Most soccer player ratings use a scale from 1 to 10, but the scale is not used evenly, and that unevenness confuses newcomers. In the traditional newspaper system, 6 is the default, average performance: a player who did his job without standing out, positively or negatively, gets a 6. A 5 is below average, a 7 is good, an 8 is very good and usually marks a player-of-the-match contender, a 9 is fantastic and rare, and a 10 is exceptional, the kind of performance that happens a handful of times in a season across all of football. Scores below 5 are reserved for genuinely poor performances, and a 1 or 2 means something went badly wrong, such as a costly error, a sending off or a performance that actively harmed the team.

The reason 6, not 5, is the midpoint is historical: on a 1-to-10 scale there are five integers below 6 and four above it, so 6 sits slightly above the mathematical middle, and the tradition stuck. Some systems use half points, and data providers such as Sofascore give ratings to one decimal place, which creates finer distinctions but the same basic shape: most players cluster between 6 and 7, good performances reach 7.5 or 8, and anything above 9 is extraordinary. If you see a set of ratings where every player got a 7 or 8, the rater is being generous; if every player got a 5 or 6, the rater is being harsh.

There are two fundamentally different kinds of ratings, and they often disagree. Journalist ratings are subjective: a writer watches the match, notes important actions, and assigns a score based on overall impression, with goals and assists typically lifting a player by a full point and major errors costing a similar amount. The writer's method is usually simple: start from 6, add for positive contributions and subtract for negative ones, then adjust for the result, because a player in a winning team is more likely to gain a point than a player in a losing team is to lose one. That subjectivity is both the strength and the weakness of journalist ratings: they capture things data misses, such as positioning, leadership and off-the-ball movement, but they are vulnerable to bias, reputation and the temptation to reward famous players.

Data ratings, such as those from Sofascore, are algorithmic: hundreds of data points are collected in real time and after the match, grouped into categories such as shooting, passing, dribbling, defending and goalkeeping, and combined by a machine-learning model into a single number. Every action is weighted by its context and impact, so a progressive pass that creates a chance counts more than a sideways pass, and a shot from a good position counts more than a speculative effort. Data ratings are consistent across players and matches, which makes them better for comparisons, but they miss context the model cannot see, such as a player's tactical discipline in holding a shape or the psychological effect of a captain's performance. The smart reader uses both, and treats disagreement between them as interesting rather than as proof that one is wrong.

Start by reading the whole team, not just the star. A set of ratings where the whole defence got 5s and the goalkeeper got an 8 tells a story of a team under siege, saved by its keeper. A set where the midfield all got 7s and the forwards got 5s tells a story of control without cutting edge. Look for the outliers first, then ask what they did to earn them: a defender with an 8 probably made crucial tackles or scored, a midfielder with a 5 probably lost possession repeatedly or failed to track runners. Then check the substitutes: a substitute who played 15 minutes and got a 7 probably changed the game, while a substitute who got a 5 probably did not, though short cameos are inherently hard to rate fairly and should be read with a pinch of salt.

Ratings cannot capture everything, and the most important gaps are worth naming. They struggle with off-the-ball work: the forward whose runs drag defenders away without touching the ball, the midfielder who covers space so well that the opponent never attacks through it, the defender whose positioning prevents a chance from ever being created. They also struggle with tactical context: a player told to sit deep and protect a lead will look less involved than a player told to attack freely, even if both executed their instructions perfectly. And they are always, to some degree, a judgement about a single match, which is a small sample: one bad rating does not make a bad player, and one great rating does not make a star.

Use ratings as a starting point for discussion, not as a verdict. If a rating surprises you, rewatch the player's key moments and check the data behind it: the passes, tackles, shots and duels that produced the number. If the rating and the data agree, you have learned something. If they disagree, you have learned something else: either the rater saw something the data missed, or the data measured something the rater undervalued. Either way, you understand the match better than you did from the rating alone, which is the whole point of reading ratings in the first place.