Blog — page 4
How prediction models actually work, what betting markets really price, and how to judge anyone's tips — including ours, whose track record is public.
What Championship matchday 2 told us about defence ratings in August
Three of the division's weakest-rated attacks put eight goals past better-rated defences. What a defence rating is actually worth in August.
Match preview
Flamengo vs Botafogo: the derby column our model does not have
A clássico at the Maracanã, and our model gives Flamengo 60.1%. Do prediction models account for derbies? Ours does not, and here is what that costs.
Match preview
Fulham vs Chelsea: how much should a new manager move a match probability?
Two managerial debuts at Craven Cottage, and a market that disagrees with our model by 17 points. How much should a new manager move a match probability?
Match preview
Köln vs Hoffenheim: when the goals line is surer than the result
Our model cannot separate Köln and Hoffenheim, yet it is confident about goals. When the goals line is surer than the result, the ratings explain why.
Match preview
Liverpool vs Nottingham Forest: what back-to-back wins at one ground are worth
Forest have won their last two visits to Anfield, a first since 1963. What back-to-back wins at one ground are worth to a model, put through a noise test.
Match preview
Tottenham vs Newcastle: when our model backs the away team
Our model backs the away team at Tottenham. Here is how big a rating gap it takes to beat home advantage, and why this particular call is fragile.
Match preview
Bayern vs Stuttgart: what a perfect opening-matchday record is worth
Bayern's Bundesliga opening-matchday record is unbeaten since 2011 and they beat Stuttgart four times last season. Our model still says 67.9%. Here's why.
Match preview
Crystal Palace vs Man City: how much one absence changes a probability
Jeremy Doku is out with a calf injury. We put a number on how much one absence changes a match probability — and on what our model never saw at all.
Model watchNew laws, old data: what a rule change does to a prediction model
Football's 2026/27 time-wasting laws are in force. Here's the honest answer to what they do to a model trained on years of matches played under the old ones.
TransparencyHow accurate are football predictions? The honest ceiling
How accurate are football predictions, really? The math that caps even a flawless model near 50-55% — and why a bigger headline number is a warning sign.
Match preview
Real Madrid vs Real Sociedad: when expected goals and the result disagree
Real Sociedad lost at Betis while creating the better chances. What our model does when expected goals and the result column point opposite ways.
TransparencyCalibration in football predictions: does 70% really mean 70%?
Calibration in football predictions, explained: what a trustworthy 70% looks like, how to read a reliability diagram, and why overconfidence is the norm.
Betting 101How to predict draws in football: the honest math
How to predict draws in football, honestly: why draw probability tops out near 35%, the two levers that actually move it, and how to judge any draw record.
Model watch
Promoted clubs: what our model knows and when it guesses
Hull arrive from a division we cover; Málaga from one we don't. Why promoted clubs are two different problems for a prediction model, in four fixtures this weekend.
Model watch
Atlético 2–0 Málaga: what one result proves about a model–market gap
Our model said 58%, the market said above 70%, and Atlético won 2–0. What one result proves about who was right: almost nothing. Here's what would.
Betting 101How to convert betting odds to implied probability (properly)
Odds are probabilities with a fee hidden inside. How to convert betting odds to implied probability, measure the margin, and strip it out — step by step.
TransparencyHow to verify a football tipster's track record
Screenshots lie, win rates mislead and hot streaks mean nothing. A practical checklist for verifying any football tipster's track record — ours included.
Betting 101Asian handicap explained: lines, pushes and real margins
Whole, half and quarter lines demystified with the real margin-of-victory distribution from 17,495 matches — and how one scoreline grid prices every line.
Match previewAtlético vs Málaga: when the market knows more than our model
Our model gives Atlético 58%; the market says 73%. The gap is Málaga — a promoted club with zero matches in our dataset. An honest look at a model limit.
LeaguesBundesliga predictions: the league where the result tells you least
The Bundesliga leads our data in goals, Over 2.5 and BTTS while its 1X2 split is dead average. That combination decides where a German prediction is worth having.
LeaguesChampionship predictions: the league where the table lies
The Championship has the tightest margins of any competition we model — just 12.2% end by three goals or more — across a 46-game season. Compression is the whole story.
Betting 101Double chance explained: safer odds, same maths
How 1X, X2 and 12 really price up, base rates from 17,495 matches, why the same bet is worth less in England than in Spain, and why safer never means better value.
LeaguesEredivisie predictions: the goals come from mismatches
The Eredivisie is our second-highest scoring league and has the highest blowout rate of any domestic competition we model — 21.6%. Those two facts are the same fact.
Model watch
Our model's five strongest calls of the 22–24 August weekend
Five home wins priced at 67% or better across five leagues — and the honest math on why even these probably won't all land.