Expected Threat (xT) Analysis: How to Use xT Metrics for Football Predictions

Introduction

Expected Threat (xT) analysis represents one of the most significant advances in football analytics over the past several years, offering a fundamentally different way of measuring attacking value and territorial progression than the metrics that preceded it. While expected goals (xG) measures the probability of a specific shot resulting in a goal, expected threat quantifies the value of any action on the pitch — pass, dribble, carry, or cross — in terms of its contribution to a team's probability of scoring from a given position. This makes xT a more holistic measure of attacking quality, capturing the progressive phases of play that build toward scoring opportunities rather than only the terminal event (the shot) itself. For football predictions, expected threat analysis provides a window into the underlying attacking effectiveness of teams that raw statistics like possession, pass completion, and even shots simply cannot match.

This guide provides a thorough examination of what expected threat measures, how it is calculated, what the metric reveals about team attacking patterns, and — most importantly — how analysts can use xT data to build more accurate and insightful football predictions. Whether applied to the identification of teams whose underlying attacking quality is under-represented by their recent results, the assessment of player-level contributions to attacking play, or the evaluation of tactical systems that generate or suppress xT effectively, the expected threat framework adds a genuinely valuable analytical dimension to any prediction methodology.

What Expected Threat Measures and Why It Matters

Zone-Based Threat Assessment in Football

The fundamental concept behind expected threat is elegant. Rather than only measuring the probability of scoring from a shot — which is what xG does — xT assigns a value to every position on the pitch representing the probability of a goal being scored within a small number of actions from that position. When a player moves the ball from a lower-value position to a higher-value position — through a pass, dribble, or carry — the difference in xT values between the starting and ending positions represents the "threat increase" created by that action. These incremental threat increases, summed across all actions in a possession, provide a comprehensive measure of how effectively a team is building attacking value throughout a move.

Why xT Outperforms Traditional Possession Metrics

To understand why this matters for predictions, consider the limitation of xG as the sole attacking metric. Two teams might generate identical xG in a match — say, 1.5 each — but via very different routes. One team might create their 1.5 xG through long shots from outside the box and unlikely headers from tight angles, while the other generates their xG through sustained progressive passing into high-value central areas followed by shots from prime positions. The xT profiles of these two teams would look very different: the second team is generating much more territorial value throughout their attacking sequences and is building their xG through high-quality progressive actions. This difference in underlying attacking methodology is predictive of future performance in a way that the identical xG figures do not capture. The team with superior xT-generating processes is more likely to sustainably produce quality scoring opportunities than the team relying on lower-quality attempts.

For prediction analytics, this distinction is enormously valuable. Teams that generate high xT relative to their actual goals or xG are typically teams whose underlying attacking quality is better than their results suggest — potential candidates for positive form reversion. Teams that concede high xT despite defending well in terms of xG-against may be building structural weaknesses that will eventually translate into increased goals conceded. The xT framework thus provides leading indicators rather than lagging indicators of attacking quality, making it one of the most forward-looking metrics available to prediction analysts. The foundational context for understanding how advanced metrics feed into prediction models is covered in our guide on data-driven predictions and statistical accuracy.

How xT Is Calculated: The Zone-Based Framework

The Pitch Zone Grid and Transition Probabilities

The calculation of expected threat was formalised by analyst Karun Singh in 2018, building on earlier work in Markov chain models of football possession. The approach divides the pitch into a grid of zones — typically 16 x 12 or similar resolution — and assigns each zone a threat value based on the historical probability of the ball moving from that zone to a shot that results in a goal within the next several actions. These zone values are derived from large match datasets that track the outcomes of possessions from every starting position on the pitch.

The intuition behind the zone values is straightforward: the central area just outside the opposition penalty box has high xT because many dangerous attacks originate from this zone. The central channel inside the penalty area has even higher xT. The wide areas near the corner flag have low xT because few goals directly result from possession in those areas. By mapping these zone values across the entire pitch, xT creates a geographic landscape of attacking value that captures the progressive dimension of attacking play.

Calibrating xT Models with Match Data

When a player receives the ball in a low-xT zone (say, 0.02) and passes to a teammate in a high-xT zone (say, 0.08), the passing action has generated 0.06 xT. If a player dribbles from a moderate-value zone to a high-value zone — as a central midfielder carrying the ball from 30 yards out to the edge of the penalty area — the xT increase of that carry is calculated from the difference in zone values. Summing all these incremental values across a possession and across a match gives the total xT generated by each team, providing a measure of how effectively they have been progressing the ball into dangerous territories. Analysts interested in how this connects to the shot-based metrics should study our detailed guide on expected goals methodology, as xT and xG are complementary tools that together provide a comprehensive picture of attacking quality.

xT as a Predictor of Future Performance

xT and Future Goal Expectation Correlation

The most powerful analytical application of expected threat data lies in its forward-looking predictive capacity. Because xT captures the process of attacking play — how efficiently teams progress the ball into dangerous zones — rather than merely the outcomes (goals scored), it is less susceptible to the short-term variance that affects goal counts and even xG. A team that consistently generates high xT per match, but has been converting fewer opportunities into goals than the xT total would predict, is likely to see positive performance reversion over the coming fixtures. Their attacking process is sound; their outcomes have been temporarily suppressed by finishing luck, goalkeeper form, or defensive errors in execution.

Research using large samples of professional football data confirms that xT-based metrics show stronger correlation with future goal-scoring performance over 5-10 match windows than raw goal totals alone. This is because goal totals are volatile — a single fortunate or unfortunate match can swing them significantly — while xT requires sustained patterns of territorial progression to produce high values. When xT and goals align — a team scoring many goals and generating high xT — the prediction of continued strong performance is robust. When they diverge — high xT but few goals — the analyst should expect reversion toward the xT-predicted scoring rate rather than the recent goals rate. For prediction analysis, teams in a positive divergence state (high xT, under-delivering in goals) represent particularly interesting analytical opportunities, as their underlying quality suggests better results are forthcoming.

Using xT to Identify Overperforming and Underperforming Teams

This principle connects directly to form guide analysis. When reviewing recent results to build a prediction, the form guide should include xT data alongside actual results and goal counts. A team on a three-match winless run may actually show a consistently high xT across those matches, suggesting that the results have not reflected the team's actual quality of play. Conversely, a team on a three-match winning run with low xT may have been fortunate, and their upcoming fixtures carry greater uncertainty than the win streak suggests. Our guide on form guide analysis using recent results provides the framework for integrating xT into form assessment alongside conventional results data.

Player-Level xT Analysis and Its Prediction Value

Dribbling and Carrying Contributions to xT

Expected threat analysis extends beyond team-level aggregates to provide meaningful player-level insights. For any individual player, xT accumulation tracks the value they personally add to attacks through their passes, carries, and dribbles. This player-level xT data reveals which players are genuinely driving their team's attacking threat generation, as distinct from players who may benefit from good positioning without themselves creating the progressive ball movement that leads to chances.

For prediction analysis, player-level xT is particularly valuable in two specific contexts. First, when assessing the impact of a key player's absence through injury or suspension, xT data quantifies how much attacking value that player contributes and how difficult it will be for their team to replace it. A central midfielder who accounts for 25% of their team's total xT through progressive passing and ball carries represents a genuinely significant absence that should depress goal expectations for their team's upcoming matches. The importance of player availability for predictions is explored in depth in our team news and injury impact guide, and xT provides the quantitative tool to calibrate how significant a specific absence actually is.

Passing Patterns and xT Generation by Position

Second, player-level xT analysis is valuable for goalscorer and assist markets. The expected assists metric (xA) — which captures the xG value of passes that lead to shots — is closely related to xT, as it represents the specific subset of threat-generating passes that result in shot attempts. Players with high xT generation who also rank highly on xA represent prime candidates for assist selections. Our dedicated guide on expected assists (xA) analysis builds on the xT framework to develop specific methodology for assist-market predictions. These two metrics together — xT and xA — provide a complete picture of a player's contribution to the attacking process from early progressive ball movement through to shot creation.

xT in Defensive Analysis: Suppression Metrics and Predictions

xT Against as a Defensive Quality Indicator

Expected threat is equally valuable from a defensive analysis perspective. Just as teams can be ranked by their xT generation in attack, they can be ranked by the xT they concede — the territorial value opposition teams generate when playing against them. Teams that allow high xT-against are structurally vulnerable to progressive attacking play, even if their actual goals conceded have been relatively low (perhaps due to excellent goalkeeping or fortuitous results). Identifying teams with high xT-against is an early-warning signal for future defensive vulnerability that precedes it becoming visible in goals conceded statistics.

Low-Block Teams and xT Suppression Profiles

Defensive xT suppression — the ability to limit the threat value progression of opposition attacks — is one of the most important defensive qualities that conventional statistics fail to capture. A team might have a low shots-against total, but if the shots they do allow come from high-xT positions because their defensive structure is being broken down regularly, their defensive quality is lower than the shots count suggests. Conversely, a team allowing high shots-against from low-threat positions is not as vulnerable as the shot count implies. For prediction analysis, the xT-against metric provides a more accurate basis for assessing defensive quality and forecasting goals conceded than shots-against or even raw xG-against alone.

Combining team-level offensive xT with defensive xT-against provides a prediction framework for both teams scoring in any given match. A match between a high offensive xT team and a high xT-against team has structural features that support both teams creating genuine scoring threats, which informs both-teams-to-score analysis. Our comprehensive BTTS prediction guide provides the analytical framework that xT data enhances when assessing whether both sides are genuinely likely to create and convert quality opportunities.

xT in Tactical Systems Analysis

High-Press Systems and xT Creation Patterns

Different tactical systems generate expected threat in characteristically different ways, and understanding these system-specific xT profiles is a powerful tool for prediction analysis. Teams that play a high-pressing, rapid-transition system — like Liverpool's archetypal Jurgen Klopp approach — generate significant xT through quick vertical transitions that move the ball from defensive or midfield positions to attacking positions in one or two actions. Their xT profile shows high values from midfield zones resulting from quick switches of play and direct vertical passes into the penalty area channels.

Counter-Attacking Teams and Concentrated xT Zones

By contrast, teams that build through sustained positional play — Barcelona, Manchester City under Pep Guardiola — generate xT through gradual territorial accumulation, moving the ball across the pitch to create overloads and patiently progressing through zones rather than seeking direct transitions. The xT profile of such teams shows high cumulative values from many individual moderate-value actions rather than fewer higher-value individual plays. Neither approach is inherently superior in xT terms — they can both generate equivalent total xT — but they produce very different prediction implications for how goals are likely to arrive and whether the match environment will be fast and open (transition-based teams) or controlled and patient (possession-based teams).

Understanding these system-level xT patterns helps analysts predict how specific matchups will develop. When a transition-heavy team faces a possession-dominant team, the tactical interplay often creates a specific match environment with predictable goal timing patterns — the possession team generating slow-burn xT accumulation while the transition team seeks to exploit high-value counter-attacking positions. The tactical formations analysis guide provides the complementary tactical vocabulary for interpreting these matchup dynamics in conjunction with xT data.

Available xT Data Sources for Analysts

The practical use of expected threat analysis depends on access to data sources that track and report xT metrics. As of the current analytical environment, xT data is available through several routes. StatsBomb — whose data is used extensively in academic football analytics — provides xT-related metrics in their event data product, which is available to clubs, media organisations, and analysts through commercial licensing. FBref, the free analytical website that draws on StatsBomb's public data, provides xT-related statistics for players in major leagues, accessible without a paid subscription. Wyscout and InStat provide event data with carrying and progressive passing metrics that approximate xT-like values for scouting and analysis purposes.

For analysts who cannot access proprietary xT data directly, proxy metrics — such as progressive passes, progressive carries, and passes into the final third — provide indirect insight into xT-generating patterns. A team consistently ranking in the top tier for progressive passes and carries is almost certainly generating high xT, even without direct xT calculation. These proxy metrics are available through free sources including FBref, Sofascore, and WhoScored, making xT-informed analysis accessible even without premium data subscriptions. The approach to building prediction models from available data — including proxy metrics where primary data is unavailable — is covered in our guide on building your own prediction model.

Expert Insight: Analysts who have worked with expected threat data across multiple seasons identify one consistent finding that stands out for its prediction value: the divergence between xT generation and goals scored is most predictive at the 5-8 match sample level. At shorter time horizons — one to three matches — xT divergence is too noisy and volatile to be reliably predictive. At longer horizons — 20 or more matches — the regression to mean is typically already visible in the results data, meaning the opportunity has already partly manifested. The sweet spot for xT-based prediction analysis is the 5-8 match window: teams that have generated consistently high xT across five to eight recent matches but whose results under-deliver on this quality are statistically likely to see improved outcomes in their next three to five fixtures, making them systematically more interesting as prediction subjects than their recent results alone would suggest. Building this kind of xT-divergence tracking into a regular analytical routine — reviewing team xT figures weekly alongside results — creates a structured, forward-looking signal that complements conventional form analysis with genuine predictive power beyond what the result history provides.

Analyst Note: When incorporating expected threat analysis into your prediction workflow, the following practical notes apply. First, prioritise xT data sources that cover the specific competition you are analysing — not all leagues are covered equally, and using proxy metrics where direct xT is unavailable is a reasonable workaround. Second, always contextualise xT by opponent quality — high xT against weak opposition is less predictive than high xT against strong opposition, so weight the matches that generated the xT data by the defensive quality of the teams being played against. Third, look for the divergence signal: teams with high xT but below-expected goals are the most analytically valuable targets for positive reversion predictions, while teams with low xT but above-expected goals warrant caution as their results may be unsustainable. Fourth, integrate player-level xT into your injury and suspension analysis — a player in the top 10% for xT generation in their league represents a material absence when unavailable, and models should reflect this. Fifth, note that xT is most useful for predicting medium-term trends rather than single-match outcomes; its value increases when applied to analytical questions about which teams are genuinely better than their record suggests, rather than being used to predict any specific match result with precision. For analysts working across multiple metrics simultaneously, our PPDA pressing metrics guide provides a complementary framework that, used alongside xT, gives a comprehensive picture of team attacking and defensive quality.

Case Studies: xT Analysis in Real Football Prediction Contexts

In the 2022-23 Premier League season, Brentford emerged as one of the most analytically interesting xT case studies. Through the first 15 matches, Brentford's xT generation was consistently above the league median, driven by Ivan Toney's ability to receive and carry the ball in high-value zones while creating xT through layoff passes to runners. Their actual goal output lagged slightly behind this xT, with several matches producing fewer goals than the xT-suggested range. Analysts tracking this divergence would have noted Brentford as a positive reversion candidate, and their subsequent improvement in results over the following ten matches was broadly consistent with the xT picture. This illustrates the predictive value of xT at the team level for identifying form reversal candidates.

At the player level, consider the case of a central attacking midfielder playing in the Championship who consistently generates high xT through progressive ball carries and key passes, but plays for a team with limited finishing quality. Their xT contribution is high, but the downstream conversion of this threat into goals is poor due to the teammates receiving in the final third. Prediction analysts tracking player-level xT would correctly identify this player as generating genuine quality despite poor team results — making their team an interesting analytical subject in matches where the opposition's defensive quality is also poor, creating an environment where the xT generation might finally convert into goals more efficiently.

A third case study comes from the defensive xT application. A Premier League side in the 2021-22 season conceded relatively few goals in the first half of the season but showed consistently high xT-against — opposition teams were regularly progressing the ball into high-threat zones but failing to convert the resulting opportunities. By the second half of the season, this underlying defensive vulnerability manifested in an increased goals-conceded rate as opposition conversion luck normalised. Analysts who had tracked the xT-against data mid-season would have correctly anticipated increased defensive fragility before it became visible in the goals-conceded statistics — a genuinely forward-looking analytical insight with clear implications for goal-total and BTTS predictions involving this team.

xT and Asian Handicap Analysis

The application of expected threat data extends naturally into handicap market analysis. Teams with significant xT advantages over their opponents in a given matchup are structurally positioned to create more attacking value throughout the match, which supports handicap analysis favouring the xT-superior side. When the xT differential between two teams is large and consistent — not a product of one unusual match but of sustained patterns over multiple fixtures — this provides a robust analytical basis for evaluating whether a handicap selection appropriately reflects the genuine attacking quality gap between the two sides.

Conversely, when xT analysis suggests greater quality parity than the market price implies — perhaps because one team has a high current price based on recent results that don't reflect their underlying xT quality — this creates analytical opportunities in handicap markets. The Asian handicap methodology guide provides the complete framework for handicap market analysis, and xT data enhances this framework by providing a more reliable measure of true quality differential than results-based form alone.

Expert Insight: Expected Threat data becomes most valuable in prediction work when used to identify directional mismatches rather than absolute strength comparisons. A team with strong xT generation in central zones playing against a team that consistently allows high xT in those same zones represents a genuine predictive edge — the overlap between attacking strength and defensive weakness is the most reliable signal xT analysis can provide. Analysts who use xT purely as a league ranking tool miss the comparative matchup dimension that drives real prediction value.

Conclusion

Expected threat analysis represents a genuine advance in the analytical toolkit available to football prediction analysts. By measuring the territorial value of every action on the pitch — not just shots — xT provides a more complete and forward-looking picture of attacking quality than any single conventional metric. Teams generating high xT are building genuine scoring threats throughout their attacks; teams conceding high xT-against are structurally vulnerable even when short-term goal totals have not yet reflected this. At the player level, xT quantifies individual contributions to the attacking process in ways that connect directly to injury-impact analysis, player form assessment, and assist-market predictions.

The practical application of xT in prediction analysis requires access to the relevant data — either through direct xT metrics from providers like StatsBomb or through proxy measures available on free platforms — and the discipline to use the metric appropriately as a medium-term predictor rather than a short-term result predictor. Analysts who build xT tracking into their regular pre-match preparation and use it systematically to identify quality divergence from results will consistently produce more accurate analytical assessments than those relying on conventional statistics alone. For a complete advanced analytics framework, xT should be used alongside the related metrics covered in our guides on expected goals, expected assists, PPDA pressing metrics, and shot maps and heat maps. Together, these metrics build a comprehensive statistical portrait of team and player quality that enhances prediction accuracy across all major football markets.