Early Premier League: When the Table Doesn't Tell the Full Story, xG Becomes a Temporary Compass

On a late-September afternoon, when the Premier League season had just passed its sixth matchweek, I sat in my Manchester office with a stack of notes and a worn black notebook. Beside me, a screen showed the latest league table. One side sat top with a perfect record. Another occupied a Champions League place. On forums, people had already begun calling them genuine candidates for the season.

But in my notebook, on a page devoted to a match played weeks earlier, a small underlined line read: 'This team controls possession well, but most shots come from outside the box. Two goals arrived from set pieces.' I circled it in pencil. It did not match the story the table was telling everyone.

I remember the colour of the grass that day before I remember the score. It was a stadium with a low roof, sunlight slanting through steel beams just enough to make you squint at the scoreboard. In the stands, supporters sang the team's name. On the pitch, players moved in patterns only those who sit long enough notice. I observed that the hosts, despite leading, did not create many genuinely dangerous chances in open play. They scored from a corner and a long-range shot into the top corner.

That was the moment I understood the early-season table needs to be read through a different lens.

Last season, Liverpool topped the table at this stage with five wins from five, and Tottenham sat third. By the end, Liverpool finished fifth, while Tottenham fell to seventeenth, only a few points above the relegation zone. Same starting point, two entirely different endings. The rhythm of a season does not live in goals but in each repeated Saturday, in matches nobody remembers once the campaign closes.

That is why I became interested in a concept that, little more than a decade ago, only a small group of analysts in England used: expected goals, or xG.

What xG is and why it matters at this stage

xG measures chance quality, based on the number and quality of shots a team creates and faces in a match. When a player shoots from the edge of the box in a comfortable stance, that shot's xG value might be 0.08. When a player faces the keeper from five metres, it might reach 0.7 or higher. Add every value together and you get a team's xG for a match.

What makes the metric useful is that it strips out two things the table cannot separate: a player's finishing ability and luck. A team can win three straight games through shots from outside the box, own goals by opponents, or situations where the opposing keeper makes a mistake. The table records three wins. xG records a different picture.

There are three related metrics I track. First, xG, measuring the quality of chances a team creates. Second, xGA, measuring the quality of chances it allows. Third, xGD, the difference between the two, acting as a proxy for how dominant a side has been across a season.

A subtlety I always remind myself of when reading these numbers is individual finishing quality. Elite strikers such as Mohamed Salah of Liverpool or Erling Haaland of Manchester City routinely score more than their xG, because their finishing sits at a different level from the league average. Son Heung-min of Tottenham is another example of a player who has converted chances above the mean across several consecutive seasons. This does not make xG wrong, but it shows the metric must be read alongside an understanding of the individual quality within a squad.

To newcomers, the abbreviations xG, xGA and xGD sound dry. But placed next to the table, they tell another story. A team fourth in the standings but twelfth in xGD may be living on luck. A team fifteenth but sixth in xGD may be underrated. These gaps do not mean results are meaningless. They mean results, early in a season, are a noisy signal that needs context.

One factor that complicates matters is fixture difficulty. A side can top the table after five games because it faced only weak opponents. Another can sit near the bottom because it had to travel to Manchester City, Liverpool and Arsenal in three straight weeks. The table does not distinguish these cases. Metrics like xG do not automatically distinguish them either, but they open a space in which to ask questions.

In recent years, xG has travelled from a specialist concept into mainstream conversation. Broadcasters show xG charts during punditry segments. Major newspapers include xG in post-match analysis. Fans on social media debate whether their club is 'playing well' or merely 'getting lucky' based on these numbers. That popularity brings benefits and risks.

The benefit is that supporters gain another tool to understand a match beyond the scoreline. The risk is that the tool can be misused, turned into an absolute prophecy rather than a relative measure.

The Liverpool and Tottenham case

Return to last season's Liverpool and Tottenham. Liverpool opened with five wins from five, top of the table. Tottenham sat third. Reading the table, one might assume both were enjoying successful seasons. But the story ended in ways few predicted: Liverpool finished fifth, Tottenham seventeenth.

I do not have detailed xG data for either side from that period in my notebook, but I remember what I observed. Liverpool won early games through individual moments of brilliance and a touch of luck in decisive situations. Tottenham collected points through a defence that held firm at key moments, even though their control of matches was not entirely convincing. Both were sides whose results outstripped the quality of chances they created and faced.

The gap between results and process, when it persists across many games, tends to correct itself in the opposite direction. That is the principle of regression to the mean, one of the most reliable laws in sport.

Regression to the mean does not mean every club will land exactly where its xG suggests. It means extreme results, whether positive or negative, tend to soften over time. A team winning five straight through late goals will struggle to sustain that run if the quality of its chances does not change. A team losing three straight yet creating many clear chances is likely to improve its results.

This is why analysts often say the early-season table is a good descriptive metric but a poor predictive one. It describes accurately what has happened. It does not forecast well what will happen, particularly when the sample is small.

Premier League history offers many examples of sides that started strongly and faded, and sides that started slowly and surged. Leicester City in 2026/16 is the reverse case: a team not rated highly in theory that sustained extraordinary form across an entire season. That case shows data is never the whole story. Will, squad spirit and hard-to-quantify human factors can produce results far beyond prediction.

Exceptions do not invalidate data's value. They remind us that data is a tool, not an absolute truth.

How to read the early table sensibly

So how should one read the early table sensibly? I usually apply a three-step process.

Step one is separating results from process. Results are points. Process is how a team creates and prevents chances. The two can diverge, and when they do early in a season, I pay more attention to process. This does not mean results do not matter. It means results alone are not yet sufficient for a conclusion.

Step two is assessing the fixture list. A side top of the table after facing only weak opponents is less convincing than one eighth after travelling to the three hardest stadiums in the league. Early fixtures are often overlooked in debates, yet they are essential to the picture.

Step three is waiting for an adequate sample. I rarely form firm judgements about a team until roughly ten matchweeks have passed. Before that threshold, every conclusion is provisional. After it, trends become clearer and forecasts more reliable.

This process does not guarantee I am always right. It simply helps me avoid the rushed conclusions media and fans often reach after a few rounds.

xG's limits and provider differences

An important aspect is variation between data providers. Not every xG model is identical. Opta, StatsBomb and other vendors can produce slightly different xG values for the same shot, depending on how they define and process data. This does not make xG meaningless, but it means comparing xG across different sources requires care.

In my work, I usually select one data source and use it consistently across a season. This helps me avoid distortions caused by methodological differences.

There is another limit of xG I always keep in mind. The metric measures chance quality based on location, angle and other factors, but it cannot measure everything that constitutes a chance. It does not know a player is carrying a minor injury. It does not know a club has just endured a stressful week in the media. It does not know a manager changed formation in the first half.

These factors cannot be quantified, but they influence results. That is why I always read data alongside what I observe on the pitch and in the press room.

The inverse case: when process beats results

There is a dimension xG debates often overlook, and I consider it the most interesting: the inverse case.

When people discuss xG, they usually focus on sides whose results exceed their xG, the ones 'living on luck'. But the more interesting group is sides whose xG exceeds their results. These are teams playing well, creating chances and controlling matches, yet still losing or drawing for reasons beyond their control.

I remember a specific case in a season I followed closely. A mid-table club began the campaign with a winless run, and pressure on the manager rose quickly. But looking at the data, I saw the side created more quality chances than opponents in most matches. They simply lacked efficiency in front of goal and suffered a touch of misfortune at decisive moments. By mid-season, they began collecting points steadily and climbed to safety.

The reverse also happens often. A side starts strongly through long-range goals and set pieces, but the quality of its open-play chances is low. When luck runs out, results decline. Fans call it a 'form crisis', but in truth it is a team returning to its real level.

The asymmetry in how we read these two cases is striking. When a team's results exceed its xG, we tend to call it 'character' or 'a winning mentality'. When a team's xG exceeds its results, we call it 'inefficient' or 'lacking character'. The same statistical phenomenon, two different interpretations.

I believe the second interpretation is usually more accurate over the long run. A team with good process is more likely to improve its results than one relying on luck. But in the short term, the first interpretation tends to be favoured because it fits the story results are telling.

The media cycle and pressure on managers

This leads to another dimension: pressure on managers. In English football, a slow start can quickly become a media crisis. Pundits begin asking about the manager's future. Fans on social media call for change. Newspapers publish lists of potential replacements.

But if data shows the side is performing better than its results, that pressure can be unjust. The manager is doing his job, but results have not yet reflected it. In such cases, a sacking can be a strategic mistake.

Conversely, a manager whose team is winning through luck may receive more praise than deserved. When luck ends and form dips, he may be held responsible for a situation that is really a return to the true level.

This is one reason clubs increasingly consult data when evaluating managerial performance. They look not only at results but at the process producing them. A decision grounded in both is more likely to be sound than one grounded in either alone.

Pressure is not only on managers. Players face it too. A striker scoring three goals in four opening games may be hailed as a new discovery, but if those goals came from low xG, that praise may be temporary. Another striker who fails to score in four opening games yet carries high xG may be closer to a breakout than people think.

In my notebook, I record these details. They help me track long-arc stories rather than reacting only to what happens in a single week.

Connecting data to human stories

One thing I always try to remember when writing about football: data is a tool for understanding people, not a replacement for them.

Behind every metric is a player who has trained for hours on the pitch to improve his finishing. Behind every trend is a manager who has spent weeks building a tactical system. Behind every gap between results and expectation is a group of people trying to turn what they believe is right into results on the pitch.

When I look at a team's xG, I try to look beyond the numbers. I try to understand why they create chances but do not score. Is their striker short of confidence? Are they choosing suboptimal options at decisive moments? Are they carrying the effects of a key injury?

These questions have no answers in any data table. They require going to the training ground, talking to people inside, and listening to the silences between answers. That is the part of the job I love most.

Signals to track in the coming weeks

As a new season begins and the table takes shape, I remind myself of one thing: what we see in the first ten matchweeks is not the essence of the season, but part of the process of understanding it.

To write a true story, I must stand where nobody else stands. In this case, that is the space between the table and the data. The table tells me what happened. Data tells me what may be happening behind it. Both are necessary, and both have limits.

The signal I will track in the coming weeks is the gap between league position and xGD position. Teams with large gaps between the two are the ones most likely to shift position over time. That is not a firm forecast, but a way to notice what may be overlooked.

I will also track how media reacts to sides whose results exceed their xG. If the narrative around them centres on 'character' and 'winning spirit' while ignoring luck, I know a correction is approaching. Conversely, if a side whose xG exceeds its results is criticised as 'inefficient', I know the opportunity lies on the other side of perception.

And perhaps the most important thing I have learned after years of watching football is humility before the unknown. Data gives us a better picture, but not a complete one. What remains, always, is the human story on the grass, something no metric measures in full.

The press tower is the highest place to look from, but it is not the closest place to understand. I still prefer sitting in the lowest tier of the stand, where I can hear players breathing between halves, and see the colour of the grass before the scoreboard has time to change.