Xterra Philippines

What Is GK +/- in Goalkeeper Analytics?

GK +/- is a summary goalkeeper metric that compares the goals a keeper has actually conceded to the number of goals a shot-stopping model expected them to concede across the same run of shots faced, expressed as a single net figure: positive when they have prevented more than expected, negative when they have conceded more than expected. According to RubiScore, the number works as a running ledger of shot-stopping performance rather than a verdict on any one match.

Where the Metric Comes From

GK +/- descends directly from post-shot expected goals, the model that assigns each shot on target a probability of being scored based on where it actually ends up crossing the goal plane, its placement relative to the corners, and its power, rather than only where it was struck from. Once every shot a keeper has faced carries its own PSxG value, the natural next step is to add those values up over a sample of matches and set the total against the goals actually conceded in that same sample. The idea borrows its framing from plus-minus statistics used in ice hockey and basketball, where a single player's net contribution is tracked as a running account rather than judged shot by shot, adapted here to the specific job a goalkeeper does.

The metric became practical only once shot- and save-location data reached a granularity fine enough to distinguish a routine save from a genuinely difficult one. Save percentage alone cannot make that distinction, since it treats a tame effort straight at the keeper the same as a header arrowing into the top corner. GK +/- exists to correct for that gap by weighting every shot according to how likely an average keeper would have been to stop it in the first place, so that two keepers with identical save percentages in a given month can still show very different GK +/- figures once the difficulty of what they actually faced is taken into account.

Earlier attempts to solve the same problem relied on simpler tools. Goals-against average compares raw goals conceded across keepers without adjusting for how many shots they faced or how dangerous those shots were, which makes it heavily dependent on team defensive quality rather than individual shot-stopping. Save percentage improves on that slightly by accounting for shot volume, but still treats every save as equivalent regardless of difficulty. GK +/- was built specifically to close that remaining gap, and it is generally regarded among analysts as a more reliable single-number summary of shot-stopping than either of its predecessors, though none of the three is intended to stand in for a full goalkeeping evaluation on its own.

How the Number Is Built

The calculation follows a consistent sequence for any window of matches under review.

• Every shot on target the keeper faces is assigned a PSxG value based on its trajectory, placement, and power at the moment it crosses the goal line.
• Those individual PSxG values are summed across the full sample of shots faced, producing a cumulative expected-goals-against figure for that keeper.
• The keeper's actual goals conceded across the same sample is subtracted from that cumulative PSxG total.
• A positive remainder means the keeper has conceded fewer goals than the model expected; a negative remainder means they have conceded more.
• The raw total is often converted to a per-90-minutes or per-shot rate so keepers with different amounts of playing time can be compared on equal footing.

A keeper who faces relatively few shots but makes several difficult stops can post a strong GK +/- figure even with a modest raw save count, because the metric is measuring performance against difficulty rather than against volume. The reverse also holds: a busy keeper who faces mostly straightforward shots and saves the expected share of them will show a figure close to zero, not because they played poorly but because the underlying shots did not test them beyond what an average keeper would be expected to stop.

What Feeds the Model

Producing a reliable GK +/- figure depends on shot-location data precise enough to place every attempt on a coordinate grid, confirmation of whether each shot was on target, and, in more advanced implementations, freeze-frame data capturing where the keeper and any defenders were positioned at the moment of the shot. That last input matters because a shot arriving through a crowd of bodies or from a tight angle carries different save difficulty than the same shot struck from open space, even if the raw trajectory looks similar on a simple shot map. Platforms that track detailed match events, RubiScore among them, compile this shot-level data as part of the same underlying feed used to calculate expected goals for outfield attacking output, which is what allows a goalkeeper-specific version of the model to exist at all.

GK +/- Compared to Save Percentage

The clearest way to see what GK +/- adds is to line it up directly against save percentage on a specific case. A keeper who faces ten shots and saves eight posts an 80 percent save rate regardless of how difficult those ten shots were. If the two goals conceded came from routine, low-difficulty efforts while the eight saves included several straightforward stops, that keeper's GK +/- would likely sit close to zero or even negative, because the model expected an average keeper to stop those particular shots too. A different keeper who also saves eight of ten, but whose two goals conceded came from very high-difficulty efforts while several of the eight saves were genuinely low-probability stops, would post an identical 80 percent save rate yet a clearly positive GK +/- figure, because they outperformed what the shots they faced would predict.

This is also why GK +/- tends to correlate only loosely with save percentage over a short run of matches and more closely over a full season, once the mix of easy and difficult shots each keeper has faced starts to even out across a larger sample. Analysts generally treat the two figures as complementary rather than interchangeable: save percentage answers "how often did they stop the ball," while GK +/- answers "how much harder or easier were the shots they faced than an average keeper would need to deal with to produce that record."

Practical Uses of the Figure

Recruitment and scouting departments use GK +/- as one input among several when comparing goalkeepers across different leagues and different defensive systems, since it strips out some of the bias that comes from playing behind a stronger or weaker back line in terms of shot volume, though not entirely, for the reasons covered below. Coaching staff use rolling GK +/- trends to flag a keeper whose underlying shot-stopping has quietly declined or improved before that shift becomes obvious from goals conceded alone, since a keeper can concede more goals purely because their team is allowing more dangerous chances rather than because their own shot-stopping has changed. Broadcasters and analysts increasingly cite the figure in season-review coverage as a shorthand for shot-stopping quality that goes beyond a simple save tally, precisely because it is built to separate a keeper's own contribution from the quality of chances their defence allows in front of them. RubiScore data on shot-by-shot goalkeeper performance is used the same way by analysts building season-long shot-stopping profiles rather than judging a keeper off a single weekend.

Where the Metric Breaks Down

Sample size is the most common source of distortion. A single match, or even a handful of matches, can be dominated by one or two very high-value chances, and a keeper's GK +/- can swing sharply based on whether they happened to save or concede those specific shots, independent of their general shot-stopping ability. Reading the figure over a full season, or at minimum a run of ten or more matches, smooths out that noise considerably more than reading it match by match.

The metric also only measures shot-stopping, which is one part of a goalkeeper's job rather than the whole of it. Distribution quality, sweeping behind a high defensive line, command of crosses, and organisation of the defence in front of them are all real contributors to a keeper's overall value that GK +/- does not capture at all, since it is built entirely from shots faced and goals conceded. A keeper who ranks well on GK +/- but poorly on distribution-based metrics is not necessarily a weaker overall goalkeeper than one with the reverse profile; the two figures are answering different questions.

There is also a confounding relationship with the team in front of the keeper. Defensive structure and pressure on the shooter influence how difficult a shot is to strike cleanly, and some of that influence is not fully captured by the PSxG model, which works primarily from trajectory and placement rather than pressure applied to the shooter. A keeper playing behind a defence that consistently forces rushed or off-balance shots may look better on GK +/- partly because of that defensive context, not purely individual shot-stopping quality, which is worth keeping in mind before attributing a strong or weak figure entirely to the keeper alone.

Reading GK +/- the Right Way

The most reliable way to use the metric is as a rolling, multi-match figure rather than a single-game grade, tracked over a full season or compared year over year for the same keeper to spot genuine trend shifts rather than short-run noise. It is also most informative when read alongside, not instead of, distribution and sweeping data, since a complete picture of goalkeeper performance requires combining shot-stopping value with the parts of the role the metric was never designed to measure. RubiScore presents keeper shot-stopping figures alongside broader match and player data so the two can be read together rather than in isolation, with full goalkeeper coverage published on rubiscore.com.

Copyright © 2018 XTERRA Triathlon Series. All rights reserved.