Settings

Language, shortcuts & timezone

Language

Timezone

Time zone changes apply on this device. A new device may start at GMT+8; check match times before reading.

Skip to content
Referrals

Settings

Customize your preferences

Field Tilt in Football: Territory Is Not Chance Quality

By Opta Tips Team Updated

Read field tilt through event definitions, raw counts, game state and chance quality, and see why territorial share is different from possession.

Generated illustration of red-shirted players circulating the ball near a blue defence and goal, with the OptaTips logo at lower right.

Field tilt in football measures a team’s share of a specified kind of attacking-territory activity. It can help you see which side established play near the opposition goal. It does not, on its own, tell you who created the better chances, controlled the whole match or deserved to win.

The useful question is what happened with that territory. Repeated passes outside a compact defence can produce a large territorial share without a clear shot. A team with less time in the attacking third can still create dangerous transitions. Both patterns can exist in the same match.

Before interpreting a percentage, check the provider’s definition, the underlying event counts and the match state. Without those details, two charts labelled “field tilt” may be describing different things.

Start with the events behind the percentage

Field tilt is not one universally standardised event definition. A presentation might use final-third touches; another might use completed passes in the final third. Those measures overlap in purpose, but their counts are not interchangeable.

An Atlanta United analysis of territorial dominance discusses the share of attacking-third touches or passes. Analyst Scott Willis’s discussion of field tilt uses final-third touches and considers why volume and game context matter. Read the definition attached to the chart you actually have.

For the worked examples below, we will choose one explicit convention: each team’s completed passes in its own attacking third, divided by the combined count for both teams. These are hypothetical examples, not match data or an OptaTips product definition.

Under that convention:

A’s field tilt = A’s attacking-third completed passes ÷ both teams’ attacking-third completed passes × 100.

If A records 80 and B records 20, A has an 80% share. The denominator is 100 qualifying passes. It is not 100 minutes, all passes on the pitch, or all occasions on which either team possessed the ball.

“Attacking third” also follows each team’s direction of play. You are comparing A’s activity near B’s goal with B’s activity near A’s goal. Counting both teams’ passes in one fixed physical end would answer a different question, particularly after half-time.

Why 80% field tilt is not 80% possession

Possession and territorial activity concern different parts of a match. A side can circulate the ball deep in its own half without adding anything to a final-third count. Another can play more directly and reach attacking territory with fewer intervening passes.

Consider an invented match in which A has 55% overall possession and an 80% share of the chosen final-third events. There is no contradiction. A’s share becomes larger when the sample is restricted to activity near the opposition goal.

Nor does that 80% mean the ball spent 80% of the match in A’s attacking third. Event counts do not measure elapsed time directly. Ten quick passes and one long spell without a qualifying pass are counted differently even if they occupy similar time.

This distinction matters when a match report calls a side dominant. The claim might be about possession, territory, chances or results. Ask which one the evidence supports. A percentage from one category cannot silently substitute for the others.

It certainly is not an 80% win probability. A descriptive territorial share does not become a forecast simply because it is expressed as a percentage.

Check volume before praising the share

Eight qualifying passes against two produces the same 80% as 80 against 20. The share is identical; the amount of observed attacking-third activity is not.

Hypothetical periodA’s countB’s countA’s share
Low-volume period8280%
Higher-volume period802080%

A few events can move the first percentage substantially. Adding two passes for B changes it to 8 out of 12, or 66.7%. Adding two for B in the second changes it to 80 out of 102, or 78.4%. The same addition has a different effect because the denominator differs.

Report both counts where possible. If neither side records a qualifying event, the formula divides by zero and the share is undefined. A display may use a placeholder, but an analyst should not turn that placeholder into evidence of balanced territorial control.

Short live windows need particular care. A quiet five-minute period and a sustained attacking spell can show similar shares with very different amounts of information behind them. Overlapping rolling windows also reuse events, so several consecutive readings are not several independent matches.

Keep territorial pressure separate from chance quality

Imagine A finishes with an 80% territorial share but 0.60 expected goals, while B has 20% and 1.20 xG. These invented figures describe a possible contrast: A accumulated more qualifying activity near the opposing goal, while B’s shots received a greater combined model value.

The next step is to examine the events, not force one number to overrule the other. Did A pass around the outside without entering useful shooting positions? Did B create a small number of strong counterattacking chances? Did a penalty account for much of either total?

The guide to expected goals explains what shot-based estimates measure. Territorial events can describe attacks that never produced shots; xG usually cannot assign shot value to an attempt that did not occur. Conversely, counting passes does not directly evaluate the quality of the shot at the end.

Video can clarify the relationship. Note where the ball entered the box, whether receivers faced goal and how the defence reacted. A repeated backward pass can preserve territory without disorganising the opponent. A quick attack can do the reverse with very few counted events.

Neither pattern proves a repeatable strength from one match. It gives you a more specific question to investigate across comparable opponents and circumstances.

A leading team may willingly concede territory

A whole-match share can hide a change in incentives. A side protecting a lead may defend deeper and allow passes outside its block. The trailing side may push more players forward. Their territorial shares can change even without a sudden change in player quality.

Split the match into periods before and after goals where the data permit it. Preserve the actual timestamps, then compare counts and chances within those periods. Do not treat a 30-minute spell chasing a two-goal deficit as equivalent to 30 minutes at 0–0.

Dismissals create another comparison problem. If one side plays with ten for most of the second half, its opponent’s territorial share needs that context. The percentage remains a valid description of the chosen events; it becomes less informative as a standalone comparison of the teams at full strength.

A low block is not automatically evidence of failure, just as circulation against one is not automatically evidence of success. Look for whether the defence allowed dangerous receptions and shots, rather than judging its effectiveness from the opponent’s passing share alone.

This also limits comparisons between matches. Opponents, scorelines, dismissals and venue can alter the conditions in which territorial activity occurs. A change in the percentage invites investigation; it does not identify the cause by itself.

Average percentages and pooled counts answer different questions

Suppose A has eight qualifying events to two in one match and 20 to 80 in another. Its shares are 80% and 20%. The simple average is 50%, giving each match equal weight.

Pooling the raw counts gives A 28 of 110 events, or about 25.5%. This gives more influence to the match with more qualifying activity. The arithmetic differs because the weighting differs, not because one calculation is necessarily dishonest.

Label which approach you use. “Mean match share” and “share of pooled events” describe the distinction more clearly than an unexplained season percentage. Keep the event definition and time coverage consistent before combining either.

Do not merge a touch-based series with a pass-based series just because both have the same chart title. Provider changes or altered pitch-zone definitions can break comparability. If the earlier definition is unavailable, acknowledge the gap rather than inventing a conversion.

The same care applies to per-90 figures. Counts per unit of playing time can add information about activity, but they do not replace the opponent’s count needed for a territorial share. Show the denominator relevant to the claim.

Build a review another reader can reproduce

Record the provider, definition, period and underlying counts first. Add the score and player numbers during that period. Then examine shots and chance quality, keeping any penalty adjustment explicit.

For the invented low-volume example, a useful sentence would be: “A completed eight of the ten combined attacking-third passes in this period; that 80% share describes a small sample and does not establish which side created the best chance.” It preserves the observation without inflating its meaning.

When looking beyond one game, the home-and-away comparison guide helps identify another source of uneven comparisons. If you want to use territorial measures in a forecasting model, follow the backtest checks and evaluate genuinely later matches. A plausible explanation of the past is not an out-of-sample result.

Field tilt earns its place when it narrows the question: where did a side establish play, and what did it achieve there? The answer becomes more useful when counts, chances and match context remain visible beside the percentage.

For adults aged 18 and over. All numerical examples are hypothetical and do not represent OptaTips performance records. References explain analytical concepts, not a data-supply relationship with OptaTips.

Reader discussion

Comments 0

No comments yet.

Leave a comment

Your email address will not be published. Required fields are marked *