The Comfortable Save | Where Shots Go vs Where Goals Come From

Shots on target sounds like a useful number. It's the level of 'making things happen' as a footballer that comes after "you miss 100% of the shots that you don't take". A player puts six of ten shots on target last night. That's good, right? But "on target" treats a 25-yard volley that arrows into the top corner and is majestically saved by the keeper the same as a scuffed effort that rolls straight to the keeper from 5 yards out. Both show up the same way in the stats, but our eyes tell us differently.

This is where expected goals (xG) offers a better picture: it factors in where the shot was taken from. Distance, angle, body part, and other context. But xG is assigned at the moment the shot leaves the foot. It knows nothing about where the ball is heading.

That's where post-shot expected goals (psxG) comes in. Also known as xGOT (expected goals on target), it asks: given where this shot on target is actually going, how likely is a goal to result? An effort with pace headed for the bottom corner from 25 yards might have an xG of 0.03 but a psxG of 0.45. The placement itself multiplied the chance of a goal by 15 times.

The gap between psxG and xG is the placement value: how much a player adds through where they aim, beyond what the shooting position gave them.

Outside-box shots are where this matters most. These are naturally low-xG chances: the shooting position gives you very little, so placement is the only thing that can rescue value. That makes them the cleanest test of shooting technique from distance. It also means the data skews toward midfielders and deep-lying playmakers who shoot from range, rather than strikers who get their chances closer to goal.

We looked at nearly 24,000 outside-box foot shots across eight Premier League seasons to map where they actually go and what happens when they get there.

The goal, from the keeper's perspective

Each cell shows the conversion rate for shots landing in that zone. Use the toggles to switch between what keepers see and what scouts should see.

How to read this

The goal is shown from the goalkeeper's perspective: left and right are mirrored from the shooter's view.
We combined both sides of the goal (near post / far post) into one half, since the data is symmetric.
Near post = within 1.5m of the post. Inner third = 1.5–2.5m. Central = more than 2.5m from either post.
Height bands: Ground (< 0.8m), Mid (0.8–1.6m), Upper (1.6–2.44m), Over bar — above the crossbar, off target vertically.
Wide — outside the posts horizontally, off target. Shown as a greyed column to the left of the post.
23,856 outside-box foot shots, 2017/18–2024/25.

The comfortable save zone

The mid-height band runs from 0.8 to 1.6 metres (2.6 to 5.2 feet) off the ground, around hip to shoulder height on a goalkeeper. A shot at this height amounts for 38% of all outside-box shots. Keepers save 82% of them. Comfortable.

At its horizontal extreme, toward the posts, this is standing diving height. The keeper doesn't always need to leave their feet or adjust their set position. The ball arrives in the zone where their hands already are or can quickly be.

By contrast, the upper corners (1.6m to the crossbar at 2.44m) receive just 8% of outside-box shots yet convert at 17%. Near the post and above 1.6m, the rate jumps to 51%, every attacking footballer's "name in headlines" zone.

The ground is the middle path: 31% of volume, 9% conversion. Low driven shots near the post convert at 27%, because keepers have to get down quickly and a firm enough strike can skip under a diving hand. A set defence also offers a sea of limbs you have to shoot through when hitting the ball low. Central ground shots (straight at the keeper) convert at 5%. Those tend to stick in the memory.

Where different players aim

Height profiles for selected players, showing how their outside-box shots distribute across the four vertical bands. The population average is shown for reference.

Sergio Agüero had the opposite profile: 28% upper corner, only 16% mid-height. He almost never gave the keeper an easy save.
Erling Haaland runs on ground power. His upper% is 0%, but 48% of his outside-box shots are ground-level, converting at 24%, well above the ground average.
James Ward-Prowse has the highest upper% (15.3%) among 200+ shot players. His +10.33 cumulative psxG-xG is the largest placement value in the dataset.
Bruno Fernandes has 306 outside-box shots, the most of any PL player in the data. More than half land at mid-height, in the comfortable save zone.
Antony is the paradox. His 11% upper-corner rate is nearly double the population average, and his 14% over rate is the lowest among high-upper shooters. He threads the crossbar zone without sailing over it. That generates a cumulative psxG-xG of +3.95. Yet he converts at just 1.6%. He buried little at United, let alone Caesar.

Sample sizes

Shot counts (n=) shown per player. Players with fewer shots have noisier profiles — a high upper% from 30 shots is less reliable than the same rate from 200. Conversion rates in particular are volatile at low volumes.
Data: 2017/18–2024/25 Premier League outside-box foot shots.

The correction that was already happening

Aston Villa scored 13 outside-box goals from 5.3 xG in 2025/26. A goals-minus-xG of +7.7, the highest in the Premier League by a factor of two. They converted at 10.5%, more than double the league average of 4.8%.

The popular explanation is injuries. McGinn, Tielemans, and Kamara, Villa's three senior central midfielders, all suffered serious injuries around the turn of the year. Without them, a collapse is entirely explainable. Except outside-box shooting tells a different story.

Cumulative overperformance for Aston Villa's outside-box foot shots in 2025/26. When the green area rises, Villa are beating xG. When it flattens, the overperformance has stopped.

Reading the chart

Green area (Goals − xG): cumulative overperformance. Steep rise = scoring well above expectation. Flat = regression.
Amber line (psxG − xG): cumulative placement value. When positive, shot placement is adding value beyond what position alone would predict. When it drops, shots are heading to easier zones.
psxG data available through ; later matches pending.

Between September and November, Villa's outside-box placement value was +2.28: their psxG outstripped their xG by more than two goals. They were picking corners. Nine goals from 2.01 xG.

By December, before a single central midfielder was injured, the non-CM group's placement value had flipped to −0.51. Rogers and Buendía were still shooting from the same positions, at the same rate, but their shots were landing in the comfortable save zone. The keeper didn't have to move.

What the injuries did was remove the safety valve. McGinn, Kamara, and Tielemans took only 17% of Villa's outside-box shots all season, but they converted at 14.3%. All three goals were precisely placed. Even without volume, McGinn and Tielemans offer the threat of a long shot that affects play, something difficult to quantify with data or trendlines. What we do know, is that when they fell out of the starting XI, the remaining players compensated with volume: 6.6 shots per match, double the December rate. Ten against Brentford. Nine against Leeds. Zero goals.

The mechanism wasn't personnel loss. It was high quality placement decaying, followed by increasing long shot volume to compensate. Exactly the pattern the comfortable save zone predicts. More shots into the same space the keeper already occupies doesn't magically rescue value. It dilutes it.

Morgan Rogers has gone cold indeed. Source: Aston Villa

The same stat line can mean completely different things depending on where a player's shots actually go. Which brings us back to the three metrics we started with.

Three stats, three valuable ideas

Shots on target tells you a player found the frame. It rewards mid-height efforts that keepers comfortably save. A player with 65% on-target rate might be consistently shooting into the comfortable save zone: volume without danger. We need something to tell us how likely they are to score from where they shoot.

xG tells you the quality of the chance based on where the shot was taken. It captures distance, angle, and other context. But it has nothing to say about what the player does with the ball after pulling the trigger. Two identical 0.05 xG shots can have completely different outcomes based on placement. We want something to tell us how likely they are to score from where they shoot.

psxG tells you the quality of the placement: where the ball is actually heading. A player whose psxG consistently exceeds their xG is adding value through their technique. Ward-Prowse's career +10.33 psxG-xG means his shot placement alone generated roughly ten goals' worth of value beyond what his shooting positions gave him.

None of these metrics is wrong. Each answers a different question. But each is also a single number, and single numbers often hide the shape of what's underneath. Two players with the same xG total can have completely different placement profiles: one peppering the comfortable save zone, the other picking corners with otherworldy reliability. Writ large, this applies to teams as well and can help explain why you can only outrun your xG for so long before reality catches up. If you're very lucky, it lasts a season. If you're not, then well, mayhaps you should start practicing attacking set pieces instead.

Want to know the latest news and stats from your club? Track them all season.