THE SHORT ANSWER
The Seattle Seahawks hold No. 1 heading into Week 4 at 1694 ELO, even after a 33–31 upset loss in Washington. Buffalo (1664), San Francisco (1618) and Minnesota (1614) follow, and all three are 3–0. Full table below.
Every week, Spreadspoke ranks every NFL teams by ELO rating, a single number that rises with wins and falls with losses. How far it moves depends on the margin of victory and on how strong the opponent was. The league average is 1500.
NFL ELO power rankings (all 32 teams)
Ratings are before this week's games. Move is the change in rank from last week's rankings.
| Rank | Move | Team | Record | ELO | Week ± | Season | Last Game |
|---|---|---|---|---|---|---|---|
| 1 | – | Seattle Seahawks | 2–1 | 1694 | −19 | +2 | L 31–33 @ WAS |
| 2 | – | Buffalo Bills | 3–0 | 1664 | +6 | ▲ +41 | W 24–16 vs LAC |
| 3 | – | San Francisco 49ers | 3–0 | 1618 | +6 | ▲ +53 | W 36–30 vs ARI |
| 4 | ▲ 1 | Minnesota Vikings | 3–0 | 1614 | +14 | ▲ +57 | W 23–16 @ TB |
| 5 | ▲ 2 | Denver Broncos | 2–1 | 1610 | +13 | −8 | W 30–26 vs LAR |
| 6 | ▲ 3 | Jacksonville Jaguars | 2–1 | 1594 | ▲ +32 | ▲ +28 | W 35–6 vs NE |
| 7 | ▼ 1 | Los Angeles Rams | 1–2 | 1585 | −13 | ▼ −37 | L 26–30 @ DEN |
| 8 | ▼ 5 | New England Patriots | 1–2 | 1580 | ▼ −32 | −22 | L 6–35 @ JAX |
| 9 | ▲ 1 | Detroit Lions | 2–1 | 1563 | +7 | −4 | W 31–24 vs NYJ |
| 10 | ▲ 5 | Chicago Bears | 2–1 | 1562 | ▲ +29 | ▲ +33 | W 27–7 vs PHI |
| 11 | ▲ 2 | Kansas City Chiefs | 3–0 | 1558 | +22 | ▲ +66 | W 24–10 @ MIA |
| 12 | – | Baltimore Ravens | 2–1 | 1553 | +10 | 0 | W 34–31 @ DAL |
| 13 | ▼ 5 | Philadelphia Eagles | 2–1 | 1552 | ▼ −29 | −17 | L 7–27 @ CHI |
| 14 | ▼ 3 | Houston Texans | 0–3 | 1542 | −13 | ▼ −77 | L 17–19 @ IND |
| 15 | ▼ 2 | Cincinnati Bengals | 2–1 | 1524 | −12 | ▲ +46 | L 27–30 @ PIT |
| 16 | ▲ 1 | Pittsburgh Steelers | 2–1 | 1519 | +12 | +9 | W 30–27 vs CIN |
| 17 | ▼ 1 | Green Bay Packers | 1–2 | 1472 | ▼ −51 | ▼ −61 | L 14–35 vs ATL |
| 18 | ▲ 12 | Atlanta Falcons | 1–2 | 1450 | ▲ +51 | −13 | W 35–14 @ GB |
| 19 | ▲ 4 | Washington Commanders | 1–2 | 1446 | +19 | −11 | W 33–31 vs SEA |
| 20 | ▼ 1 | Dallas Cowboys | 1–2 | 1445 | −10 | −5 | L 31–34 vs BAL |
| 21 | ▲ 3 | Indianapolis Colts | 1–2 | 1436 | +13 | −21 | W 19–17 vs HOU |
| 22 | ▼ 2 | Los Angeles Chargers | 0–3 | 1436 | −6 | ▼ −92 | L 16–24 @ BUF |
| 23 | ▲ 6 | Las Vegas Raiders | 3–0 | 1434 | ▲ +31 | ▲ +105 | W 35–27 @ NO |
| 24 | ▲ 1 | Cleveland Browns | 2–1 | 1431 | +12 | +22 | W 21–18 vs CAR |
| 25 | ▼ 3 | Carolina Panthers | 1–2 | 1425 | −12 | +12 | L 18–21 @ CLE |
| 26 | – | New York Giants | 2–1 | 1425 | +8 | +16 | W 12–7 vs TEN |
| 27 | ▼ 9 | New Orleans Saints | 1–2 | 1425 | ▼ −31 | +3 | L 27–35 vs LVR |
| 28 | ▼ 7 | Tampa Bay Buccaneers | 0–3 | 1424 | −14 | ▼ −52 | L 16–23 vs MIN |
| 29 | ▼ 1 | Arizona Cardinals | 1–2 | 1399 | −6 | ▲ +25 | L 30–36 @ SF |
| 30 | ▼ 3 | Miami Dolphins | 0–3 | 1386 | −22 | ▼ −65 | L 10–24 vs KC |
| 31 | – | New York Jets | 1–2 | 1352 | −7 | +14 | L 24–31 @ DET |
| 32 | – | Tennessee Titans | 0–3 | 1283 | −8 | ▼ −46 | L 7–12 @ NYG |
Tied ratings are ordered by unrounded ELO.
Who has moved most since Week 1
ELO points gained or lost since the start of the season.
Sample code: build the rankings yourself
The rankings come from spreadspoke_enhanced.csv, which carries pre-game and post-game ELO for both teams in every game.
import pandas as pd
# Spreadspoke enhanced game file: one row per game, 1966 to present
games = pd.read_csv("spreadspoke_enhanced.csv", low_memory=False)
# Week 3 of the 2026 regular season
wk3 = games[
(games["schedule_season"] == 2026)
& (games["schedule_week_str"].astype(str) == "3")
]
# Stack home and away sides into one row per team
home = wk3[["home_team_id", "elo_home_pre", "elo_home_post"]]
home.columns = ["team", "elo_pre", "elo_post"]
away = wk3[["away_team_id", "elo_away_pre", "elo_away_post"]]
away.columns = ["team", "elo_pre", "elo_post"]
teams = pd.concat([home, away], ignore_index=True)
# Rank 1 to 32 on post-Week 3 ELO
teams["wk3_change"] = teams["elo_post"] - teams["elo_pre"]
teams = teams.sort_values("elo_post", ascending=False).reset_index(drop=True)
teams.index += 1
print(teams.round(0))
The ELO update behind each post-game number uses K = 20, a 65-point home-field edge (none at neutral sites) and a margin-of-victory multiplier. Here it is as a function you can run on any game:
import math
def elo_update(home_elo, away_elo, home_pts, away_pts, neutral=False, k=20, hfa=65):
diff = home_elo - away_elo + (0 if neutral else hfa)
p_home = 1 / (1 + 10 ** (-diff / 400)) # home win probability
result = 1 if home_pts > away_pts else 0 if home_pts < away_pts else 0.5
winner_diff = diff if result == 1 else -diff
mov_mult = math.log(abs(home_pts - away_pts) + 1) * 2.2 / (winner_diff * 0.001 + 2.2)
shift = k * mov_mult * (result - p_home)
return home_elo + shift, away_elo - shift
# Week 3: Atlanta won 35-14 at Green Bay
print(elo_update(1523, 1399, 14, 35)) # (1472.4, 1449.6)
GET THE DATA
The file behind these rankings, is included in the Spreadspoke ANALYST package. It covers every NFL game since 1966 with scores, closing lines, weather, and pre- and post-game ELO, so you can run the code above and build your own power rankings or test ELO against the spread.
See ANALYST pricing →Frequently Asked Questions
What is a good ELO rating?
ELO centers around 1500 for an average team. Strong playoff-caliber teams sit above 1600; rebuilding teams fall below 1400. The gap between two teams maps to an expected point spread.
Methodology
All figures are computed from the Spreadspoke Analyst data package: spreadspoke_enhanced.csv (results with pre- and post-game ELO: elo_home_pre, elo_home_post, etc.) and nfl_elo_handicapping_master.csv (ELO, closing spreads and model edge per game). Coverage is all 48 regular-season games from Weeks 1–3 of 2026. The model uses K = 20, a 65-point home-field advantage (none at neutral sites, such as Dallas vs. Baltimore in Week 3) and a margin-of-victory multiplier. League average is 1500. This page is published weekly while games are played.
Explore: Power Rankings Tracker · ATS Standings · Over/Under Records · All 32 Teams