The Football Analysis Hub for Metric Definitions and Limitations

Most football analysis metrics are published to sell something. A broadcast package, a player’s transfer value, a narrative that keeps you watching past the eighty-fifth minute. Ask two things of any statistic before you trust it. Who benefits when you believe it? And what can the number not prove?

Every metric in this section is defined, sourced to a named data supplier, and paired with a statement of its ceiling. No invented examples dressed as case studies. No “deserved to win” arguments built on a single xG delta from one match. The guides linked from this hub answer specific questions: “Was my team actually the better side?” “Why did a high press fail in the second half?” “How much of that transfer fee was guaranteed money?” You will not find match reports, transfer gossip, or fantasy projections. You will find a framework for reading a football match as a system rather than a story.

If you are trying to understand chance quality, possession, progressive actions, or the statistical pitfalls that make one number look like proof, start with the metric that matches your question. Each guide routes you by what it answers, not by its title.

The single biggest mistake analysts make is treating the number as an end. A metric is a reduction. It takes 90 minutes of 22 bodies, a ball, and a set of laws and collapses it into one column. The value of the reduction depends entirely on whether it answers the question you actually asked.

This hub sorts the catalogue by what a reader is trying to do. Need to know whether a team created good chances or just took a lot of shots? Reach for shot quality models. Need to measure territorial control without being fooled by sterile sideways passing? Use field tilt and progressive actions. Need to understand why a defensive shape broke down? Look at pressing intensity and rest-defence structure. The guides that follow each own a single question and give it the depth it demands.

Every metric in this section is tagged with its data supplier. Where the supplier is Opta, the number tracks what a trained operator logged from a video feed. Where it is StatsBomb or Wyscout, the methodology may differ enough that two “expected goals” figures for the same shot do not match. That is not error. It is model variation, and it is a feature of football analysis, not a bug. A guide that does not name its supplier and explain the gap is not a guide.

Football match analysis begins with a specific question about 90 minutes. “Who was the better side?” is not specific enough. “Did the low block work because the defensive shape held or because the opponent’s shot selection was terrible?” That question can be answered. The guides in this section each take one question and walk you through the metric that answers it, the metric that will mislead you if you use it instead, and the sample size you need before the number means anything.

Possession Percentage and What It Hides

Possession percentage is the most misused number in football match analysis. A side can hold 65 percent of the ball and lose 2-0 by design. The opponent ceded the ball in areas that do not hurt them, sprung transitions into the space left behind, and scored from two high-quality chances. The possession number captured none of that. The guide on possession explains what it actually measures: per cent of total match passes completed, from tracking data logged by Opta or an equivalent supplier. It does not measure territorial advantage, threat, or control. A team with 35 percent possession and a coherent defensive shape can win consistently. The metric alone proves nothing.

Shot Quality Models and Expected Goals

Shot quality models, led by expected goals, fix one part of the problem. xG weights every shot by the probability that a shot from that location, under those conditions, results in a goal. The model is built on tens of thousands of historical shots. A shot worth 0.02 xG is a prayer from 35 metres. A shot worth 0.76 xG is a tap-in from six yards. Sum the xG across a match and you have a measure of chance creation quality that a raw shot count cannot give you. What xG cannot do is tell you who deserved to win. A single-match xG delta is a description of chances, not a verdict. Over a season, xG correlates with points. In one game, a 2.1 to 0.8 xG advantage and a 1-0 defeat is normal variance.

Game State Changes Everything

The guide on game state explains why all of this changes when the score is not 0-0. A team leading 2-0 in the 70th minute stops pressing high, drops into a mid-block, and concedes shots from low-value locations. The xG tally for the final 20 minutes will favour the losing side, but that tells you about the tactical choice to protect a lead, not about who was the better side over the whole match. A football match analysis that ignores game state is analysing a different game from the one that was played.

Every metric has a ceiling. The football data limitations that matter most are not gaps in the technology. They are the structural reasons a number stops being evidence at a certain threshold.

Sample Size and Stabilisation

Sample size is the first and most ignored. A striker who has scored four goals from 2.1 xG across five matches is finishing well. That same number across five matches is noise. Shot conversion rates do not stabilise until you have roughly 50 shots. Defensive metrics stabilise even more slowly. A team’s xG conceded across a ten-match run tells you something about the run. It does not tell you whether the defensive shape is good. You need half a season, minimum, before a defensive number is descriptive rather than anecdotal. The guide on sample size gives you the stabilisation thresholds for each metric category.

Why the Data Supplier Matters

Data supplier methodology is the second structural limit. Opta defines a “progressive pass” as one that moves the ball at least 25 percent closer to the opponent’s goal relative to the starting position. StatsBomb uses a different threshold. Wyscout uses another. The same match processed by three suppliers will yield three different progressive action counts. None of them is wrong. Each is measuring a different thing. A guide that presents “progressive passes” without naming the supplier and the definition is presenting a number whose error bars are unknown. That is worse than presenting no number at all.

Tracking Data Precision

Tracking data introduces a third category of limitation. Optical tracking systems capture the coordinates of every player and the ball at 25 frames per second. That data can measure physical output, tactical shape, and off-ball movement that event data misses entirely. It can also produce numbers that look precise and are not. Distance covered totals are sensitive to the tracking system’s calibration, the stadium’s camera configuration, and the algorithm that assigns a coordinate to each player when bodies overlap. Two systems in the same stadium can disagree by half a kilometre per player per match. The number is useful for within-system comparisons across a season. It is not useful for comparing a Premier League player’s distance covered to a Bundesliga player’s distance covered unless both numbers come from the same supplier using the same hardware.

The guides in this section each carry a specification table. Soccer analytics explained means knowing how to read the table before you read the number. The fields are ordered the way they matter, not the way they are printed.

The first field that matters is the metric’s definition. A metric without a written definition is a label, not a measure. Expected goals is a probability-weighted shot quality model. PPDA is passes allowed per defensive action. Progressive actions are carries or passes that advance the ball toward goal by a defined threshold. If the definition is absent or fuzzy, the number has no meaning. Move on.

The second field is the data source. The same metric from Opta and from StatsBomb can differ by a margin that changes the tactical conclusion. The guide names the source for every metric it uses. Where a metric is available from multiple suppliers, the guide explains the gap.

The third field is what the metric cannot prove. This is the field that most analysis omits, and it is the field that separates a useful tool from a rhetorical device. Possession percentage cannot prove control. xG cannot prove a deserved result. PPDA cannot prove a team pressed well, only that it pressed often and how quickly it closed down. The guide owns the limitation explicitly.

The fourth field is the stabilisation threshold. How many matches before the number describes the team rather than the run of fixtures? Finishing metrics need roughly 50 shots. Defensive metrics need half a season. Possession stabilises faster. A guide that does not tell you when to trust the number is giving you a tool without a manual.

The final field is the comparable attribute. Where does the number sit on its scale? Higher is not always better. A high distance covered figure often indicates tactical disorganisation. A low PPDA can mean an aggressive press or a low block that never engages. The comparable attribute tells you which direction is desirable and under what conditions.

Football performance metrics break into three families. Chance quality metrics tell you whether the shots were good. Possession and progression metrics tell you where the ball was and whether it went forward. Defensive pressure metrics tell you how the out-of-possession shape functioned. Each family answers a different question. Using a metric from the wrong family is the most common error in match analysis.

Possession and Territory

The possession family includes ball share, territorial advantage, and field tilt. Ball share is the simple percentage of passes completed. Territorial advantage measures the share of passes completed in the final third. Field tilt is the share of total final-third passes, a measure of which team spent more time camped in the attacking end. None of these measures control for game state, and all of them are inflated by teams that recycle the ball in safe areas. The guide on possession explains when each measure is useful and when it is noise.

Progression Metrics

The progression family covers progressive passes, progressive carries, and deep completions. A progressive pass moves the ball at least 25 percent closer to goal from its starting position. A progressive carry does the same with the ball at a player’s feet. Deep completions are passes that reach the final 20 metres of the pitch. These metrics measure intent to attack rather than intent to keep the ball. A team that completes few progressive passes but scores from transitions is not a bad attacking side. It is a side that attacks differently. The metric captures the style, not the quality.

Pressure and PPDA

The pressure family is built on PPDA and its variants. PPDA counts the number of passes an opponent is allowed to complete before a defensive action is taken. A low PPDA means the team closes down quickly. A high PPDA can mean a passive defensive shape or a low block that does not engage until the opponent reaches a certain zone. The number alone does not distinguish between an aggressive press that forces turnovers high and a mid-block that concedes sterile possession. The guide on pressing intensity explains how to read PPDA alongside defensive line height and turnover location to understand what the defensive shape was actually doing.

• Core section rule: Every metric is defined, sourced to a named supplier, and accompanied by a statement of what it cannot prove. No invented data.
• Data suppliers tracked: Opta, StatsBomb, Wyscout. Methodology gaps between suppliers are documented, not hidden.
• Stabilisation thresholds: Shot-based metrics need roughly 50 shots. Defensive metrics need roughly half a season. Possession stabilises within 5-8 matches. Every guide states its threshold.
• What the guides do not cover: Match reports, transfer gossip, betting odds, fantasy projections, live scores. The catalogue is a reference architecture for analysis, not a news feed.
• Model variation: xG, PPDA, and progressive action counts differ by supplier. The guides treat this as a feature to be understood, not an error to be corrected.

The moments that decide football matches are often the ones event data does not capture well. Transition, the phase when possession changes hands, lasts a few seconds. The shape of a defensive block during a sustained attack is a relationship between bodies in space. Neither is a discrete event that an operator logs with a timestamp.

Tracking data is the only source that can measure transition directly. It captures player coordinates at 25 frames per second, which means the spacing between a back four when the ball is lost, the distance between the pressing forward and the centre-back receiving the pass, and the acceleration profile of the counter-attacking runner. The guides on transition and rest-defence use tracking-derived metrics. They also state clearly when a claim is based on event data and therefore carries a structural blind spot.

Defensive shape metrics suffer from a different problem. A back four that holds a perfectly straight line 35 metres from goal looks organised on a tracking visualisation. If the two centre-backs are five metres apart and the opposition striker is standing between them, the shape is broken regardless of how neat the line appears. The metric that captures the gap, inter-player distance, is sensitive to the tracking system’s coordinate accuracy. A gap of 4.8 metres and a gap of 5.2 metres may be the same tactical reality measured by two different installations. The guide on defensive shape explains which metrics are robust to measurement error and which should be read as directional indicators only.

The catalogue is organised by what you are trying to do, not by the name of the metric. Start with your question. If you do not have a specific question, start with the guide on football data limitations. It will give you the vocabulary to form one.

If your question is about chance creation quality, follow the route through expected goals, shot quality, and game state. The xG guide defines the model and its suppliers. The shot quality guide explains post-shot xG and why a shot that is on target is worth more than one that is blocked. The game state guide explains why every metric changes when the score is not level.

If your question is about control of the match, follow the route through possession percentage, territorial advantage, and field tilt. Each guide defines the metric, names the supplier, explains the game state interaction, and states the stabilisation threshold.

If your question is about pressing and defensive organisation, follow the route through PPDA, defensive line height, and turnover location. The pressing guide explains how to read PPDA alongside the context that determines whether a low number means an effective press or a low block that never engages.

If your question is about a specific competition or a specific data source, use the index. Every guide is tagged by the question it answers. The index routes you by question, not by title.

Football analysis suits the tactics student who needs to understand pressing structures, defensive shape, and chance creation as repeatable mechanisms, not as match-specific stories. Suits the amateur analyst writing about the sport and needing a checkable reference for what expected goals actually measures, why Opta and StatsBomb figures diverge, and when a sample size is too small to support a conclusion. Suits the governance nerd who wants to know how the CIES Football Observatory calculates academy productivity or what the UEFA coefficient’s five-season rolling window actually weights. Suits the newcomer who has watched a World Cup, heard terms like “PPDA” and “field tilt” on a broadcast, and wants a definition that does not assume prior knowledge.

Skip this section if you are looking for live scores, transfer gossip, injury updates, or match reports. Use a dedicated news service. Skip it if you need betting odds, tips, or form guides. Go to a licensed betting information provider. Skip it if you need fantasy football price changes and points projections. Use a fantasy platform. Skip it if you want career advice on becoming a professional footballer or agent. Consult a sport career specialist. The catalogue is a reference architecture for understanding the sport’s analytical tools. It documents what the numbers mean, what they cannot prove, and how to use them to answer a specific question about a match or a season.