Rex's Rankings · Results
Week 3 Fantasy Football Rankings Results — How Rex's Rankings Did
27 players graded · PPR scoring · Boards published
Week 3 is scored: 27 ranked players graded against what they actually did in PPR scoring, across 8 boards. The QB board tracked the results most closely (rank correlation 0.93, average miss 26.5 spots); the RB board tracked them least closely (0.21, 39.8 spots). Across QB, RB, WR and TE, Rex's board was closer than the season points per game baseline on 0.0% of 22 contested players.
Board accuracy
| Board | Graded | Rank correlation | Mean rank error | Points error | Start-call hit rate |
|---|---|---|---|---|---|
| QB | 4 | 0.93 | 26.5of 4 | 5.05 | — |
| RB | 5 | 0.21 | 39.8of 5 | 9.26 | 100% vs 480% by chance1 start call, top 24 |
| WR | 9 | 0.80 | 78.6of 9 | 5.78 | 100% vs 400% by chance2 start calls, top 36 |
| TE | 5 | 0.37 | 36.8of 5 | 3.20 | — |
| FLEX | 19 | 0.55 | 156.1of 19 | 6.02 | 0% vs 50% by chance2 contested pairs |
| Superflex | 23 | 0.57 | 174.3of 23 | 5.85 | 0% vs 50% by chance2 contested pairs |
| Defense | 2 | — | 16.5of 2 | 6.12 | 100% vs 600% by chance1 start call, top 12 |
| Kicker | 2 | — | 17.5of 2 | 4.50 | 100% vs 600% by chance1 start call, top 12 |
Start-call hit rates assume a 12-team league starting 1 QB, 2 RB, 3 WR, 1 TE, 1 DST and 1 K. Rank correlation compares the order the board set against the order players actually finished in. Mean rank error is the average number of spots each player finished away from his ranking; it scales with board size — compare it within a board across weeks, not between boards. Rank correlation is the scale-free measure. FLEX and superflex have no fixed lineup slot, so no start-call hit rate applies. Flex-call rate instead asks how often the board put the better player on top when two closely ranked players finished far apart. Chance is what picking that many names at random would hit. A rate close to chance means the board showed no edge at that position this week, not that it failed.
Against the obvious alternatives
Chance is a low bar. The real question is whether this beats what you would do yourself — so every week, the same players are also ranked by season points per game and by last week's points, and scored the same way.
| Position | Lineup points a week | Rank error | Closer on disagreements | Weeks scored |
|---|---|---|---|---|
| QB | 0.0 a week18.5 vs 18.5 by season average | 26.51.0 by season average | 0.0%across 4 players ranked 10+ places apart | 1 |
| vs last week's points | 0.0 a week18.5 vs 18.5 by last week's points | 26.51.0 by last week's points | 0.0%across 4 players ranked 10+ places apart | 1 |
| RB | 0.0 a week39.1 vs 39.1 by season average | 39.80.8 by season average | 0.0%across 5 players ranked 10+ places apart | 1 |
| vs last week's points | −7.2 a week39.1 vs 46.3 by last week's points | 39.80.4 by last week's points | 0.0%across 5 players ranked 10+ places apart | 1 |
| WR | 0.0 a week72.0 vs 72.0 by season average | 78.61.8 by season average | 0.0%across 8 players ranked 10+ places apart | 1 |
| vs last week's points | 0.0 a week72.0 vs 72.0 by last week's points | 78.61.6 by last week's points | 0.0%across 8 players ranked 10+ places apart | 1 |
| TE | +3.0 a week6.6 vs 3.6 by season average | 36.80.8 by season average | 0.0%across 5 players ranked 10+ places apart | 1 |
| vs last week's points | −2.5 a week6.6 vs 9.1 by last week's points | 36.80.8 by last week's points | 0.0%across 5 players ranked 10+ places apart | 1 |
| Kickers and defenses — close to noise for every method | ||||
| DST | 0.0 a week-2.0 vs -2.0 by season average | 16.51.0 by season average | 0.0%across 1 players ranked 10+ places apart | 1 |
| vs last week's points | 0.0 a week-2.0 vs -2.0 by last week's points | 16.51.0 by last week's points | 0.0%across 1 players ranked 10+ places apart | 1 |
| K | 0.0 a week2.0 vs 2.0 by season average | 17.51.0 by season average | 0.0%across 1 players ranked 10+ places apart | 1 |
| vs last week's points | 0.0 a week2.0 vs 2.0 by last week's points | 17.51.0 by last week's points | 0.0%across 1 players ranked 10+ places apart | 1 |
Lineup points are what a manager would have scored starting the top of each board — 1 QB, 2 RB, 3 WR, 1 TE, 1 DST, 1 K. Closer on disagreements counts only players where the two methods differ by ten or more places; agreement proves nothing about either.
What we got right
Smallest rank error among players ranked inside the top 24 of their board.
| Player | Team | Ranked | Finished | Points |
|---|---|---|---|---|
| Green Bay Defense | GB | DST6 | DST2 | -2.0 |
| Christian Watson10 touches · 80% of snaps | GB | WR6 | WR2 | 22.6 |
| Daniel Carlson | NO | K10 | K2 | 2.0 |
| Terrance Ferguson8 touches · 76% of snaps | LA | TE17 | TE2 | 6.6 |
| Deshaun Watson1 touch · 100% of snaps | CLE | QB18 | QB1 | 18.5 |
| Omarion Hampton31 touches · 72% of snaps | LAC | RB22 | RB1 | 35.3 |
Surprises
Ranked low and delivered anyway — an error in the other direction, leaving points on the bench.
| Player | Team | Ranked | Finished | Points |
|---|---|---|---|---|
| Omarion Hampton31 touches · 72% of snaps | LAC | RB22 | RB1 | 35.3 |
| Deshaun Watson1 touch · 100% of snaps | CLE | QB18 | QB1 | 18.5 |
| Terrance Ferguson8 touches · 76% of snaps | LA | TE17 | TE2 | 6.6 |
| Daniel Carlson | NO | K10 | K2 | 2.0 |
| Green Bay Defense | GB | DST6 | DST2 | -2.0 |
| Christian Watson10 touches · 80% of snaps | GB | WR6 | WR2 | 22.6 |
Every graded Week 3 board
Underlying NFL data via nflverse. Betting lines from Lee Sharpe's games file. Rankings are FantasyWeek.us originals.