Expected Threat (xT) Explained: How Football Teams Create Danger

Expected Threat, usually shortened to xT, is a football metric that values how passes and carries move possession between areas of the pitch. A typical xT model divides the pitch into a grid and gives each zone a value based on the probability that possession there will eventually lead to a goal.

When a player moves the ball into a more dangerous zone, the action receives positive xT. If the destination is less threatening, its immediate xT value is negative. This allows analysts to measure progression before a shot occurs, but the result still depends on the model’s grid, data, definitions and treatment of turnovers.

What Does Expected Threat Mean in Football?

Expected Threat estimates the attacking potential of possession in different areas of the pitch.

The central idea is that having the ball in some locations is more valuable than having it in others. Possession beside a team’s own corner flag is unlikely to produce a goal during the same attacking sequence. Possession between the opposition midfield and defence is usually more promising. Reaching a central position inside the penalty area increases the potential further.

An xT model assigns estimated values to those locations. The value of a pass or carry is then calculated from the difference between its starting and finishing zones:

xT added = destination xT − starting xT

If a player moves the ball from a zone valued at 0.02 into one valued at 0.08, the action adds 0.06 xT under that model.

This does not mean that the player literally created six per cent of a goal. It means the model’s estimate of the possession’s future scoring potential increased by 0.06.

How an Expected Threat Grid Works

There is no single universal xT grid or definitive xT value for every pitch location. Models can use different grid sizes, datasets, possession definitions and scoring horizons. Most basic approaches nevertheless follow a similar process.

1. Divide the pitch into zones

The pitch is divided into a grid. Each cell represents a different location from which the team in possession might shoot, pass, carry the ball or lose possession.

A coarse grid is easier to calculate and interpret but can group meaningfully different positions together. A detailed grid captures more spatial variation but requires more data to estimate each zone reliably.

2. Study what happens from each zone

Using historical event data, the model estimates what teams tend to do when they have possession in each cell. This can include:

• How frequently they shoot.
• How often they pass or carry into another zone.
• Where successful ball movements finish.
• How often possession is lost.
• How likely shots from the zone are to become goals.

3. Estimate the zone’s scoring potential

If a player shoots, the model considers the probability of that shot becoming a goal. If the ball moves to another zone, it considers the future scoring potential of the new location.

The calculation is repeated across the grid until each zone has an estimated threat value. Central areas near goal usually receive the highest values. Deep and wide areas normally receive lower values.

4. Compare the starting and finishing zones

Once the grid has been estimated, successful passes and carries can be valued by subtracting the starting-zone value from the destination-zone value.

The metric therefore rewards changes in attacking potential, not simply distance travelled.

A Worked xT Passing and Carrying Example

Consider the following illustrative attacking sequence. The values demonstrate the calculation and are not taken from a real match or presented as a universal xT scale.

The complete sequence adds 0.065 xT because possession moves from an initial value of 0.010 to a final value of 0.075.

The backwards pass receives negative positional value, but that does not prove it was a poor decision. The midfielder may have been surrounded by defenders with no viable forward option. Retaining the ball might have been preferable to forcing a pass and exposing the team to a counterattack.

xT describes how the location-based threat changed. It does not fully judge the tactical quality of the decision.

How Passes Receive xT Value

A successful pass receives positive xT when its destination zone has greater scoring potential than its starting zone.

This can reward:

• Passes that break an opposition line.
• Central passes between midfield and defence.
• Passes into the penalty area.
• Switches that find space on the opposite side.
• Cutbacks into valuable central positions.

Long passes are not automatically worth more. A 40-metre pass towards a crowded touchline may add less threat than a short pass into a central pocket between several defenders.

Basic xT normally credits the player who moves the ball between the zones. It may not credit the teammate whose run created the space or the earlier player who disrupted the defensive structure.

How Carries Receive xT Value

A carry adds xT when a player retains the ball while moving it into a more threatening zone.

This can identify players who create progression by travelling with the ball rather than passing. Attacking full-backs, dribbling wingers and midfielders capable of moving through pressure may therefore generate substantial carrying xT.

Separating passing xT from carrying xT helps describe different progression styles. Two players can add similar overall threat while contributing in very different ways.

Carrying value still depends on how the provider defines a carry. Minimum distance, action boundaries and the treatment of touches can vary, so outputs from different datasets are not automatically comparable.

Expected Threat Compared With xG, xA and Possession Value

xT versus xG

Expected Goals values the quality of shots that are actually taken. Expected Threat works earlier in the attacking process and values how possession reaches dangerous territory.

A team may generate high xT through repeated entries into threatening areas but fail to produce good shots. That could indicate weak final passes, poor penalty-area movement, ineffective crosses or strong last-line defending.

Conversely, a counterattacking team may create reasonable xG from relatively little accumulated xT. Neither metric replaces the other.

xT versus xA

Expected Assists generally assigns the xG value of a resulting shot to the pass that created it. If the receiving player does not shoot, the pass usually creates no xA.

xT can value a pass into a more threatening area even when the next action is another pass, a carry, a turnover or no shot at all. It therefore covers more of the progression before the final chance.

However, xA is connected to the quality of a real shot. Basic xT values the destination zone and may not distinguish a precise cutback to an unmarked striker from a difficult pass to a tightly marked receiver in the same area.

xT versus possession value

Possession value is the broader modelling family. Expected Threat is one specific type of possession-value model, normally centred on the scoring potential of pitch locations.

Richer possession-value models can include the probability of conceding, defensive pressure, recent actions, player positions and unsuccessful events. They may value tackles, interceptions and turnovers as well as successful passes and carries.

It is therefore accurate to describe xT as a possession-value approach, but not to treat every possession-value output as xT.

Expected Threat Compared With Field Tilt

Field tilt measures which team has the greater share of advanced territorial possession. xT estimates how much particular passes and carries change the attacking potential of that possession.

A team can record high field tilt by sustaining pressure in the final third without progressing into the most dangerous zones. Another can have limited advanced possession but generate substantial xT from a small number of penetrative transitions.

Field tilt helps answer who controlled advanced territory. xT helps explain how effectively possession was moved into more threatening positions.

What Expected Threat Reveals About Players

xT can identify players who consistently move possession towards more dangerous areas, including contributors who receive limited credit from goals and assists.

It may highlight:

• Defenders whose distribution breaks the first line of pressure.
• Midfielders who find teammates between opposition lines.
• Full-backs who progress into advanced wide areas.
• Wingers who carry the ball towards the penalty area.
• Creative players who connect build-up with the final third.

Total xT should not become a simple ranking of creativity or player quality. A player in a dominant team will usually receive more possession, operate in better territory and benefit from stronger teammates.

Useful supporting views may include:

• xT per 90 minutes.
• xT per touch or possession.
• Passing xT compared with carrying xT.
• xT by starting and destination zone.
• Positive xT compared with turnover costs.
• Output by game state and opponent quality.

No adjustment removes every role and team effect. Per-90 totals may favour players in possession-dominant teams, while per-touch figures can overstate high-risk specialists.

What Expected Threat Reveals About Teams

At team level, xT can help explain how a side creates attacking danger rather than simply recording how much it shoots.

Analysts can examine:

• Which areas generate the most progression.
• Whether attacks develop through central or wide positions.
• How effectively the team enters the final third.
• Which players connect build-up with attack.
• Whether territorial threat becomes high-quality shots.
• Where opponents are able to progress against the defence.

A team with strong xT but modest xG may be progressing effectively without converting that territory into good chances. This is a prompt for further analysis, not proof that goals will inevitably follow.

Shot maps can help test what happened after the progression. They show whether threatening possession eventually produced central chances, blocked attempts, speculative shots or no shot at all.

Turnover Risk and Unsuccessful Actions

The treatment of unsuccessful actions is one of the most important differences between xT implementations.

A simple analysis may award positive xT only to successful passes and carries. That can make an ambitious player look highly productive when successful without recording how frequently the same player surrenders possession.

Other approaches may:

• Assign negative value to the threat lost through a turnover.
• Estimate the opponent’s counterattacking threat after possession changes.
• Report successful progression and failed-action costs separately.
• Calculate a net on-ball contribution rather than positive xT alone.

These are materially different measures. A ranking based on successful xT added should not be interpreted as though it already includes turnover risk.

The location of the loss matters as well. Failing with a difficult pass near the opposition penalty area may be less costly than surrendering possession during a central defensive build-up.

Expected Threat and Defensive Context

xT can also be applied to defensive analysis by examining the threat a team allows opponents to generate.

Possible measures include:

• xT conceded per match.
• xT conceded per opposition possession.
• Threat allowed through central and wide areas.
• xT conceded following turnovers.
• The relationship between xT conceded and xG conceded.

A defence may allow opponents to hold the ball while forcing progression towards low-value wide areas. Another may suppress shots but repeatedly allow entries into dangerous central zones, relying on last-ditch interventions to prevent attempts.

Basic xT still provides only a partial defensive view. It may not credit:

• A defender whose positioning removes a passing option.
• A pressing run that forces the ball backwards.
• A covering player who allows a teammate to challenge.
• Compact team spacing that prevents central progression.
• Communication that maintains the defensive structure.

Event-data xT values recorded ball movement more readily than off-ball prevention. Tracking data can capture more spatial context, but the output still depends on modelling choices.

How Game State Influences xT

Expected Threat totals are shaped by the score and match situation.

A team trailing late in a match may commit more players forward, accept greater risk and generate more attacking possession. Its xT can rise because its incentives have changed, rather than because its underlying ability has improved.

The leading side may defend deeper and accept territorial pressure while protecting central areas. It may create little sustained xT but remain dangerous on the counterattack.

Red cards, venue, opponent quality and competition format can distort comparisons further. Analysts should interpret xT alongside game state and, where possible, separate output by:

• Winning, level and losing periods.
• Equal-strength and red-card periods.
• Home and away matches.
• Opponent quality.
• Open play and set pieces.

Why xT Varies Between Providers

There is no definitive public xT model. Provider outputs can differ because of:

• Pitch-grid dimensions.
• The historical competitions and seasons in the training data.
• The definition of a possession.
• The treatment of passes, carries and dribbles.
• Whether failed actions receive negative value.
• Whether the model includes conceding risk.
• The scoring horizon being estimated.
• Whether contextual or tracking information is included.

Two providers can evaluate the same action differently without either calculation necessarily being wrong. Comparisons across players, teams and seasons should use a consistent source and methodology.

Analysts should also confirm whether the reported number represents positive xT, net xT, xT from passes, xT from carries or another provider-specific variation.

The Main Limitations of Expected Threat

Basic models value zones rather than complete situations

Two passes can start and finish in the same grid cells while creating very different receiving situations. One may find an unmarked player facing goal. The other may reach a player surrounded by defenders with their back to goal.

A location-only model may award the same value to both.

Off-ball movement receives limited credit

A run that drags a defender away can create the space for a high-value pass without the runner touching the ball. Event-data xT generally credits the player moving possession rather than every player responsible for creating the opportunity.

Volume can be confused with efficiency

Players and teams with more possession have more opportunities to accumulate xT. High totals can reflect ability, tactical role, team dominance or a combination of all three.

High xT does not guarantee a good shot

Moving possession into a dangerous zone increases attacking potential. It does not ensure that the team selects the correct final action or produces a high-quality attempt.

Negative actions may be missing

If a dataset reports only xT from successful progression, it may overstate players who attempt large numbers of risky actions. Turnover costs must be examined separately unless they are explicitly included.

Using xT in Match Predictions

xT can provide information that goals, results and shot totals miss. Potential forecasting signals include:

• A team repeatedly progressing into threatening areas despite a poor scoring run.
• A successful side creating little territorial threat and relying on efficient finishing.
• A defence allowing frequent high-threat entries despite conceding few recent goals.
• The absence of a player responsible for a large share of progression.
• A tactical match-up that exposes the zones through which an opponent creates threat.

These are investigation signals rather than automatic predictions.

xT may describe what has happened without adding reliable forecasting power. The sample could be small, the schedule unusually weak or the pattern driven by game state. The next opponent may defend the relevant spaces differently, and the betting market may already reflect the information.

A disciplined match-analysis framework should combine xT with shot quality, team strength, tactics, player availability, market expectations and uncertainty.

Using xT in Recruitment

xT can help recruitment teams identify players whose progression is understated by goals and assists. It may reveal defenders who break pressure, midfielders who progress centrally or wide players who create threat through carries.

Recruitment comparisons still need to account for:

• Team possession and territorial dominance.
• The player’s tactical role.
• Opponent and competition strength.
• The availability of forward passing options.
• Turnover frequency and defensive consequences.
• Whether the contribution would transfer to a different system.

A high xT total does not independently prove that a player is creative, adaptable or suitable for the recruiting club. It identifies actions for further statistical, contextual and video analysis.

How to Interpret Expected Threat Properly

Before relying on an xT figure, ask:

• Which model and data provider produced it?
• How is the pitch divided?
• What future outcome is the model estimating?
• Which passes and carries are included?
• Are unsuccessful actions and turnovers penalised?
• Does the metric include defensive or conceding risk?
• Is the number a total, per-90 rate or per-action measure?
• How much does team role and possession volume explain?
• Has game state and opponent quality been considered?
• Does xT add information beyond xG, results and market prices?

This is more useful than treating xT as one universal ranking. As with every advanced metric, the important question is whether it provides relevant information for the decision being made.

Key Takeaways

• Expected Threat assigns values to pitch zones based on the future scoring potential of possession.
• Passes and carries add xT when they move the ball into more threatening locations.
• xT measures progression before a shot, while xG evaluates shots and xA values passes that directly create shots.
• xT is one type of possession-value model, not a synonym for every action-value framework.
• Longer or more progressive-looking actions are not automatically more valuable.
• Turnover costs and unsuccessful actions are handled differently across models.
• Provider outputs vary because grids, datasets, targets and action definitions differ.
• High xT does not guarantee good shots, goals, player quality or future results.
• Prediction and recruitment uses require tactical context, consistent data and evidence that the metric improves the intended decision.

Related Guides

• Possession Value Explained
• Field Tilt Explained
• Shot Maps Explained
• How Professional Football Bettors Build a Match Analysis Framework

Stay Ahead of the Market

Subscribe to GoalIQAI for evidence-led football analytics, betting-market analysis and practical explanations of how data, tactics and probability fit together.