How many points do you need to win the Premier League?
By FootInsights · Published · 6 min read
Ask how many points to win the Premier League and you will be handed a round number — ninety, usually, sometimes ninety-five. It is a reasonable summary and a poor answer, because it describes something that has already happened rather than something a club can aim at. Across the twenty-team, thirty-eight-game era, winning totals have run from the mid-seventies to a hundred. That is a spread of roughly a quarter of a season, and no club chose which end of it they landed on.
The honest version of the question is a probability. For any final points total, what is the chance it proves to be enough? That curve is more useful than a threshold, and — unlike the folklore number — it explains itself.
How many points to win the Premier League: you chase a rival, not a total
A title is a ranking outcome. Nothing in the competition rules mentions a points line. The rules say the club at the top of the table wins, and whether 88 points puts a side first or third depends entirely on what the clubs around them did over the same thirty-eight weekends.
So the bar is not a property of the champion. It is a property of the field. Two things are uncertain at once: your own final total, and the total you have to beat. The probability you actually want is the chance the first lands above the second — and the spread of that difference is wider than the spread of either piece alone.
This is the same structure as the question at the other end of the table, where the safety line is set by rivals rather than by rule. The mechanics differ in one important way, and that difference is the whole story of why title totals climbed.
The bar you must clear is a maximum, not an average
Survival asks you to finish above the third-worst club. A title asks you to finish above the best one. Those are very different statistical objects.
Take a hypothetical division with three credible challengers besides you. Suppose each is genuinely strong and each would, on average, finish on 82 points, with a standard deviation of 7 points around that from ordinary match-to-match variance.
The bar is not 82. The bar is the highest of those three totals, and the maximum of several random draws sits systematically above any single one of their averages. For three independent draws from a normal distribution, the expected maximum lands about 0.85 × 7 ≈ 6 points above the individual mean — so the bar averages roughly 88 points, not 82.
Now add two more challengers of the same quality. The expected maximum of five draws sits about 1.16 × 7 ≈ 8 points above the mean, pushing the typical bar to roughly 90.
Nothing about any individual club changed in that second scenario. No one got better. The number rose purely because there were more chances for someone to have an excellent year. That is the cleanest available explanation for why the points required to win a title drift upward as a league’s top tier gets deeper — and it is why the answer is a moving target rather than a constant.
Building your own side’s distribution
The other half of the calculation is your own distribution, and a model builds it from individual matches rather than from a season-level guess.
Take a hypothetical title contender whose per-match probabilities, blended across home and away fixtures, are 68% win, 18% draw, 14% defeat. We are stipulating these — there is no model error anywhere in this example.
Expected points per match are 0.68 × 3 + 0.18 × 1 = 2.22. Over 38 matches that is an expectation of 84.4 points.
The variance matters more than the expectation. For one match, E[X²] = 0.68 × 9 + 0.18 × 1 = 6.30, so the variance is 6.30 − 2.22² ≈ 1.37. Across 38 independent matches the standard deviation of the season total is √(38 × 1.37) ≈ 7.2 points.
A 95% range therefore runs from about 70 to 99 points — for a club whose strength we fixed by assumption and never allowed to change. The same side, replaying the same season, can come up short of ninety or set a record, and the difference is chance rather than merit.
That range also lets you answer the original question properly. The probability this contender reaches 90 points is the chance a draw from a distribution centred on 84.4 with a spread of 7.2 clears that mark — roughly 24%. Not impossible, not likely, and nothing like the certainty a single headline number implies.
Why a strong season and a title are different things
Put the two halves together and an uncomfortable result falls out. Our contender expects 84.4 points. The bar in the five-challenger league averages about 90. A club can play to its full ability, bank a total that would have won plenty of past titles, and still finish second — not because it underperformed, but because the field contained more opportunities for an outlier than it contained titles.
This is also why chasing a fixed points target is poor reasoning during a season. Ninety points is not a goal; it is a percentile of a distribution that other clubs are still writing. A model that runs the whole fixture list forward — the approach behind season-long simulations — updates both halves at once: your remaining schedule and everyone else’s.
Reading a pace instead of a target
The practical unit is points per game, because it converts cleanly in both directions and does not care how many matches have been played.
Points per game × 38 = final total. A pace of 2.37 projects to 90 points; 2.11 projects to 80; 1.84 projects to 70. A club running at 2.0 needs to lift to roughly 2.6 over a remaining half-season to reach ninety — which is a far more concrete statement than “they need to win most of their games,” and it exposes immediately when a target has become arithmetically unrealistic rather than merely difficult.
Pace is also the right input for a projection because it carries the variance with it. Thirty-eight matches of noise is ±7 points; nineteen matches of noise is about ±5. The narrowing of that band as a season progresses is most of what makes a title race feel decided, long before the points gap alone would justify it.
The two numbers a title race actually has
The useful answer to how many points win a title is a distribution with two moving parts: what you are likely to accumulate, and how high the best rival’s season is likely to reach. A single number collapses both into folklore.
That is the same discipline behind every published probability on this site — a forecast is only meaningful if it states its uncertainty and is then checked against what happened. Our track record keeps that check public, and the same league-level reasoning sits behind the model output on the Premier League page.
When someone quotes you a points target, the right follow-up is not “is that right?” but “how often is it right?” The gap between those two questions is where most confident football statements quietly live.