What Are Goals Added (G+)?
Those who have been following this site are likely well aware of my appreciation for data. With the correct understanding of the data points and taken within context, it can help add to one's understanding of a match and a team’s performance. Previously, I had written a guide on one of the most popular advanced soccer statistics, expected goals (xG). xG certainly is a great statistic, however, its measure is limited to goal-scoring sequences. In the 90+ minutes of action on the pitch, there is a lot more happening. So many actions go unnoticed and/or underappreciated in traditional soccer statistics (1). Build-ups and midfield plays are largely a non-factor in this stat. Recognizing the gap, the brilliant minds at American Soccer Analysis set to work and ultimately created a metric that would capture these opportunities. Goals added (G+) measures a player’s total on-ball contribution in attack and defense by calculating how much each touch changes their team’s chances of scoring and conceding (2).
G+ is a rather new metric but other similar models have come before it (VAEP, Possession Value Framework, Expected Threat (xT), and Expected Possession Value (EPV)). While they all vary in their approach, they are all attempting to solve the same problem of assigning value to non-goal scoring opportunities. It was initially created to answer "How much is any given player contributing to his team's success?" (3). Steve McPherson said, “If goals show you what happened, and expected goals (xG) show you what probably should have happened, goals added seeks to show you everything that went into getting to the point where something should have happened.” (4). Much like xG, the G+ model relies on referencing similar possessions in the past to see what its overall contribution (good or bad) is valued at. The model assigns all actions into one of six categories; shooting, receiving, passing, dribbling, interrupting, and fouling (2). Every action is then assigned a value based on its context and its likelihood of contributing to the team scoring or being scored on. While xG does not do much for defenders or midfielders not as involved in the attack, G+ can measure the value of their actions. It is based upon the likelihood of scoring and not the actual act of scoring itself. It doesn't give players any credit for actually scoring (2).
The above visual is a simple explanation of how one sequence is measured. Based on the historical data in the model, it has awarded +0.019 G+ for the pass due to the difference in Seattle’s chance of scoring and their chance of conceding. Build-up sequences or possession chains can now be assessed in a way that was previously not possible. To tie it back to Louisville City, think of a player like Corben Bone. While his play did not always net him goals or assists, you can tell by watching that he is an important piece of the puzzle. Cameron Lancaster can’t score without service and he is one of several to help make that happen. For the 2021 season, his cumulative actions netted him a G+ value of +0.28 with the majority coming from receiving and dribbling.
While Brian Ownby’s goal and assist counts in 2021 already paint a picture of a valuable player, his +2.29 G+ value (2nd highest on the team, 20th in the league) helps to even further that argument. As previously mentioned, his goals and assists did not influence this value but rather his cumulative actions netted him one of the top scores in the league. Looking at some of the individual actions, Sean Totsch had the 2nd highest passing G+ score, Paolo DelPiccolo had the 3rd highest receiving score, and Cam had the 7th highest shooting score in the league.
The cumulative actions from individual players can be summed together to get a collective score for a team. Louisville City ended with the 2nd highest G+ For and the lowest G+ Against, producing them the top G+ difference. In short, the players’ individual actions were the second-best for helping to produce goals and the best and preventing goals against. You will notice a positive correlation between the top teams and their placement in these rankings.
Like all other data points, it does have its limitations. For starters, the model is built on event data that can't see what's happening off the ball, so it's not very precise over a single game, but over enough minutes it's good at estimating the scoring probabilities in any given situation on the field (4). Like xG, it is more accurate and effective with a larger sample size.
Another limitation is its ability to capture defense. Given that defense particularly is about denying space, metrics based on event data like g+ are always going to be incomplete (4). This ties in very closely with the point above. Defense in soccer is largely predicated on keeping danger from ever happening — a thing that is vanishingly hard to measure with math that follows the ball (4). There are some other opportunities such as pass origins and pass and receipt values, however, I’ll let the creators at ASA tell you about those if you really want to get into the weeds (5). It’s a bit much if you are just trying to understand what G+ is and how it works at a relatively high level.
Goals added (G+) is a step further into the deep end regarding soccer and statistics but a fundamental understanding can help you to better dig into various performances. Having the ability to understand the positive and negative impacts of on-ball actions provides us with a way to more robustly measure how a player or team is doing. The game is much more than the shots and assists. G+ provides a value for the lead-up to the events based on historical performances. It’s a new model and one can fully expect that it will continue to see evolutions to it to further increase its value and validity.
I hope that this primer on G+ has helped your understanding and appreciation for the numbers behind the beautiful game! Big thanks to USL Tactics and Zach Allen-Kelly for their assistance with this piece!