Paris Saint-Germain had 27 shots against Liverpool in March 2025. Liverpool managed two. The final score was 1-0 to the visitors, courtesy of an 87th-minute goal from Harvey Elliott. Anyone who had spent the evening watching the numbers rather than the match might have wondered how on earth Liverpool had won.
The expected goals figures made the result even more striking. PSG finished with 1.82 xG, compared with Liverpool’s 0.27. Alisson made nine saves, the French champions dominated possession, and practically every attacking statistic pointed in their favour. Yet the only number that mattered to the result was on the scoreboard.
It’s an example worth remembering when checking football tips for today. Statistics can help identify strong performances, questionable favourites and teams whose recent results hide underlying problems. But a convincing set of numbers can also encourage more confidence than the evidence deserves. Understanding the difference is particularly important when money is involved.
Football has become remarkably good at measuring what happens on the pitch. The harder question is whether supporters, pundits and punters have become equally good at interpreting those measurements.
The problem with taking statistics at face value
The PSG-Liverpool match illustrates why raw numbers need an explanation. According to the Premier League’s analysis of the game, PSG’s 27 attempts produced 1.82 expected goals. That works out at roughly 0.067 xG per shot, although the quality of individual chances varied considerably.
Liverpool defended deep for much of the second half, restricting space around their penalty area and forcing several attempts from distance. PSG certainly created enough opportunities to win, but the volume of shots made their dominance look even more overwhelming than the quality of those attempts suggested.
This is where football statistics can become misleading. A team recording 20 shots sounds considerably more dangerous than one managing eight. But if most of those attempts are speculative efforts from outside the box, the comparison tells us very little about which side created better opportunities.
Possession figures are another familiar trap. A side that spends 65% of the match with the ball might be controlling proceedings. It might also be passing harmlessly across midfield while its opponent protects a lead and waits for a counter-attack.
The distinction becomes particularly relevant when analysing results after the event. A team that loses despite dominating possession and shots may have been unfortunate. Alternatively, its opponent may have deliberately allowed those numbers to accumulate in areas where they posed relatively little danger.
You need to know how the statistics were produced before deciding what they mean.
Expected goals are useful, but they don’t settle arguments
Expected goals, usually shortened to xG, has become one of football’s most frequently quoted statistics. It estimates the probability of a shot resulting in a goal by comparing it with similar chances from historical data.
According to Opta’s explanation of its xG model, the calculation draws on nearly one million recorded shots and considers factors including distance, shooting angle, defensive pressure and goalkeeper positioning.
A chance worth 0.20 xG would therefore be scored approximately 20% of the time across comparable situations. That doesn’t mean the player should score it, or that five similar chances will necessarily produce one goal in a particular match.
Over a substantial number of games, xG can be extremely helpful. It offers a clearer picture of attacking performance than goals alone, especially when a team has been creating chances without converting them.
The trouble starts when a useful long-term measurement becomes a verdict on one afternoon’s football.
Consider a striker who has scored twice from chances worth a combined 0.6 xG. He might be enjoying an exceptional finishing spell, benefiting from a little good fortune, or both. You cannot establish which explanation is correct from two goals.
The reverse is equally true. A forward who has scored once from 3.5 xG isn’t automatically due a hat-trick next weekend. Previous missed chances don’t increase the probability of his next shot going in.
There are limitations to the metric itself. Standard xG is based on shots that actually happen. It cannot fully capture an excellent attacking move that ends with a poor final pass, or a dangerous run that never leads to an attempt.
Different data providers also use different models, which explains why the same match can produce slightly different xG figures across platforms. Those differences don’t necessarily make one provider unreliable, but they are another reason to avoid treating a number displayed to two decimal places as an unquestionable fact.
Recent form can be a particularly convincing illusion
Football fans love a form table. Five consecutive wins look impressive, while four defeats suggest a club in trouble. Unfortunately, those sequences are often stripped of the details that explain them.
Imagine two Premier League teams preparing to meet. The home side has won four of its previous five fixtures, mostly against struggling opponents. The visitors have taken only five points over the same period, having faced several of the league’s strongest teams.
Looking exclusively at recent results, the home side appears the obvious favourite. Examine the opposition, chances created and performances within those matches, and the difference may become much smaller.
The same applies to defensive records. Three consecutive clean sheets sound encouraging until you discover that the opposition created several excellent chances and repeatedly failed to finish them.
None of this means results should be ignored. Winning matches is ultimately the point of football. But form becomes more meaningful when you account for the quality of the opposition and the circumstances surrounding each performance.
Team news can undermine an otherwise convincing statistical case, too. A side’s attacking figures across ten matches might look excellent, but those numbers become less relevant if its leading striker and most creative midfielder are unavailable. As Today News has previously explored in its coverage of late squad updates, information arriving shortly before kick-off can change the assumptions behind an earlier assessment.
This is why an analysis based on last month’s results should never be treated as complete simply because the spreadsheet looks impressive.
Bookmakers don’t pay out for statistical superiority
For anyone placing a bet, there is another distinction that matters more than most: identifying the likeliest winner and finding a worthwhile price are separate exercises.
Consider a hypothetical Premier League fixture in which the home team has stronger recent xG figures, more shots on target and a better league position. You decide that it has a 35% chance of winning.
A bookmaker offers odds of 1.5/1. A £10 stake would return £25, including £15 profit, if the team wins.
Those odds imply a break-even probability of 40%. If your 35% assessment is accurate, the bet has a negative expected return of £1.25 per £10 staked over many equivalent bets. The statistics might support the selection, but the available price doesn’t.
This is also why an odds-on favourite isn’t automatically a sensible bet. The favourite may win more often than it loses, yet repeatedly backing it at prices shorter than its true chances justify can still produce losses.
Bookmakers build a margin into their markets, and their prices already reflect a considerable amount of public information. Spotting that one club has created more chances than another is unlikely to reveal something entirely overlooked by the market.
There is also the uncomfortable matter of variance. Even a genuinely favourable bet can lose, sometimes repeatedly. Football contains too many unpredictable events for any statistical approach to eliminate that risk.
The best analysis begins with a better question
Statistics have made football discussion more informed. Supporters can challenge lazy assumptions, recognise promising performances hidden behind defeats and examine trends that would previously have gone unnoticed.
But the growing availability of data has also encouraged a particular kind of overconfidence. A figure looks authoritative simply because it is precise. A graph that rises steadily feels like evidence of inevitable progress. Neither deserves automatic trust.
Before relying on a team’s attacking numbers, consider who they played and whether the performances were consistent. When looking at expected goals, examine the chances behind the total. If recent results appear unusually strong or weak, look for explanations rather than immediately predicting a reversal.
Most importantly, separate what the statistics say about previous matches from what you believe they imply about the next one. The second judgement requires additional information, and there will always be uncertainty.
For punters, that uncertainty should influence both the decision to bet and the size of the stake. Betting is for adults aged 18 and over, and the sensible limit remains what you can afford to lose.
Liverpool’s victory in Paris didn’t prove that football statistics were useless. PSG really did create more opportunities, and Alisson really did produce an extraordinary performance. The figures helped explain why the result was so surprising.
The mistake would have been expecting those figures to decide the score in advance.
The next time a statistic appears to settle a football argument, ask what it actually measures. That’s usually worth more than another decimal place.
David Prior
David Prior is the editor of Today News, responsible for the overall editorial strategy. He is an NCTJ-qualified journalist with over 20 years’ experience, and is also editor of the award-winning hyperlocal news title Altrincham Today. His LinkedIn profile is here.












































































