PSxG, xA and xThreat: modern football metrics in simple terms (with examples)

Football has always been about moments, but modern analysis tries to measure what leads to those moments. Three numbers you’ll see more and more in 2026 are PSxG (post-shot expected goals), xA (expected assists) and xThreat (expected threat). They’re not “magic” stats that replace watching matches, but they are useful for separating finishing from chance creation, and for putting value on actions that don’t end in a shot. This article explains what each metric actually measures, where it helps, where it can mislead, and how to read it with real, easy examples.

PSxG: measuring shot quality after the strike

Classic xG estimates the chance of scoring at the moment the shot is taken, based on factors like distance, angle and body part. PSxG goes one step later: it rates the shot after contact, using information about where the ball is headed (placement) and often how fast it travels. In practice, PSxG is widely used to assess finishing and goalkeeping because it focuses on the quality of the shot on target, not just the chance that existed before the shot.

Here’s a simple way to think about it: two players can take the “same” shot (same location, same angle), but one rolls it straight at the keeper and the other bends it into the top corner. Their pre-shot xG could be identical, yet their PSxG would differ because the second effort is genuinely harder to save. That’s why PSxG is handy when you want to distinguish “he got a good chance” from “he produced a genuinely high-quality shot”.

PSxG can also stop you over-crediting a goalkeeper who faces lots of weak finishing. If most shots on target are poorly placed, the keeper’s save total might look impressive, but PSxG-based evaluation can reveal that many saves were from low-danger shots after contact. As of 2026, different data providers implement PSxG slightly differently, so you should treat values as comparable within the same source rather than mixing models across sites.

PSxG in practice: finishing vs goalkeeping (worked examples)

Imagine a striker takes a shot from 12 metres, central. Pre-shot xG might be 0.30. If the player scuffs it and it dribbles into the keeper’s hands, a PSxG model could rate the on-target shot as something like 0.05 because the placement makes it easy. In one action you can see the split: the chance was decent (0.30), but the execution was poor (0.05).

Now flip it. Same shot location, same pre-shot xG around 0.30, but this time it’s rifled into the side-netting just inside the post. PSxG might jump to 0.70 or higher because the keeper has very little margin. If the keeper saves that, you can credit the goalkeeper for stopping an on-target effort that was genuinely difficult, rather than assuming all saves are equal.

For goalkeepers, analysts often compare goals conceded to the sum of PSxG faced on target (you may see labels like “PSxG – Goals Allowed” or similar). If a keeper concedes fewer goals than the PSxG suggests, it can indicate strong shot-stopping over a sample. The key phrase is “over a sample”: single matches swing wildly, and even half a season can be noisy, so it’s best used alongside video review and context (deflections, screens, set-piece chaos, defensive errors).

xA: crediting the pass that creates the shot

xA (expected assists) estimates how likely a shot is to become a goal, and assigns that value to the player who made the final pass leading to the shot. If you create a chance that has 0.40 xG and the striker misses, xA still credits the creator with 0.40. This makes xA useful for judging chance creation without being overly dependent on teammates finishing perfectly.

xA is especially helpful when comparing players with different roles. A winger who consistently slips cut-backs to the penalty spot may rack up high xA even if the team’s finishing is inconsistent. Meanwhile, a player might have a modest assist tally simply because teammates are converting below expectation, not because the chances are poor. xA doesn’t “prove” quality on its own, but it makes the conversation fairer.

There are limits. xA is usually tied to the shot’s xG, which means it inherits any model assumptions: how headers are valued, how defensive pressure is treated, and so on. It also undercounts creation that happens one pass earlier (the “pre-assist”), and it can miss the value of a carry that breaks the line before the final pass. That’s why you should read xA alongside other indicators of involvement, including touches in dangerous zones, progressive actions and (for some providers) secondary chance-creation metrics.

xA in practice: what a creator “deserves” from their service

Suppose a full-back overlaps and squares the ball to a forward six metres out, slightly wide. If the shot is valued at 0.45 xG, the passer gets 0.45 xA, regardless of whether it’s scored. Over ten matches, a player repeatedly making that pass could accumulate 3.0–4.0 xA even if the actual assist count is low, signalling repeatable chance creation.

Now consider a “cheap assist”: a sideways pass 30 metres from goal, then the shooter dribbles past two defenders and scores. Many models would assign the final pass a low xA because the shot’s xG at the moment of the shot may still be low or the creation is driven mostly by the dribble and finish. That doesn’t mean the passer did nothing, but it prevents the stats from treating every assist as identical in difficulty.

When comparing players, look for consistency and context. High xA with low assists can be bad luck or poor finishing around them; low xA with high assists can be hot finishing streaks or a few spectacular long shots. In 2026, clubs commonly use xA-style measures in recruitment shortlists, but they still validate with video: the “shape” of chances (cut-backs, through balls, crosses) matters for tactical fit.

xThreat: valuing actions that move the ball into danger

xThreat (expected threat) aims to measure something xG-based stats can miss: the value of actions that increase the likelihood of scoring later, even if they don’t immediately create a shot. It typically uses pitch zones and asks a simple question: “If the ball moves to this location, how much more likely is a goal to happen in the next phase?” Carries, passes and even smart lay-offs can gain value if they move the attack into a zone where goals are more likely.

This is why xThreat can highlight players who progress play rather than rack up shots or assists. A midfielder who consistently breaks lines with passes into the half-space might have modest goals and assists, but high xThreat contribution because those actions repeatedly shift the ball into high-leverage areas. Similarly, a winger’s carry to the byline can spike threat even before the cut-back is played.

xThreat is not a single universal formula. Some models are more “zone-grid” based, others incorporate sequence context more heavily. What stays consistent is the idea: it’s a bridge between possession and shots, putting numbers on territorial and structural advantage. In 2026, xThreat-style measures are common in public analytics circles and internal club work, but the best use is comparative: within the same model, who is reliably increasing danger and from where?

xThreat in practice: why a carry can matter as much as a key pass

Picture a team recycling possession near the touchline 35 metres out. A simple sideways pass might add almost no threat. But a midfielder receives under pressure, turns and carries ten metres into the half-space, forcing a defender to step out. Even without a shot, the ball is now in a zone where cut-backs, through balls and low crosses become realistic, so the carry gains xThreat.

Another example is the “third-man” pattern. Player A passes into Player B between the lines; B lays it off; Player C then plays the killer ball. Traditional xA credits only Player C. xThreat can credit Player A and B as well because their actions moved the ball into a more dangerous state and opened the defence. That’s useful when you want to describe how a team creates chances beyond the final pass.

To read xThreat sensibly, always pair it with role and team style. High xThreat from a full-back might mean they are constantly advancing into the final third, but it can also reflect a system that funnels progression wide. Use maps (where the threat is generated), check volume vs efficiency (many small gains vs fewer big jumps), and remember that game state matters: chasing a goal changes risk, and risk changes threat patterns.