Goalkeepers Expected Goals Prevented and Post-Shot xG

You know that moment when a keeper pulls off a save that leaves you speechless? The one where you’re certain the ball was heading for the top corner, and somehow, they’ve got a hand to it. For years, we’ve described those moments with words like “world-class” or “instinctive.” But in the modern analytics landscape, we now have two metrics that attempt to quantify exactly what happened in that split second: Expected Goals Prevented (xG Prevented) and Post-Shot Expected Goals (PSxG). These numbers don’t just tell us if a goalkeeper made a save—they tell us how difficult that save actually was. And as you’ll see, they completely change how we evaluate shot-stoppers.

Let’s start with a distinction that’s often blurred. Standard Expected Goals (xG) measures the quality of a chance before the shot is taken. It looks at the position of the shooter, the type of assist, the angle, and a dozen other variables to assign a probability that a shot from that situation will result in a goal. That’s incredibly useful for outfield players and team analysis, but it has a blind spot: it doesn’t care where the ball actually goes. A shot from 20 yards out that balloons into the stands gets the same xG as a curling effort that kisses the post. That’s where Post-Shot Expected Goals (PSxG) comes in. PSxG re-evaluates the chance after the shot, considering the actual placement, velocity, and trajectory. If a striker hits a low-driven shot into the bottom corner, PSxG will be higher than the initial xG because the execution was perfect. If they scuff it straight at the keeper, PSxG will drop.

Now, here’s the crucial link to goalkeeping. The difference between the total xG a goalkeeper faces and the actual goals they concede is traditionally called “Goals Prevented” or “Goals Saved Above Average.” It’s a simple calculation: expected goals against minus actual goals against. If a keeper faces 50 xG but only concedes 40 goals, they’ve prevented 10 goals. Sounds good, right? The problem is that xG doesn’t account for shot quality. A keeper might face a high xG total because the defense allows dangerous positions, but if those shots are all straight at the keeper’s chest, the xG is misleading. PSxG solves this. By using PSxG, we can calculate a more accurate measure: the difference between the total PSxG faced and the actual goals conceded. This is often called “Post-Shot Expected Goals Prevented” or simply “PSxG Prevented.”

To make this tangible, imagine two scenarios from a recent match. In the first, a striker receives the ball 8 yards out, unmarked. Standard xG for that chance might be 0.35. The striker takes a weak shot straight at the keeper. PSxG would downgrade that to maybe 0.10 because the placement was poor. The keeper makes an easy save. Traditional xG Prevented would credit the keeper with preventing 0.35 goals (0.35 xG – 0 goals = +0.35). PSxG Prevented would only credit them with 0.10 (0.10 PSxG – 0 goals = +0.10). The keeper didn’t do anything special; the striker missed a good chance. The second scenario: a striker shoots from 18 yards, xG of 0.08. But they curl it into the top corner. PSxG jumps to 0.40 because the placement is elite. The keeper gets a fingertip to it but can’t keep it out. Traditional xG Prevented would punish the keeper: 0.08 xG – 1 goal = -0.92. That’s brutal—they conceded a goal that looked unlikely. PSxG Prevented would show: 0.40 PSxG – 1 goal = -0.60. Still negative, but much fairer. The keeper faced a high-quality shot from a low-quality position. The metric recognizes the execution, not just the starting point.

Let’s look at how these metrics perform across different tactical systems. A goalkeeper in a high-pressing team that uses a 4-3-3 formation often faces fewer total shots, but those shots tend to be high-quality counter-attacks. The defense compresses the pitch, and when it breaks, the striker is through on goal. That keeper’s PSxG per shot will be high because the chances are dangerous. A keeper in a deep-block 3-5-2 system, on the other hand, faces more shots, but many are from distance with bodies in the way. The PSxG per shot will be lower. If you only look at standard xG Prevented, the 4-3-3 keeper might look average because they concede more goals from fewer shots, while the 3-5-2 keeper looks great because they face high xG totals but save most of them. But PSxG Prevented reveals the truth: the 4-3-3 keeper is facing harder shots and doing well to save any, while the 3-5-2 keeper is facing easier shots and should be saving most of them. The metrics untangle the system from the individual.

To illustrate the difference between these metrics, consider a hypothetical comparison of two keepers over a season:

In this example, Keeper A looks terrible by traditional xG Prevented, but PSxG Prevented shows they were actually slightly above average. Keeper B looks like a hero by standard metrics but is merely solid by PSxG. The narrative flips entirely. This is why scouts and analysts now prioritize PSxG-based metrics over raw xG Prevented. It’s not that xG Prevented is useless—it’s that it measures a combination of goalkeeper skill and defensive structure. PSxG Prevented isolates the goalkeeper’s impact on shot-stopping.

There’s also a growing conversation about the limitations of PSxG. The model relies on accurate tracking data for shot placement, velocity, and trajectory. Not all data providers capture this with the same precision. A shot that clips the inside of the post and goes in might be recorded differently by different systems. Additionally, PSxG does not account for the goalkeeper’s positioning before the shot. A keeper who reads the game well and is already in the right spot might make a save that looks routine but is actually the result of exceptional anticipation. That intelligence isn’t captured by PSxG. Some analysts argue that we need a metric like “Goalkeeper Expected Positioning” to complement PSxG, but that’s still in development.

For bettors and fans, understanding this distinction is crucial. When you see a keeper touted as having “the best save percentage in the league,” ask yourself: what’s their PSxG Prevented? A high save percentage against low-quality shots is less impressive than a moderate save percentage against elite chances. Similarly, when a keeper is criticized for conceding “too many goals,” look at the PSxG they faced. They might have been hung out to dry by a defense that allowed high-quality attempts. The market often overcorrects for these narratives, creating opportunities for informed analysis.

Let’s also touch on how these metrics interact with set pieces. A corner kick in a 4-2-3-1 setup often generates a high xG because of the proximity to goal, but the actual shot placement varies wildly. A header that goes straight at the keeper has a low PSxG; a glancing header to the far post has a high PSxG. A keeper who excels at reading crosses and positioning themselves to minimize the PSxG of headed efforts is incredibly valuable, even if their raw shot-stopping numbers are average. This is where the marriage of traditional scouting and analytics shines.

Finally, a word on responsible use. While PSxG Prevented is a powerful tool, it’s not a crystal ball. A keeper can have a season where they outperform their PSxG by a significant margin—that’s often called “overperformance” or “regression candidate.” But it doesn’t mean they’re guaranteed to drop off. Some keepers genuinely have elite reflexes and positioning that allow them to consistently save shots that the model expects to go in. The key is to look at multi-season samples. A single season of +10 PSxG Prevented could be luck; three seasons of +8 to +12 is a skill.

Risk Disclaimer: Sports betting involves financial risk. The statistical patterns and metrics discussed in this article, including PSxG Prevented and xG Prevented, are analytical tools and do not guarantee future outcomes or betting success. Past performance is not indicative of future results. Always gamble responsibly and within your means.

In summary, PSxG Prevented is the gold standard for evaluating goalkeeper shot-stopping in modern analytics. It corrects the flaws of traditional xG Prevented by accounting for shot placement and execution. It reveals that not all saves are created equal, and not all goals conceded are the keeper’s fault. Next time you watch a match and see a keeper make a string of saves, don’t just count them—ask what the PSxG was on each shot. The answer might surprise you.

For more on how defensive systems influence shot quality, check out our guide on Expected Goals Conceded (xGC) and Defensive xG. And if you want to understand how pressing intensity affects the chances a keeper faces, read our breakdown of Pressing Triggers and Counter-Press Success Rate. For a broader look at the metrics that define modern football, visit our Player and Team Statistics hub.