Premier League predictions: the league that refuses to draw
Published · Updated 2026-08-29
The Premier League is the most-watched, most-bet and most-analysed football competition on earth, which makes it the hardest league to find an edge in and the easiest one to write nonsense about. So here is the one structural fact our data gives it, and what follows from it.
The lowest draw rate on our board
Across 1,911 Premier League matches in our dataset:
- Home wins 44.3%, draws 23.8%, away wins 31.9%. That draw rate is the lowest of every competition we cover — a full 3.7 points below Brazil’s 27.5% and 3.3 below Serie A’s 27.1%.
- 2.93 goals per match, 56.7% over 2.5, both teams scoring 55.3%. Fast, behind only the Bundesliga and Eredivisie among the leagues we model.
- 18.2% of matches end with a three-goal margin or more, against 12.2% in the Championship one division below.
England’s top flight decides matches. That is the base rate every serious prediction has to respect: a model handing out 30% draw probabilities in a league that draws 23.8% of the time is miscalibrated before kick-off, and no amount of good team ratings rescues it.
What a 23.8% draw rate does to double chance
The draw rate is not just a number in the middle column; it sets the price of every market built on top of the three outcomes.
In the Premier League, home-or-draw covers 68.1% of matches, draw-or-away 55.7%, and home-or-away — the “no draw” bet — 76.2%. Compare that last figure with La Liga’s 73.8% or Brazil’s 72.5% and the gap is where the money is: the same double chance selection is worth materially less in England than in Spain, because the outcome it is insuring against happens less often.
The practical version: a double chance covering the draw buys you less protection here than the identical bet elsewhere, and the market knows it. If you are carrying a mental price for these markets built on continental football, England will look like poor value everywhere, and mostly it will be your reference point that is wrong.
Home advantage, measured rather than assumed
Home advantage in England is real and unremarkable: home sides win 44.3% and away sides 31.9%, a gap of 12.3 points. That sits close to the middle of our range, which runs from Ligue 1’s 11.3 up to Brazil’s 21.1.
The number matters because “home advantage” is usually quoted as a mood rather than a quantity. Twelve points of spread is worth roughly a quarter of a goal in a scoring model — enough to swing a genuinely even fixture into a favourite, and nowhere near enough to rescue a weaker side against a stronger one. When a preview leans on a home crowd to explain a prediction, that is the size of the thing being invoked.
It also decays in a specific way our model has to handle: the home term is fitted at league level from thousands of matches, so it is one of the most stable parameters in the fit, while the team ratings around it move week to week. A shift in a published probability almost never comes from home advantage. It comes from the ratings.
Three promoted clubs, every August
Every season three teams arrive from the Championship with no top-flight history in the fit — and the two divisions are not close statistically. The Championship averages 2.54 goals a match to the Premier League’s 2.93, and produces a three-goal margin in 12.2% of matches against 18.2%.
A promoted club therefore brings a rating estimated in a slower, more compressed competition into a faster, more spread-out one. Our model starts those sides near the league average, regularised toward the mean, and lets them earn a true rating match by match. That is honest about the uncertainty, and it is why our early-season probabilities involving promoted clubs carry less conviction than their crispness suggests — a limitation we set out in full in what our model knows about promoted clubs.
The sharpest prices in football
Premier League closing odds are the most efficient in the sport. Enormous liquidity grinds out mispricing faster than anywhere else, and any model claiming a persistent edge here is making an extraordinary claim.
We publish our probabilities next to the market’s on every match page precisely so the disagreements are visible rather than asserted, and we are direct about how hard this particular market is to beat in value bets and closing odds. The useful posture is not “we beat the market” but “here is where we differ, judge the record”.
Where the accuracy actually lives
Every Premier League fixture gets 1X2, Over/Under 2.5 and both-teams-to-score probabilities, refreshed through the day as results and odds move, plus form, head-to-head and the model’s verdict, on the Premier League hub.
We do not quote a win rate in a blog post. Static text cannot be audited, and a figure written today would be stale within weeks. Every prediction we make is settled after full time and published per league and per market on the track record, with the misses given equal billing — and what “accurate” even means for a probability is worth reading about in how accurate are football predictions before judging any forecaster, us included.