What Is xA (Expected Assists)? The Complete Guide for Football Fans

Ask a football fan what makes a great playmaker and you'll get a version of the same answer. Vision. The pass no one else sees. The ability to put a teammate through on goal in a moment everyone else would have played safe. What you won't usually get is a number. Playmaking has historically been the part of football hardest to quantify; by the time goals get scored, the creator is often already on the way back to his half, and the final stat sheet lists the assister only if the pass happened in the exact sequence immediately before the shot.

Expected assists, or xA, tries to fix that.

It's the closest thing modern football analytics has to a creative-player metric. Not perfect, not proprietary to any one provider, and frequently misread in exactly the same way xG is misread: used as a verdict instead of as a probability, or treated as a report card when it's really a distribution. This article walks through what xA actually measures, how to read it properly alongside xG, and the traps that catch even analysts who should know better.

What xA actually is

xA attaches a probability score to every pass in a match. The score answers one question: how likely is it that an average shooter, receiving this pass in this location under this defensive pressure, takes a shot that becomes a goal?

A square pass in the centre circle with nobody running in behind has an xA of roughly 0 because no shot is created. A through ball that drops a striker six yards out with an open angle might score 0.45 xA. Not because this particular striker scores from that kind of chance 45% of the time, but because across thousands of similar passes in the training data, the resulting shot went in 45% of the time.

Three things follow from that definition.

First, xA is a measurement of the pass, not the pass-giver. A midfielder who plays the same weight of pass into the same zone twenty times in a season accumulates roughly the same xA each time, regardless of how his teammates finish. That's a feature, not a flaw, but it trips up fans who expect xA to reward the combined quality of passer plus shooter.

Second, only passes that lead to shots are measured. The threaded ball that a teammate takes a heavy touch on and turns into a lost possession scores 0 xA in most public models. The moment was creative. The shot didn't happen. xA doesn't see it.

Third, xA and xG are two halves of the same chance. If a forward takes a 0.30 xG header from a cross, the cross that created the header also has an xA value (typically the same 0.30, because the pass is being graded on the quality of chance it produced). xA = xG of the resulting shot, conditional on a shot being taken.

That last point is where readers often go wrong. Reading xA as if it were a separate flavour of xG leads to double-counting. A 0.30 xA pass into a 0.30 xG shot is one chance, described from two angles.

How xA is calculated, in outline

xA models are trained on enormous libraries of passes, each tagged with contextual features and with the outcome of any shot that followed.

The features most public xA models rely on are broadly consistent across providers:

• Start location of the pass. Where on the pitch did the pass originate, measured as distance and angle to goal.
• End location. Where did the pass arrive. This is the dominant driver. Passes that land in the box carry higher xA than passes that land outside it, at least for shots that follow.
• Pass type. Through ball, cross, cutback, set-piece delivery, chipped ball, simple horizontal pass. Each pass type conditions the expected shot quality differently.
• Defensive pressure and body shape of the intended receiver. Some models include proxies for how closed-down the receiver is. This moves xA up or down.
• Game state. Open play, fast break, set piece rebound. As with xG, these phases have different conversion profiles.

More sophisticated models trained on tracking data can incorporate defender positioning relative to the pass and the receiver's body orientation. Public models without tracking data use simpler proxies.

Tactiq reads event-level pass data from licensed sports feeds covering more than 1,200 competitions. The per-pass xA values that feed into the analysis are derived from those event records alongside the broader match context the product looks at. The specific way xA combines with other signals inside the analysis stays within the product.

Why xA matters

A goals column rewards finishers. An assists column rewards the last pass before a goal. Both of those are noisy. A creative midfielder who plays a dozen through balls in a match, watches three reach the striker in strong positions, and finishes the game with zero assists because the striker missed all three did not have a quiet game. The goals column says he did. xA says he didn't.

xA matters for fans in several concrete ways.

It separates creation from finishing luck. A playmaker who racks up 8 actual assists off 4.5 cumulative xA is finishing above the model's expectation because of his striker's form, not because his passes are especially good. A playmaker who posts 2 actual assists off 6.0 cumulative xA is delivering elite creation but being failed by his shooter. Over a season, assists and xA tend to converge; in 10-fixture samples, they diverge heavily, and which direction they diverge tells you a useful story.

It makes creative midfielders visible. Number 10s and deep-lying playmakers who don't pile up raw assists often pile up xA. The gap between their xA ranking and their assist ranking is usually the difference between creation and finishing context.

It travels across leagues. A through ball that produces a 0.30 xA chance in the Dutch Eredivisie is recognisably the same creation as a 0.30 xA through ball in the Italian Serie A. The metric is portable in the same way xG is, which is what makes it useful for cross-league scouting and international comparison.

It rewards sustained creation over one moment. An assist rewards the pass that directly precedes the goal. A build-up that takes five passes to produce a shot gives the first pass zero assist credit, even if that first pass was the creative act that cracked the defence open. xA captures more of the chain, because every pass that leads to a shot gets measured, not just the final one.

Where xA misleads

This is the half most xA explainers skip. Being honest about where the metric breaks tells you more about how to use it than any definition of what it measures.

Small samples lie. Twenty passes is not a sample. A midfielder can post 1.2 xA in a match where his teammates missed a hatful and record 0 actual assists, while another midfielder gets 2 actual assists on 0.4 xA because his striker converted two weak chances. Neither result tells you about the underlying creation ability; it tells you about shooter outcomes that fixture.

Shooter quality is hidden. The xA formula assumes an average finisher. Playing alongside Haaland, Salah or Kane inflates your xA-to-assists conversion because these shooters beat the average. Playing alongside a weak finisher suppresses it. Cross-team and cross-era comparisons that don't correct for shooter context mislead by more than they clarify. The correction exists in advanced models but not in most public xA dashboards.

Pre-assists are not assists. A pass two moves before a goal is often the creative act that unlocked the move, but the xA model credits the pass immediately before the shot. Some modern "expected threat" and "possession value" models try to distribute credit more fairly across a possession; xA itself does not. Using xA to judge deep-lying playmakers who start attacks from midfield understates their contribution compared to final-third playmakers whose passes directly create shots.

Set pieces distort the headline. A corner taker who delivers 8 corners in a match, producing three headers from inside the six-yard box, accumulates high xA independent of creativity. The delivery is technical, not creative in the playmaker sense. Stripping set-piece xA out of open-play xA produces a cleaner picture of what a creator does in live play. Most public dashboards do not.

Crosses inflate volume over quality. A winger who bombs 15 crosses into the box in a match, of which the striker heads 3 at goal from difficult angles, will post higher xA than a winger who threads two through balls to a striker inside the box. The cross-heavy style accumulates xA via volume; the through-ball style accumulates xA via shot-quality density. Both can be right for a tactical context; xA alone doesn't tell you which.

Penalties and direct free kicks skew things. A drawn penalty that the penalty-taker converts is usually not logged as an xA event (the foul was drawn, not a pass played). A direct free-kick assist is rare but heavy when it occurs. These edge cases mean cumulative xA can occasionally deviate from a reader's intuitive sense of who "made the chance."

Late-game state effects apply, the same way they apply to xG. A team chasing a goal in the final fifteen minutes generates desperation passes into the box that inflate xA without reflecting sustainable creation. A team protecting a lead produces low xA because they're not trying to create. Full-match xA smears these phases together.

It's a team-level signal often misread as a player report card. A midfielder with 0.9 xA this match may have played four good passes into the box, none of which were heavy chances. Or one great through ball and eight sideways passes. The distribution matters. Cumulative xA over a single match hides that.

The rule that falls out of all of this: xA is most useful over a rolling window of several matches, read alongside xG of the resulting shots, with shooter quality held in mind and set-piece distortion stripped where open play is the question. It's least useful as a standalone verdict on a single fixture or a single season without context.

How Tactiq uses xA in the analysis

Tactiq treats xA the way this article has just described it: as one piece of underlying creation data, not a standalone playmaker verdict.

Inside a match analysis, xA signals contribute to the picture of which teams are generating meaningful chances versus which are churning passes that don't lead anywhere, which creators are performing above or below their underlying quality, and how the shape of a matchup looks through the lens of creation rather than finishing. xA sits alongside xG, form indicators, head-to-head context and other inputs. None of them is treated as the answer.

The specific way xA blends with the rest of what Tactiq looks at, the weights, the rolling windows, the open-play versus set-piece splits, the way unstable signals get flagged, stays inside the product. Published methodology gets copied and miscalibrated within weeks; what reaches the user is a confidence-qualified analysis with the reasoning explained in plain language.

What the user sees on the match card:

• Expected goals for each side, with the expected-assists context in the creation side of the read. You don't typically see an "xA: 1.8" number on screen; you see the effect of the creation picture on the confidence-qualified read.
• Probability triples for the outcome, with a visible confidence indicator that reflects how stable the underlying signals are for this specific fixture.
• Written analysis that names the creation context in plain English: "Home side's recent creation trend has lifted over their last four matches, though finishing has lagged, so the xG-to-goal gap has been wider than the underlying chance quality suggests."
• No external market data anywhere. No redirects to third-party platforms. No virtual currency. The frame is statistical analysis.

The intent is that a reader comes away with a sharper read on whether a team's finishing underperformance is a shooting problem or a creation problem, rather than a single decimal to copy somewhere else.

How to read xA like a pro

Six habits turn xA from trivia into a lens.

1. Always pair xA with xG and actual assists. A three-column view ("xA / xG of the resulting shots / actual assists") over a rolling window is more informative than any one column in isolation.
2. Adjust for shooter quality. Elite teammates inflate your conversion; weak teammates suppress it. If you're comparing playmakers, check whose strikers are finishing above expectation and whose aren't.
3. Strip set-piece xA when you care about open-play creation. A corner-taker with 0.9 xA from deliveries didn't create chances in the playmaking sense.
4. Read a rolling window, not one match. Four to eight matches smooths out the noise. One match is anecdote with a number attached.
5. Don't compare deep-lying playmakers to final-third creators on raw xA alone. The pass two moves before a shot matters. xA doesn't fully credit it. Models like "expected threat" capture that better; raw xA doesn't.
6. Weight recent form over season totals. A playmaker who hasn't produced chances in six weeks is a different player than his season-total xA suggests, regardless of what the cumulative number says.

Applied together, these habits turn xA from a number on a leaderboard into a piece of evidence that sharpens how you see the game.

The takeaway

xA is a probability about chance creation, not a report card on playmakers. Used inside a rolling window of several matches, read alongside xG and actual assists, adjusted for shooter context and stripped of set-piece inflation when open play is the question, it's one of the cleanest lenses football analytics offers on the creative side of the game.

Used as a standalone verdict, or as a leaderboard number without context, or as proof that a playmaker is or isn't delivering on a single season's evidence, it misleads. The metric is honest about what it measures. The reading is the part most analysts get wrong.

Tactiq is built around that reading. The app surfaces the creation picture inside a confidence-qualified match analysis, explains in plain language what the creation-versus-finishing gap means for a specific fixture, and never blends it with external market data. 1,200-plus competitions, 32-language localisation, free tier of eight analyses per day, no credit card required.

Three articles in a row now form the foundation of how we read the numbers. If you haven't yet, start with how AI predicts football matches and what xG actually measures. xA is the creation-side companion to that xG guide, and the three together cover the metrics that the rest of the blog keeps building on.