About me

Sport has interested me all my life, but purely from the analyst side. What caught me was never who wins, but why. I deliberately chose computer science. I earned my bachelor degree with distinction at Eszterházy Károly Catholic University, and I am currently a final-year student on the software technology master programme (MSc) at ELTE, so that I have the most current toolkit for deep, data-driven modelling of sporting events. The more confident I grew at programming, the more I turned towards research.

One question would not leave me alone: can the betting market be beaten, given a good enough model? For a long time I believed it could, and I went looking for a method of my own. My thesis is where I measured exactly that. Six forecasting models and five staking strategies run across 539 real matches, in 300,000 simulations in total. One pairing was consistently profitable on the sample examined, and that result became a peer-reviewed international paper. My work has since also been cited in an international scientific book chapter.

The paper also states its own limits, though, and they are serious. Three months of matches is too little to conclude from, the simulation always assumed the best available odds, and I never tried a single result with real money. So I did not drop the subject, I widened the study: 11 seasons, more than 50,000 matches, walk-forward evaluation, seven structurally different methods. On that sample none of the routes covered the bookmaker margin over the long run. I publish that fact too, numbers and failures included, on the research page.

The subject has not let go of me since. I have given talks in English on these same questions at international conferences, I have further papers under review, and as a master student at ELTE it is this work I am preparing for the 2027 national round with. My current research, which took first prize at the student research conference (TDK), looks specifically at black-box testing of the risk-management and account-limitation algorithms bookmakers use.

LabOfMetrics grew out of exactly that work. Not because I have a winning system, but because the search itself built a toolkit that carries real value on its own: the data, the probability estimates of the models, the back-measurement and a method you can use to look at your own results transparently. That is what I wanted to make available to everyone. The other reason is far more personal. This industry is dangerous and capable of ruining families. If someone loses less after seeing the measurements, or stops playing altogether, that is a better outcome for me than a subscription sold.

During the open test period every feature is free; the plan is that the site will later run on paid plans. The question is fair: if I do not want people to lose money, why would it cost anything? The first half of the answer is that what prevents the most loss is free now and, per the plan, will stay free later. The measurement showing that no durable edge can be demonstrated from public data is readable by anyone. Without registration the course offers the whole of Module I and Module VIII, plus five further chapters: the mathematics of luck, cognitive biases, the tipster industry, responsible gambling and taxation. That is 19 chapters out of 47, including the three things whose ignorance costs the most: odds, margin and the mathematics of luck. Whoever reads only that and never pays me anything is already better off than before.

The second half: somebody funds a free product too. In this industry that somebody is almost always the bookmaker, through affiliate commissions, and that money arrives because the visitor started betting. A site funded that way is interested in you playing more even while it urges caution. The planned subscription is the price of avoiding that: you will be the customer, not the product, and I can calmly write that you should stop. If I lived off ads, that sentence would cost me money.

The rest is simply cost. The data feed, the servers and running the AI analyses cost money every month; during the open test period I cover that out of my own pocket. Per the plan, what will be paid for later is what takes work off your shoulders: the full course and the analytics toolkit.