Expected Goal Involvements

One number for two different jobs

Expected goals measures the chances a player takes. Expected assists measures the chances they create for someone else. Both are expressed in the same currency, the probability that a shot becomes a goal. That shared unit is what makes adding them together a genuine total, not a rough approximation.

xGI = xG + xA

Nothing more complicated than that sits underneath it. The value of the combined number is what it lets you compare on one scale:

• A poacher living almost entirely off high-value chances inside the box, contributing almost nothing to xA.
• A creator rarely shooting themselves, contributing almost nothing to xG.
• A hybrid forward doing a bit of both, neither number dominant on its own.

All three can post an identical xGI total while playing completely different roles, which is the trade-off for the convenience of one figure.

xGI rewrites three verdicts

Same total, different players. A striker posting 0.45 xG and 0.05 xA and a winger posting 0.10 xG and 0.40 xA can land on the same xGI number. Neither is more valuable than the other in the abstract. The number that matters is which profile a specific team actually needs.

Per 90 versus season total. A player with 60 minutes across three substitute appearances can post an eye-catching per 90 figure from one hot run of play. The season total tells a steadier story; the per 90 figure needs the minutes behind it checked before it means much.

Form before it shows in the box score. A player's rolling xGI over their last five matches, read against their season rate, often moves before their actual goals and assists do. A burst well above their normal level is a signal a slump in output is about to end, not a sign the player has been getting lucky.

The mix-ups, corrected

"xGI is the same as goal involvements." Goal involvements count what actually happened: real goals plus real assists. xGI counts the quality of the chances behind them, whether or not they were taken or converted. A player can lead the league in one and sit mid-table in the other. That gap is exactly what the over and under performance numbers exist to describe.

"A high xGI guarantees end product." It describes opportunity and chance quality, not certainty. A player generating strong xGI through missed chances and unconverted assists is doing the hard, repeatable part of the job. The finishing, theirs and their team mates', is a separate question layered on top.

"It flattens every attacking style into one number." That is a real limitation, not a myth. A single combined figure cannot tell a poacher from a creator on its own. Pair xGI with the xG and xA split beneath it, or with expected threat for the build up before either, to get the detail back.