Non-Shot Expected Goals (NSxG)

You probably know about expected goals (xG) since it is a metric that has gone mainstream in the sports betting world. But what about non-shot based expected goals? It’s another model that can be used to measure a team’s performance, and it is highly interesting.

What is NSxG?

Non-shot based expected goals are a metric used to calculate the “fair” number of goals scored by a team. Unlike regular expected goals, the non-shot based model focuses on everything else than a shot – dribbles, passes, crosses, interceptions, and most important possession. It shares many similarities with the expected possession goals model which is also fairly interesting.

It makes sense to look into NSxG because regular expected goals are not always telling the full story. A player might score from a lucky rebound at a comfortable position, and it will add huge points to the regular xG score. In the non-shot based model, there is less luck involved, and it appreciates dangerous possession and well-timed attacks a whole lot more.

Where you can find non-shot expected goals

Unfortunately, it can be a bit difficult to obtain stats for non-shot based expected goals. It has not yet become mainstream like the regular xG model.

But some sites do share stats for NSxG.

The best example is FiveThirtyEight’s Global Soccer Prediction where they do measure non-shot expected goals in the major leagues. You can find the data for free by looking up the matches that have been played.

Infogol does not show NSxG in its tables and general statistics, but the site has brought articles about the concept and measured some matches based on the metrics.

Several Twitter accounts share NSxG so by searching the top football analytics on Twitter, you might come up with some regular data.

A big problem is that many of the sites vary in their estimations. FiveThirtyEight might have drastically different NSxG scores than another site. The reason is that they might have different values when measuring possession, and maybe some data providers value passes more than others. There could be several reasons.

The most serious way is to measure NSxG is by analyzing all matches yourself and putting up maps for how much a certain area is worth when a player is possessing the ball or passing the ball into that area. It will require a lot of work though, and that makes it a bit unrealistic. But if you follow one team closely, you can try making some NSxG maps and measuring that one team’s performance and see how your NSxG estimation matches up with FiveThirtyEight and other major providers.

How to use the data

When you have some NSxG data, you can use it for several things. We will assume you are simply finding the data in the FiveThirtyEights predictions section, or with some other provider.

Firstly, it allows you to look back at a match and see which team “won” based on its non-shot based actions. It is an alternative to regular xG that focuses only on the shot. NSxG fills you in on the rest of the action. This can be useful when analyzing previous bets – so that you can find out if you picked the correct side or not, regardless of what the actual score was.

Then you can use NSxG to compare two different teams. By keeping track of their stats, you can see if there is an edge that the rest of the market is not spotting. For example – in a match between team A and team B, team A has a lot of NSxG since it dominates possession and has many dangerous attacks, while team B allows a lot of possession. In this case, you can quickly and efficiently see that team A is likely to dominate this match, even if they are even in the table or close on regular xG. Most players don’t check up on NSxG, so it is a way to have a small edge.

If you are very serious, you can set up a betting model based on NSxG. It can be used as a factor next to several others. The bookmakers and the syndicates are using normal xG in their models, so if you have your own model, it doesn’t hurt to add in more data. NSxG definitely says a lot about a team’s attacking and defensive skills, so if you have correct data, it is something that can be used to your advantage.

Issues and risks

NSxG is far from a perfect metric.

As mentioned earlier, many data providers disagree massively on their NSxG maps, and that will result in wildly different outcomes when they analyze each match. Even more so than regular xG, where there are also some differences between providers, but NSxG is bringing even more disagreements.

If you base your model around NSxG, even if it is just one factor, you absolutely need to make sure the data is reliable.

Then there is the issue that not a lot of data providers have this metric yet. You do find non-shot expected goals and expected possession goals with some sites, but it is not anywhere near as common as xG. We hope that it becomes more common, but so far it is still a niche area.


Non-shot expected goals are a very interesting metric, and it could rival regular xG with time. We recommend you to research further and to find good providers – FiveThirtyEight will be a great site to get a first impression of NSxG and to check up on the NSxG scores from previous matches.

If done right, non-shot based expected goals can give us better knowledge of how strong a team is offensively and defensively. And that will be useful in betting, especially for those of you who are using models.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top