The method by a team is simple in winning the bet slot. They start by assuming that the bookmakers themselves are good at setting odds and that the prizes they offer are an accurate reflection of the true odds of winning, drawing or losing, plus their own margin.

In that case, a good measure of these odds is a simple average of the odds offered by all bookmakers, a kind of wisdom of the crowd. This gives the average odds, which are a remarkably accurate reflection of the true odds, according to the researchers.

Then it is easy to analyse all the probabilities that are offered and find the outliers. Next, Kaunitz and his colleagues calculate how favourable the outlier probabilities are. If they are good enough, then the gamble should pay off, at least in the long run.

And that is exactly what they have done. They created a web crawler that compiled the odds offered by online betting companies on soccer matches from around the world. They calculated the average odds, found any outliers, and then calculated whether or not a bet would favour them.

Before betting real money, the researchers tested the idea on a 10-year history of data on the closing odds and outcomes of 479,440 soccer games played between 2005 and 2015. This simulation generated pay-outs 44% of the time and returned a return of 3.5% over the 10-year period. " For an imaginary amount of 50 euros per bet, the profits generated would be 98,865 euros in a total of 56,435 bets," they say.

An important question is whether this result could have been pure chance. Could they just have been lucky? The team then compared their results to 2,000 simulations in which they placed random bets on the same games. In that case, the bets generated profits 39% of the time with a return of -3.2%, which is equivalent to a loss of 93,000 euros. That allowed the team to calculate the probability that their first result was a fluke. The research details: "The probability of obtaining a return greater than or equal to 98,865 dollars euros in 56,435 bets using a random betting strategy is less than one in 1,000 million."

So Kaunitz and his colleagues began to hope that their method would work in the real world, but there was a problem. Normally bettors cannot always bet on closing odds, which can vary significantly in the run-up to a match. So, the team decided to create a simulation of this situation. The article describes: "We decided to run a more realistic simulation where we place bets on odds available from one to five hours before the start of each match."

How odds vary in the run-up to matches is not publicly available, so the team created a bot that collected these odds from betting websites around the world from September 2015 to the end of February 2015. 2016. They then tested their approach on the set of this information.
The results were even better. His bets generated payouts 47.6% of the time with a return of 9.9%. The team celebrates: " If each bet placed was €50, our strategy would have generated €34,932 in profit on a total of 6,994 bets."

Interestingly, a random betting strategy using the same data yielded a return of 0.2% and a profit of €825. This could be the result of intense competition between online betting companies, which sometimes offer more favourable odds to entice bettors on a friendly loss-leader policy.