How SquadRanks Builds a SquadScore
The full methodology behind SquadRanks: every layer of the SquadScore, from club form to tournament form, with the weight and the scale of each one.
Last updated 17 Jul 2026On this page6 sections
Updated 30 September 2026.
What the SquadScore is
The SquadScore is one number per squad, on the same scale as a player match rating. It answers one question: on the evidence available today, how strong is this squad?
It is not a forecast of results, and it is not a reputation index. A full-back in great form at Leverkusen scores above a bigger name having an uneven season at a mid-table club, and a team that keeps out-creating its opponents at a tournament climbs above one living on its spring numbers.
The formula
At the 2026 World Cup the score has five layers:
SquadScore = 35% club form + 5% market value + 15% national-team form + 20% tournament pedigree + 25% tournament form.
The layer set belongs to the competition. A club has no national-team form and no tournament pedigree, so the Premier League 2026/27 board runs on two layers: 60% club form + 40% market value.
Every layer is expressed on the rating scale before the weights apply, so each weight is a real share of the final number.
Club form: 35%
The core of the score. It reads the most recent lineup the team actually fielded, the real starting XI and the real bench, never a predicted one:
Club form = 0.75 × starting XI average + 0.25 × bench average.
Each player carries one rating, built in four steps:
- Match ratings, weighted by recency and minutes. A club match counts in proportion to the minutes played, up to a full 90, and its weight halves every 60 days. An international match halves every 120 days, because national teams play less often. A World Cup or continental finals match counts in full, a Nations League match at 0.85 and a friendly at 0.6.
- League and opponent strength. Each club match rating is scaled by the quality of the league and of the opponent before it enters the average. A 7.4 in the Premier League ends up above a 7.4 in a weaker league, and the weaker the league, the harder the number is pulled toward 6.0.
- Sample size. A player needs 900 club minutes, about ten full matches, for the rating to stand at face value. Below that it is pulled toward 6.5 along a square-root curve: at a quarter of those minutes a rating keeps half of its distance from 6.5. International minutes need 300, and three caps, to count in full.
- Trend and age. The last five matches against the season average move the rating by 1.6% at most, up or down. An age multiplier follows: 1.00 up to 29, 0.98 from 30 to 32, 0.95 at 33 and 34, 0.90 from 35.
One correction covers the expensive player with few minutes. A player valued at €50 million or more, with under 900 club minutes, has the rating pulled toward the market-value score below, by up to half at zero minutes.
The two averages are not built the same way. The starting XI average is a plain mean over the starters who have a rating. The bench average is harder on thin depth: a bench player with no rating, with under 90 club minutes, or with under 270 minutes for club and country combined, counts as 6.0, and a rated bench player with a known value counts as 75% rating and 25% market-value score. A squad with fewer than 11 starters, or with more than two starters who have no rating, is moved out of the ranked list instead of being scored on thin data.
The bench carries a quarter of the layer because substitutes now decide matches. Under the five-substitution rule, teams average about 4.26 substitutions per game, with the average sub arriving near the 70th minute. At Qatar 2022, substitutes scored 19% of all goals, and per minute they sprint and run at high speed 28 to 47% more than starters. Add suspensions, injuries, extra time in knockout matches and the tactical reset the rule enables, and a team that goes deep uses most of its squad.
Market value: 5%
Each player's transfer value becomes a score on a logarithmic scale. €10 million is 6.0 and every tenfold step adds 2.0, so €1 million is 4.0 and €100 million is 8.0. The score is limited to the range 3.0 to 9.5, and the layer is the average over the squad's valued players.
It is a small prior. It separates a deep, expensively assembled squad from a thin one when the rating sample is noisy, and at 5% it cannot buy a ranking.
National-team form: 15%
FIFA ranking points, moved onto the rating scale. 1,400 points is 5.5 and every point above adds 0.004, so 1,650 points is 6.5. The scale stops at 7.0, which a team reaches at 1,775 points, and it does not go below 5.5.
The layer captures what the club season cannot see: qualifying campaigns, Nations League runs, continental titles. The ceiling is deliberate. Every team above 1,775 points sits at 7.0, so ranking points cannot separate the strongest sides. Club form and tournament form have to.
Tournament pedigree: 20%
Results at the major tournaments of the 2022 to 2026 cycle, scored by the round reached. A World Cup win is 10 points, a final 6, a semi-final 4 and a quarter-final 2. A continental championship win is 6, a final 3 and a semi-final 2. A Nations League win is 3 and a final 1.
The layer is 6.0 plus 2.5 for every 12 points, with a ceiling of 7.5. A team with no result in the cycle gets 6.0: a neutral value, not a punishment.
The ceiling is reached at 7.2 points, so it arrives early. Argentina's 2022 World Cup and 2024 Copa América are 16 points, more than twice what the ceiling needs. A second title adds nothing once a team is there.
Tournament form: 25%
The live layer. It went in on 16 July 2026, calibrated on the 102 World Cup matches played by then. For every finished match a team banks two signals:
- Chance creation and prevention. Expected goals for, minus expected goals against, adjusted for the strength of the opponent. A 1-0 over a weak side with little created moves a team less than out-creating a top seed does. A match that goes to extra time is scaled by 0.75, back to a 90-minute basis.
- Results against expectation. Points won against the points the pre-match strength gap predicted: three for a win, one for a draw, and two or one for winning or losing a shootout. A team that keeps beating its number is showing something the squad list does not.
The layer is 70% the first signal and 30% the second. Each signal is placed on the rating scale around 6.4 and limited to the range 4.8 to 8.4. Confidence grows with matches played: after one match the layer moves half the distance from 6.4 to the measured value, after three matches three quarters of it. One lopsided opening game cannot swing the board.
One guardrail applies in the knockout rounds. A team eliminated head-to-head cannot sit above the side that knocked it out when the gap separating them is under 0.05: the winner of the tie is placed just above.
The layer froze as the 2026 record when the tournament ended. Each squad's other inputs froze with it, two days after that squad's last match, so a club season that started in August does not re-rank a team that went out in the round of 16.
Data coverage
One adjustment sits on top of the layers. A squad with limited data coverage has its score multiplied by 0.97, so thin evidence reads as thin evidence and not as a strong score.
Which players count
The score uses the most recent lineup the team fielded: that starting XI and that bench. It does not project a lineup and it does not second-guess a coach.
There is one narrow exception. When an officially announced squad leaves out a player from that last lineup, the vacated place is filled from the announced squad by a fixed rule, the place is marked provisional, the team card says so, and the next real lineup replaces the adjustment.
This is why a SquadScore sometimes moves with no round of fixtures behind it: the list changed, not the ratings.
Per-position rankings: GK, DEF, MID, FWD
The GK, DEF, MID and FWD tabs rank one line at a time. A plain position average would let a mid-table squad top a tab by accident, if its weakest line were still high relative to its own XI. So each line starts from the team's overall level and moves by how that line compares with the team's own starting XI.
How far it moves depends on the minutes behind it. The line's own average gets a weight of n / (n + 1,500), where n is the minutes played by its players, each capped at 3,000. At 0.7 or more the evidence is rated high, from 0.5 medium, and below that low. A line with low evidence stays close to the team's level: poor coverage neither sinks nor inflates the ranking.
What changes on each update
During a competition the whole board is re-scored daily. Tournament form banks every newly finished match, and player ratings take in the latest club and international minutes.
Squad lists change whenever an official call-up changes. We use the latest official list, not a guess at a final roster. If a coach drops a veteran and gives a 20-year-old a debut, the board shows it at the next update.
What a SquadScore does not measure
Three things the score deliberately leaves out:
- International chemistry. The score assumes the coach can build working partnerships out of the ratings. An attack with six elite forwards who do not combine still posts a strong attack score, although a few flat performances show up in tournament form.
- Injuries in real time. A hamstring in the round of 16 shows up eventually: the player drops out of the next lineup and the rating stops accruing. There is a lag, typically one update, from the injury to the score reacting. The number on the card is a snapshot, not a medical report.
- Tactical fit. A coach moving from a back three to a back four after qualifying can unlock or wreck a defensive line without any rating moving.
That is what the per-position tabs and the surrounding cards are for: coach tier, prime window, league diversity, club concentration. The SquadScore is the base, and the tabs show where the strength sits.
Pedigree is a fifth of the score, and 2026 showed why it is not more. Argentina started at the pedigree ceiling of 7.5. Spain started at 7.25 on Euro 2024, then beat France 2-0 in the semi-final and Argentina 1-0 in the final.
