Exploring fantasy football strategy

I scored every week of the last five NFL seasons for a standard fantasy league and used it to test the advice people give. This post follows a season in order: the draft, the waiver wire, and then where the points actually come from each week, down to which NFL teams hand most of theirs to one or two players.

The setup

Everything here is one league, the most common one: 12 teams, 16 rounds, point per reception, starting 1 QB, 2 RB, 2 WR, 1 TE, 1 flex, 1 kicker and 1 defense. I scored every regular-season week from 2021 to 2025, all five of the seventeen-game seasons, from nflverse’s public stats. That is 34,810 player-weeks. Kickers get 3 points for a field goal under 40 yards, 4 for 40 to 49, 5 for 50 plus, 1 per extra point and minus 1 per miss. Defenses use the Yahoo default: sacks, takeaways, safeties, blocks and touchdowns, plus a points-allowed tier from +10 for a shutout down to −4 for 35 or more. The weekly sections also use the first seven weeks of 2026 as a check on this season.

For the draft and the waiver wire I needed to know what people expected before each season, so I used the FantasyPros PPR expert consensus rankings as they stood the week before kickoff, from DynastyProcess’s public archive. That isn’t average draft position, but it’s close, and it is the board most drafters look at.

Part 1: The draft

How points fall off at each position

Here is the five-season average at a few ranks. “Last starter” is the last player a 12-team league has to put in a lineup. With one flex spot, the flex is almost always a receiver: on average the league starts 25 running backs and 35 receivers.

Raw points make quarterbacks look like the most valuable thing in the game. They score the most, and the #12 quarterback outscores the #12 running back by 40 points. That doesn’t matter. A draft pick buys you points over what you could have had for free, and because every team starts one quarterback, the free one is good. The 13th quarterback still scores 253.

So the useful number is value over replacement: a player’s points minus the best player at the position that nobody in the league is forced to start. Measured that way, the order flips. The top receiver and the top running back are worth 13.3 and 12.3 points a week over the waiver wire. The top quarterback is worth 8.8, the top tight end 8.3, and the top kicker 2.6. Of the 24 most valuable player-seasons each year, two rounds’ worth, an average of 9.2 were receivers, 8.2 running backs, 4.2 quarterbacks and 2.4 tight ends. No kicker or defense ever made it.

The tight end cliff

Tight end falls fastest. The #1 tight end is worth 141 points over replacement. The #2 is worth 87, the #5 is worth 43, and the #10 is worth 13. The #5 receiver is still worth 122, close to the #1 tight end.

An elite tight end gives you a weekly edge at a position where everyone else starts a player barely better than free. The problem is knowing who it is. The preseason consensus ranked tight ends second worst of the six positions, with a rank correlation of 0.26 between where the top 12 were ranked and what they scored. Only 53% of the preseason top 12 finished in the top 12. In the simulation below, taking the best tight end left in round 2 cost 1.2 points a week, with a 95% interval from −3.3 to +0.9. Waiting until round 10 cost nothing measurable (−0.2).

The cliff is real. Picking the one player on the right side of it in August mostly isn’t possible.

Some positions you can predict, some you can’t

Running backs are the most predictable position, both before the season and during it. They do get hurt, but as the injury section shows, so does every other position, and a back with a big role keeps it when he’s healthy. The consensus is good at spotting who has the role.

Kickers and defenses are different. Their first half predicts their second half at 0.13 and 0.07. A defense that has been great for nine weeks is not a better bet for week ten than one that hasn’t. And across these five seasons, the preseason consensus on defenses did no better than chance.

Where each round’s value comes from

Value over replacement tells you what an elite season is worth. A pick only buys a chance at one. So I took every player the consensus ranked into each group of rounds and counted how many finished above replacement at their position.

Early on, every position hits at about the same rate. The split comes in the middle. Between rounds 5 and 9, receivers finish as starters more often than running backs (46–48% against 36–38%). Running back has the steepest and most predictable top, and then the depth runs out. Late tight ends hit half the time, mostly because the bar for a starting tight end is low.

Which position gets hurt most

I counted injuries for every player the preseason consensus ranked as a starter: the top 12 quarterbacks and tight ends, the top 25 running backs and the top 35 receivers, in each of the five seasons. A game counts as missed to injury when the player’s team played, he recorded no stats, and he was listed Out or Doubtful on that week’s injury report or was on injured reserve or PUP.

No position stands out. Drafted running backs missed 15% of their games to injury, and so did tight ends. Receivers and quarterbacks missed 14%. Season-wrecking absences of eight games or more hit 11% of running backs, 10% of receivers and 12% of quarterbacks. Counting every player-season with at least four games gives the same answer: 4.0 new injuries per 100 games for quarterbacks, 4.4 for running backs, 4.6 for tight ends and 4.9 for receivers.

The injuries differ even when the totals don’t. Running backs most often hurt an ankle (25%) or a knee (20%). Receivers most often hurt a hamstring (25%), tight ends a knee or a hamstring (26% each), and quarterbacks a shoulder (19%) or their head (13%, from only 31 injuries).

A drafted starter at any position misses about one game in seven. With seven skill-position starters, the chance that at least one of them is out in a given week is roughly two in three. You aren’t drafting around a fragile position. You’re drafting for the weeks, most of them, when somebody is missing.

600 drafts

To test strategies rather than positions, I simulated drafts. In each one, eleven teams snake-draft straight down the consensus board, taking a kicker and a defense in the last two rounds. The twelfth team follows a rule. Every rule drafted from all 12 slots in all 5 seasons: 60 drafts per rule, ten rules, 600 drafts. Then each team played the real season, starting whoever had the best points per game so far (weighted toward the preseason ranking for the first few weeks) among players who actually played that week.

A typical team scores 110 points a week, so one point a week is about one percent. A weekly win rate of 46% instead of 50% is about half a win over a 14-week regular season, often the difference between making the playoffs and missing them.

Waiting on a quarterback hurt. “Wait on QB” was the worst rule I tested, three points a week behind the league and behind in all five seasons, though only barely in 2023. My guess is the rushing quarterback: the drop from the top of the position to the #10 quarterback is now 113 points over a season. Going early didn’t win much, +0.6 a week, but going late reliably lost.

Zero RB depends on the season. It was slightly ahead in 2021 and 2022, then fell behind badly: −4.5 a week in 2024 and −7.4 in 2025, when elite running backs had big seasons. My teams make no waiver claims for skill players, and Zero RB is usually sold as a bet that you’ll find running backs on waivers, so this simulation can’t give it credit for that. Part 2 looks at how often that works.

Streaming defenses was the only rule ahead in every season, between +0.8 and +2.7 a week. The free defense picked by the Vegas line scored 7.98 points a week, against 7.12 for the average preseason top-12 defense. That’s under a point of edge, plus never losing a week to a bye.

Everything else, including Hero RB, Robust RB and all the tight end rules, lands inside the noise. Most arguments about draft strategy are about differences smaller than the season-to-season luck.

Part 2: The waiver wire

The simulation above makes no pickups, and Zero RB lives or dies on them. So I tracked every undrafted player across the same five seasons and checked which early signs turned into a real starter.

Three definitions do most of the work:

• Undrafted means outside the top 192 of the preseason consensus board, the 12 × 16 picks a league makes. Real leagues also drop players, so the real waiver pool is bigger than this.
• A hit is a pickup who, from the week after you could have added him through week 17, scored more per game than replacement level at his position and played at least half the remaining games. Replacement per game is 17.0 points for a quarterback, 11.9 for a running back or receiver, and 9.9 for a tight end.
• Volume means a starter’s workload: 15 or more carries plus targets for a running back, 20% or more of the team’s targets for a receiver, and 15% or more for a tight end. A big week is 18 or more PPR points for a back or receiver, 14 for a tight end, and 22 for a quarterback.

How much is on the wire

For each window I ranked players by points from that week on and took the starters a 12-team league needs.

Over a full season, one starter in twelve was never drafted. In the second half it is one in seven, about 12 players a year, or one per team. More than a quarter of second-half starting tight ends went undrafted, which fits the tight end section above: preseason rankings did badly there and the bar for a starter is low.

One week is noise

For every undrafted running back, receiver and tight end, I found the first week each showed a signal and checked what he did afterward. Signals are counted through week 12, so there is enough season left to measure.

One big week turned into a starter 14% of the time, and one week of volume 9%. Two weeks in a row doubles the rate or better. A third week of volume adds only four points of hit rate and costs a week of the season, so two is the useful number. Two big weeks in a row is the strongest signal, at 38%, but it is rare: 32 players in five seasons.

What waiting costs. Of the players who had one big week, 34% followed it with another big week or a week of volume. Those players hit 28% of the time. The ones who didn’t hit 7%. Waiting one game means you lose a third of the real pickups to someone else, and you skip two-thirds of the wasted adds.

A big week on a small role

None of the 19 undrafted running backs who had a big week on fewer than 15 touches went on to be a starter. A long touchdown on eight carries does not repeat. Backs with two straight games of 15 or more touches hit 24% of the time. Tight ends are the exception: a big week on a small target share still hit 23% of the time, because the bar is low and touchdowns are a large part of tight end scoring. The samples are small.

Snap share alone was weak for receivers. Two straight games on the field for 65% or more of the snaps turned into a starter 5% of the time for receivers, against 14% for running backs and 12% for tight ends. A receiver can play every snap and not get the ball. Targets are what count.

A new starting quarterback, one who throws 25 or more passes in two straight games, scored like a top-12 quarterback for the rest of the year only 10% of the time, because the replacement quarterback already scores 17 points a game.

Compared with your bench

A pickup means a drop, usually a late-round pick. So I compared both groups from week 6, the median week a two-game signal appears.

By week 6, a round 13 to 16 pick has an 11% chance of becoming a starter. A pickup with two weeks of volume is close to a round 10 to 12 pick, and one with two big weeks in a row is close to a round 7 to 9 pick. Even a one-big-week pickup beats your last few picks. This compares groups: if your round-14 pick has had volume for two weeks, he belongs in the stronger group too.

When the lead back is hurt

I found every game where a team’s lead back (the most touches in the previous game, at 12 or more) missed the game injured, took the back who played the most snaps behind him the week before, and recorded what that back scored.

The lead backs had averaged 13.3 points before they got hurt. A drafted backup roughly replaced that. An undrafted one scored 7.8, well below a replacement back. The draft market already sorted these backups: the valuable ones were drafted, and what is left on the wire is usually half of a committee.

Streaming by matchup

I ran the defense-streaming test for quarterbacks and tight ends too: each week, take the undrafted player who started the week before on the team with the highest Vegas implied total.

A good matchup doesn’t make up for a worse player at a position where the gap between players is large. At kicker and defense, where talent matters less than circumstance, the matchup is enough.

Part 3: Where the points come from

The first two parts were about players. This part is about the whole pile of points the NFL produces each week, and how it splits by position, team, game and player. The unit is a team-game: one team’s side of one game, counting every fantasy point scored by its players, its kicker and its defense. Five seasons give 2,718 of them.

The total is steady, one team is not

An average team-game produces 98.5 fantasy points, and a full 16-game slate about 3,150. The weekly average has a standard deviation of 4.4. In 90 weeks it never dropped below 87 or went above 109, and the low was week 18 of 2025, when playoff teams rest starters. Weeks 1 to 17 average 98.8. This season is at 98.6 through week 7.

What does change is how many points are on the board. A week with six teams on bye averages 2,595 points, 18% less than a full slate. Every team in your league is drawing from that smaller pool.

Single team-games swing hard. The standard deviation is 25.6 points, six times the weekly average’s, and the middle 80% runs from 67 to 131. The worst was 17, the Texans in a 40–0 loss to the Bills in 2021. The best was 211, the Dolphins in their 70–20 win over the Broncos in 2023.

By position

Receivers score a third of all fantasy points and running backs just under a quarter. The split is almost identical in weeks 1 to 6 and weeks 12 to 17, so cold weather doesn’t shift points from passers to runners league-wide. The one trend is tight ends: 12.2 to 12.4% from 2021 to 2024, 13.7% in 2025, and 14.1% so far in 2026.

What the points are made of

Touchdowns are a quarter of the points and most of the swing. They explain 43% of the variance in a team’s weekly skill points, against 39% for yards and 15% for catches. The bottom tenth of team-games averages 4.5 touchdown points and the top tenth averages 40. Yards and catches come from volume, which is steady. When a player has a huge week, check how much of it was touchdowns before you assume he’ll do it again.

By NFL team

Over five seasons the Cowboys produced 113.2 points a game and the Panthers 82.6, a 30-point gap. The best team-season was the 2021 Buccaneers at 128.2 a game. The worst was the 2025 Jets at 75.7.

Which team-season a game belongs to explains 21% of the variance in team-game totals. The opponent’s season explains 10%, so the offense matters about twice as much as the defense it faces. A team’s rank in one season correlates with its rank the next at 0.45, and its first half predicts its second half at 0.54. Seven weeks into 2026, the Panthers are at 111.9 against a five-year average of 82.6, and the Dolphins at 70.6 against 100.0. Seven games is a small sample, but a half-season correlation of 0.54 says more of that will hold than revert.

How teams share their points

That table says how much an offense produces. This section is about who on the team gets it. For each team-season I took the points scored by its running backs, receivers and tight ends and measured how much went to its top players. Quarterbacks are left out of that count, because every team starts exactly one, and get their own measure below.

On an average team, the top scorer took 23% of those points, the top two 40% and the top three 54%. The middle 80% of team-seasons fall between 33% and 47% for the top two.

The most concentrated teams pair one star running back with one star receiver. The most spread out are mostly offenses where the star was hurt or there wasn’t one.

Over five seasons, the Raiders (46%), Dolphins (45%), Bengals (44%), Eagles (44%) and Falcons (43%) concentrated their points the most. The Patriots (34%), Chiefs (35%), Broncos (36%), Texans (36%) and Commanders (37%) spread them the most. Spreading the ball doesn’t make an offense bad: the Chiefs were sixth in skill-player points per game while being the second most spread out, and the Raiders were 23rd while being the most concentrated.

By position. A team’s top non-quarterback scorer was a receiver in 50% of team-seasons, a running back in 44% and a tight end in 6%. Within each position, the lead player’s share of that position’s points:

Running back has the widest range. The 2021 Steelers gave Najee Harris 92% of their running back points; the 2025 Cardinals’ lead back, Michael Carter, had 30%. Tight end is the most top-heavy: when a team has one real tight end, like George Kittle (96% of the 49ers’ tight end points in 2023), he gets almost all of it. When it doesn’t, the points split across two or three players nobody should start.

Quarterbacks took 19% of a team’s skill-player points on average. The Bills and Eagles were highest at 25%, then the Ravens at 23%. That is Josh Allen, Jalen Hurts and Lamar Jackson running the ball.

Is it the team, or the players? If concentration were a team habit, it would carry from one season to the next, and the gap between teams’ five-season averages would be bigger than chance. To get the chance level, I shuffled each season’s values among the 32 teams 2,000 times and measured how far apart the five-season averages landed anyway.

Only the quarterback share is clearly a team trait, and that’s because it is really the quarterback. The rest is mostly the players. The top-two share has a small team effect, a bit above chance. The lead running back’s share has none at all: over five seasons, no team was a bellcow team or a committee team beyond what luck would produce. The Bengals’ 71% is Joe Mixon and then Chase Brown each getting the job, not a scheme.

Shares also shrink. Players who took 15% or more of their team’s non-quarterback points averaged 20% that season and 17% the next if they stayed on the same team. The 39 who changed teams went from 19% to 15%. Within a season, a team’s top-two share in weeks 1 to 9 predicts weeks 10 to 18 at 0.43.

Concentrated offenses score a little more, but not much and not every year. The correlation between a team’s top-two share and its non-quarterback points is 0.17, and it was slightly negative in 2023. The most concentrated quarter of team-seasons averaged about 87 skill-player points a game, the most spread out quarter about 81.

The game around it

Vegas lines predict a team’s fantasy output better than anything else I tested. Each point of implied team total was worth 3.0 fantasy points, and the line explains about 20% of the variance.

The top group averages 40 points more than the bottom, but the ranges overlap almost completely. A team implied for 28 still has a one-in-ten chance of scoring under 87.

Winners score 111 fantasy points, losers 86. Losing teams’ receivers score 30.7 against 35.3 for winners, 13% less, because a team that’s behind keeps throwing. Losing teams’ running backs and quarterbacks score 25% less, and their defenses two thirds less. Home teams average 101.4 and away teams 95.7. Domes average 101.8 and outdoor games 96.9; outdoor games with wind of 15 mph or more average 90.0. None of these are adjusted for the Vegas line, which already prices in most of them.

By player

In a typical week, about 273 skill players score at least one fantasy point. The top 60 score half of all the points, and the top 24 score a quarter. If a 12-team league could pick its 84 starters after the games, those players would score 61% of everything. The other 39% goes to backups, blocking tight ends and garbage-time receivers nobody would start.

What to do with this

In the draft

• Draft by value over replacement, not by points. The free quarterback scores 253 and the free receiver scores 172.
• Spend your first two rounds on running backs and receivers. They are 17 of the 24 most valuable seasons in a typical year, and they hit 82% of the time that early.
• Don’t wait too long on a quarterback in a one-QB league. “No quarterback before round 10” was the worst rule in this test.
• In the middle rounds, lean towards receivers. From round 5 to 9 they finish as starters about ten points more often than running backs.
• Treat an elite tight end as a bonus, not a plan. A late tight end costs you very little.
• Don’t mark a running back down for injury risk alone, or for playing on a “committee team.” Drafted backs missed the same share of games as everyone else, and no team’s backfield split carried from year to year beyond chance. Draft the player and his role this year.
• Expect last year’s dominant share to shrink. Players with 15% or more of their team’s points kept about 17% the next year.
• Take your kicker and defense last.

On the waiver wire

• Wait for the second week. One big week turns into a starter about one time in seven. Two big weeks in a row, or two straight games of volume, doubles that or better.
• Judge running backs by touches and receivers by targets, not by points or snaps.
• Watch tight ends most. More than a quarter of second-half starting tight ends were undrafted.
• Drop your last-round picks first. By week 6 they turn into starters 11% of the time, less than any waiver signal above.
• When a lead running back gets hurt, an undrafted backup is worth a bench spot, not your top claim.
• Stream your defense every week by Vegas matchup, and your kicker if you like. Hold your quarterback and tight end.

Setting a lineup

• Look at the Vegas implied total. It’s the best single predictor of a team’s fantasy output, but it explains only a fifth of the variance. Use it to break ties, not to bench a stud.
• Discount touchdown-heavy weeks. Touchdowns are a quarter of the points and nearly half of the swing.
• Don’t bench receivers on teams expected to lose. They keep 87% of what winners’ receivers score. Running backs and defenses are the ones that suffer.
• Don’t chase a hot kicker or defense, and do trust a running back’s first half.
• Take team trends seriously by midseason. An offense’s first half predicts its second half at 0.54.
• Plan for bye weeks as a smaller pool. A week with six teams off has 18% fewer points for every team in your league.

Sources

• Weekly player and team statistics, 2021–2025 and 2026 weeks 1–7: nflverse-data (stats_player_week, stats_team_week).
• Weekly snap counts, injury reports and roster status: nflverse-data (snap_counts, injuries, weekly_rosters).
• Schedules, final scores, Vegas lines, roof and weather: nflverse/nfldata games.csv.
• Preseason PPR expert consensus rankings and player ID crosswalk: DynastyProcess data (db_fpecr, db_playerids), archived from FantasyPros.