I have been posting Goal Probability charts on Twitter and it has inspired some questions about what exactly it is. This post will be my explanation of it.
Essentially in the simplest terms, it is the difference in the probability of a team scoring or conceding after an event compared to the probability of scoring at the start.
For the calculations, I broke the field into 400 zones and then looked based on open play actions how often teams score and conceded from those positions.
Below the play is going from left to right.
Other people have done similar things, you'll see it as possession value, expected threat (although that can be a bit more complicated), goals+, or on-ball value. I have another version that I called attacking value-added that is very similar but goal probability added has slightly different ways of being calculated with things borrowed from other systems.
Here is an example of how things work from the Arsenal vs Norwich match.
Arsenal starts with the ball around midfield, pass it around the back for a bit (which doesn't add much value as you can see here), get the ball up the left-wing (which again doesn't really move the needle on the chances of scoring), before the ball recirculates back to Ben White who steps up (the dashed lines here are carries) and then plays a pass to Takehiro Tomiyasu (this adds 2%) who carries the ball central (1%) before taking a shot that gets blocked.
Overall during this sequence Arsenal max out at about a 5% chance of scoring.
This metric also takes into account losing the ball, either through a bad pass or being dispossessed. This is where the probability of conceding comes into play.
This is another good example, Norwich are attacking and Todd Cantwell completes a pass to Max Aarons on the edge of the box, he carries the ball towards the end line, dribbles past Martin Ødegaard (12% goal probability added) but then his cross is unsuccessful (-8% goal probability added). Part of the reason that this isn't losing all the value of the area from the pass is that even failed passes can lead to your team scoring from those situations.
I think that this is an interesting statistic and I really like to have the ability to look at these graphics to illustrate the value-added for each action.