely close game by the final and adjusted scores, with the evidence pointing to the Jacksonville Jaguars, perhaps, being the slightly better team. The main driver of that assessment is the superior offensive success rate of the Jaguars, though the Bills were able to translate their inferior number into higher EPA efficiency. The Jaguars averaged more yards per play, (6.6 to 5.5), but the Bills generated more points through key short-yardage conversions.

The Bills had a modest turnover advantage (+1, +3.1 EPA) and were significantly better on higher variance late-down conversions. As is often the case in close games nowadays, fourth down success was key, with the Bills converting both of their tries (+4.4 EPA), both by Josh Allen plowing forward on 4th & 1, the second being a dramatic 10-yard gain setting them up at the Jaguars’ 1 yard-line for the deciding touchdown.

Read more

The adjusted scores quantify team play quality, with emphasis on stable metrics (success rate) and downplaying higher variance events (turnovers, special team, penalties, fumble luck, etc). Adjusted expected points added (EPA), in conjunction with opportunity-based metrics like total plays and drives, projects adjusted points. Adjusted scores have been tested against actual scores and offer slightly better predictive ability, though their primary benefit is explanatory.

All 2025-2022 and historical Adjusted Scores and other site metrics are available in a downloadable format to paid subscribers via Google Sheet.

Find previous advanced reviews here

** Adjusted Scores table:

• “Pass” - Pass rate over expectation (based on context of each play and historical averages

• “Success” - Success rate on offense, a key metric in adjusted score vs actual

• “H & A” - Home or away team

CAR vs LA (-10)

The Los Angeles Rams escape against the Carolina Panthers, winning on a come-from-behind touchdown late in the fourth quarter. The Rams were 10-point favorites going into the game and led 14-0 midway in the second quarter, giving them an estimated 95% win probability only 18 minutes into the game. However, the Panthers marched all the way back to take a 31-27 lead with 2:30 remaining, before the Rams got the final winning score.

Interestingly, the Rams weren’t unlucky to have such a close encounter by the advanced stats, unlike the fluky loss they suffered to the Panthers in Week 13 of the regular season. In that game, the Rams were massively more successful play-by-play on offense, but the Panthers were able to win on the back of a +3 turnover differential and +13.2 EPA converting all three of their fourth down chances.

The point totals for both teams were supported by extreme red-zone success, as the Panthers turned all four of their trips into touchdowns, and the Rams going 4-for-5. The Rams averaged 5.7 yards per play to the Panthers at 5.2, and outperformed in EPA efficiency in both the passing and rushing phases.

Read more