Before anyone sells you a “sure tip”
We went looking for a measurable answer to that question, not an opinion. We built seven forecasting models, then checked what each of them was worth across 11 seasons and 50,000+ matches. You will find the answer further down; until then, this much: it is not the one the tipster sites promise. And everything the research left behind is yours to use too: daily AI match analysis, probability estimates from seven models, pro odds, and the same test engine we measured with.
Modules I and VIII, and the chapter “The mathematics of luck”, are free to read without registration.
Important notice: This is an educational and data-analysis platform. The information published does not constitute sports-betting or financial advice. Predictions do not guarantee a result. The outcome of sporting events is uncertain; every decision is your own responsibility. Gambling is prohibited for under-18s.
Daily match analysis
The machinery outlived the research: our seven models compute the probabilities daily for the matches coming up. Not for all of them, only for those with enough reliable data. Fewer matches analysed thoroughly beats many analysed superficially. And what we once computed, we never rewrite afterwards.
For every analysed match we assemble a detailed picture: the teams’ form, the latest news, the weather expected at kick-off, and a summary of bookmaker prices across all markets (goals, corners, handicaps), together with how the odds moved. The short AI summary is built from the same material.
The model does not give a tip but a reference point: it shows where it sees something different from the market. And on the tip game’s public leaderboard you also see what that difference is worth in practice.
Probability estimate, not a tip. The decision and responsibility are yours.
The full analysis system
You get daily AI analysis for the matches our data covers, plus every relevant detail about the match and the probability estimates of our built-in models. Exactly the tools we used to look for the answer to the big question. We have been collecting the data since 2015, across 11 seasons and 142 leagues, and we re-check it retroactively, so every claim we make can be measured again.
A short summary of the day’s data-covered matches: form, latest news, absentees, expected weather and the market picture in one place. Referenced to numbers throughout, with no tips.
The full picture on one page: season statistics, standings, form and head-to-head record, absentees, coach and key players, top scorers, event log, latest news, and alongside them the odds of every market.
Seven built-in models (Elo, Poisson, Monte Carlo, machine learning and custom weightings) estimate the probabilities for the covered matches, together with the models’ consensus. You see at once where the model differs from the market.
We show how the sharpest bookmakers price a match, after their own margin is stripped out, and where the odds moved sharply. Module III’s bookmaker world in live numbers.
You record the bet before kick-off, and after that it cannot be deleted. That is how it comes out how you really perform: how much of your result is skill and how much is luck. Each month you also get an AI review of it, so there is no room for self-deception.
We measure your own estimates by the same yardstick as our models: you learn not only whether you won, but how well you judged the probabilities. Module VI in practice.
What would have happened if you had followed one strategy all along? You can replay it on past matches, filtering by league, odds range and how large a difference the model saw. With no after-the-fact tidying.
You can test our built-in models with your own settings: league, season, period, minimum history. You can check what we claim in the course, with the same engine we measured with.
The course formulas at a click, with real odds: how much the bookmaker takes, what a bet is worth long term, what stake is sensible. The point is not the arithmetic, it is understanding what comes out of it.
Sometimes two bookmakers think so differently about the same match that it is interesting in itself. We watch and document these rare, quickly closing contradictions.
Every technical term (odds, value, deviation, calibration) explained in plain language. Behind it, the full research background, with methodology and sources.
Except that these numbers are mute on their own: they say something only to those who know what they are looking at. First you need to grasp how betting works, and that is exactly why the course exists.
The course
Nine modules, 47 chapters, in plain language, with worked examples and sources. A test waits at the end of every chapter, and it generates new tasks each time, so you can practise as often as you like. Modules close with an exam, and you can follow your progress all the way.
The language and mathematics of sports betting: how a bookmaker works, what odds mean, and how to calculate with them.
Long term, only one thing counts: value. What EV and CLV mean, and what market efficiency says about your chances of winning.
Soft and sharp bookmakers, limiting, exchanges, bonuses and arbitrage: the betting industry’s machinery from the inside.
The heart of the course: the mathematics of luck. Once you understand it, it also becomes clear why the winner quits and why the loser thinks they are a genius.
Bankroll, stake sizing, the Kelly criterion, and the metrics that reveal where you really stand.
Poisson, Elo, xG, calibration and machine learning: how a probability model is built, and where backtesting tends to go wrong.
Cognitive biases, tilt, the tricks of the tip sellers and the reality of the professional world, and finally responsible gaming.
The LabOfMetrics handbook: models and AI analysis, market-analysis tools, bet log, simulator and strategy tester. We show which one to reach for when, and how to read it.
This is where it all comes together: record-keeping, taxation, and the final project, which is building and testing a strategy of your own.
How does it work?
1.
Work through it chapter by chapter: every concept comes with a worked example, and a test waits at the end of every chapter, always with new tasks. The curriculum builds from odds arithmetic up to the mathematics of luck.
2.
What you read in a chapter, you can try out on real matches and real odds: calculators, bet simulator, strategy tester, and with the very engine we use to measure our own models.
3.
The bet log shows how much of your results was skill and how much was luck, and each month you get an AI review of it. It is not us telling you where you stand, it is your own data.
The tip game is free: with virtual points and no real stake, you can pit yourself against the models and the other players.
What happened when we tried?
Before we started teaching, we measured what the tips industry keeps quiet about. We built seven forecasting models and tested each one across 11 seasons and 50,000+ matches: a model could only ever see what was known before kick-off, so it worked exactly as it would have in real time.
The result was unambiguous: bookmaker prices turned out to be more accurate than every one of our models. In plain terms, from public data (form, relative strength, rest days) nobody predicts better than the market itself over the long run. Anyone selling a “sure tip” either never measured it, or knows and does not say.
Our main finding: from public data the market practically cannot be beaten. That insight became a course, not another tip package.
So we do not hand out tips. We hand over the knowledge that lets you read the market: what an odds price reveals, how much the bookmaker takes, how large luck’s real role is, and how to measure your own results as rigorously as we measure our models.
Who ran this study? See the background and the publication →
The measurement ran on 50,529 matches, and every prediction was built solely on data known before kick-off (Brier score: the lower, the more accurate). The market won, and we publish that too.
How we measure
The model can only see what was knowable before the match. What it predicts today, we do not rewrite tomorrow.
A bet recorded in the log cannot be deleted after kick-off, and every model prediction stays, the bad ones too. No after-the-fact tidying.
If a method does not work, we write that up too. It is no failure that the market is hard to beat: that is precisely our main research finding.
Planned pricing
or €150/yr (2 months free)
The full course with tests, exams and progress tracking, plus the glossary and tip-game participation.
or €390/yr (2 months free)
Everything in Learner, and on top of it the full analyst toolkit: daily match analyses with the models’ probability estimates, bet log with a monthly AI review, simulator, pro odds, calculators.
Frequently asked questions
No. We measured it on 50,000+ matches: from public data nobody beats bookmaker prices over the long run, and neither do we. Anyone promising a “sure tip” or guaranteed profit either never measured it or is not telling the truth. The course explains exactly why that is.
Most of the methods you find online (doubling up, “banker” slips, betting on streaks) lose money over the long run. You do not have to take our word for it: in our strategy tester you can try any rule on real past matches, with no after-the-fact tidying.
Form, absentees, latest news, expected weather, and above all how the odds move. Our daily match analyses gather exactly this for the matches our data covers, together with the probability estimates of seven forecasting models.
Measure it, do not believe it: the course teaches you how to check a tipster’s past results in a few minutes, and why almost every track record offered for sale looks better than it really was.
All 9 chapters of Module I are readable free of charge, without registration: odds, margin, probability, everything worth knowing before a first bet. After registering you can use the whole platform free of charge during the open test period, no card required.