GamesNFans logo GamesNFans

Premier League xG 2026/27: Read the Table Like a Pro

Five games in, the Premier League table is misleading. Learn how to use expected goals data to spot overperformers and underperformers in 30 minutes a week.

Premier League 2026/27: What the xG Data Really Tells Us After 5 Games — illustrative featured image
Five games into the 2026/27 Premier League season, the table is already lying to you. That is not a hot take, it is arithmetic. A team can sit fourth on nine points while getting outshot every week, and another can sit fourteenth while dominating chances. By the end of this piece you will know how to read the table the way analysts do, and you will have a repeatable method you can run in about thirty minutes a week. If you have ever wondered how to get started in esports analytics, this is the same skill set, just applied to a different pitch. Here is the deal: expected goals, or xG, is the number that tells you whether a result was earned or stolen. We are going to walk through five concrete steps, name the tools, quote the prices in USD, and tell you exactly what goes wrong at each stage. Our focus is the US, UK and Europe, because that is where most of you watch from and where the data providers actually operate. ## Step 1: Understand what xG is actually measuring xG assigns every shot a probability between 0 and 1 based on distance, angle, assist type, defensive pressure and a dozen other inputs. A tap-in from six yards might be 0.7. A hopeful effort from 30 yards is 0.03. Add up every shot a team takes and you get their expected goals for the match. The mistake almost everyone makes is treating xG as a prediction. It is not. It is a description of chance quality. A team with 2.4 xG and zero goals did not get unlucky in a cosmic sense. They created enough to score twice and failed to convert. That gap is the story. **What goes wrong here:** You compare raw xG totals across teams and conclude one attack is better than another. Wrong. You need xG per shot and xG per possession to see style. Manchester City racking up 2.1 xG from 22 shots is a different animal from Brentford hitting 2.1 from seven. ## Step 2: Pick your data source and pay for it Free options exist and they are fine for a casual look. Understat covers the big five European leagues with shot-level data and a clean interface, and it costs nothing. FBref, run by Sports Reference, gives you xG plus progressive passes, pressures and a hundred other columns, also free. Paid tools change the game. Opta-powered feeds through Stats Perform are enterprise priced, often north of $5,000 a year, and that is not for you. The realistic middle ground is a subscription to a consumer analytics site. The Athletic runs around $72 a year in the US and £60 in the UK. Between the Posts and similar newsletters run $8 to $12 a month. | Tool | Cost (USD) | Best for | |---|---|---| | Understat | Free | Quick league-wide xG table | | FBref | Free | Deep per-player and per-team stats | | The Athletic | ~$72/year | Written analysis alongside numbers | | Specialist newsletters | $8-12/month | Weekly team-by-team breakdowns | **What goes wrong here:** You pay for three subscriptions, drown in dashboards, and stop reading by October. Pick one free source and one paid source. That is it. ### Step 3: Build the xG table yourself This is the step that separates people who understand the league from people who repeat what pundits say. Take every team's xG for and xG against across the five games. Subtract. That number is your expected goal difference. Now sort by it. You will get a table that looks nothing like the actual Premier League table. That divergence is the entire point of premier league table analysis done properly. Teams with a big positive gap between actual points and expected points are riding luck. Teams with the reverse are better than their record suggests. **What goes wrong here:** You use a five-game sample and treat it as gospel. Five games is noise plus a little signal. The method matters more than the conclusion at this stage. ## Step 4: Separate finishing skill from variance Some players really are better finishers than the model expects. Harry Kane has outperformed his xG for a decade. That is skill. But most overperformance over five games is variance, and it reverts. The rule of thumb: if a team is outperforming xG by more than 0.3 goals per game after five matches, be suspicious. If a striker has scored six from 3.2 xG, enjoy it, but do not build your mental model of the team around it continuing. **What goes wrong here:** You decide a hot start is a new identity. It usually is not. Check last season's overperformance numbers before you commit. ## Step 5: Turn the numbers into a weekly habit Thirty minutes, once a week, ideally Monday morning after the weekend fixtures. Pull the updated xG table, note the three biggest divergences from the actual premier league xg table, and read one analytical piece on the team that surprises you most. That is the whole routine. Do it for ten weeks and you will spot the regression before the pundits do. This is also, incidentally, the fastest way to learn how to get started in [esports analysis](/tech/blog/meta-s-muse-ai-vs-chatgpt-grok-claude-which-chatbot-wins-in-2026), because the workflow is identical: collect event data, compare to expectation, find the gap. **What goes wrong here:** You skip weeks. The method only works if you build the comparison over time. Set a calendar reminder. ## What we recommend If you want one free tool: **Understat**. Clean, fast, covers everything you need. If you want one paid subscription: **The Athletic** at roughly $72 a year. The writing is good and the numbers are contextualised, which matters more than raw data at this stage. If you want to go deeper and eventually move into esports analytics work: **FBref** plus a spreadsheet. Learn to pull the data yourself, because that skill transfers directly to how to get started in esports scouting and team analysis roles. If you want none of these yet: that is a legitimate answer. Watch ten more matches, then come back. Five games is a small sample and there is no shame in waiting until the picture sharpens. ## FAQ **Is xG a better predictor than the actual table after five games?** Yes, marginally. Expected goals stabilises faster than results because it measures process, not outcome. But five games is still a small sample. Treat it as a hint, not a verdict. **Where can I find premier league xg data for free?** Understat and FBref both publish it at no cost. Understat is easier to read at a glance; FBref goes deeper if you want per-player detail. **Can I use these skills to get into esports analytics?** Absolutely. The core loop (collect event data, model expected outcomes, find the gap) is the same in Counter-Strike, League of Legends and Dota 2. Build the habit on football, then port it.

Frequently asked questions

Is xG a better predictor than the actual table after five games?

Yes, marginally. Expected goals stabilises faster than results because it measures process, not outcome. But five games is still a small sample. Treat it as a hint, not a verdict.

Where can I find premier league xg data for free?

Understat and FBref both publish it at no cost. Understat is easier to read at a glance; FBref goes deeper if you want per-player detail.

Can I use these skills to get into esports analytics?

Absolutely. The core loop (collect event data, model expected outcomes, find the gap) is the same in Counter-Strike, League of Legends and Dota 2. Build the habit on football, then port it.

Related articles