If you search for the scientific studies on the sports betting or gambling you will soon discover that the general scientific consensus is that the sports betting is a fool's errand. Here's a story about the study that proved them all wrong.

If you are interested in the math/stats/machine learning and betting, here's an excellent resource that you can use to search for and read the scientific papers: arxiv.org 

It's been some time since I've got an idea that I can use machine learning in order to beat the bookies. At the moment, the jury is still out there on whether or not I will be able to actually do it.

All the studies I've read have come to the similar conclusion - the market is efficient. That is, the prediction model has failed to find and exploit an inefficiency in the market and thus return a profit.

Sure, some models did better than the others, showing the profitability during one season, then failing the next... I found no consistently profitable model.

And then, I found a study by Lisandro Kaunitz, Shenjun Zhong,  and Javier Kreiner.

Their idea has been brilliantly simple. The bookies are employing a lot of data scientists, mathematicians, statisticians, tipsters.... in order to create as accurate prediction models as possible and the best part of it is that they make their results available to the general public!

Let's use the bookies' numbers in order to find the market inefficiencies (+EV bets)!

Yes, the bookies do publish the odds and if you divide 1 by the odds, you'll get the probability of the event occurring according to the bookies. 

For example, home win @1.50 means that the home team has a 1/1.50 probability of winning - that is 66.66% chance.Yes, it's a bit more complicated with the vig and trying to balance the book, but that's the gist of it.

So they started to work. They gathered the data for the 479440 games across the 818 leagues around the world for the period from 2005 to 2015. By applying their betting strategy using the closing odds they've got 44.4% hit rate and the +3.5% return.

Being the scientists, they've decided to do some more tests. Each one has been successful. 

Finally, the time has come to take the bookies to the cleaners.

They've opened the online accounts, loaded the funds, and started to place the bets. During the period of the five months placing $50 bets, they've got 47.00% hit rate and the 8.50% return.

I won't blame you if you think that they are now retired millionaires, cruising the Caribbean, and enjoying their wealth. Unfortunately, their study found out one more thing:

                                 The bookies are dirty bastards. The lowest of the low!

Yes, their accounts started to get limited and they had to suspend their betting. And there you have it, a totally disappointing conclusion to this very interesting study.

P.S - I recommend that you read the original study here.


The Sports Betting Community - join us!

Vote stimp1024 and btb for delegates!

                           Thank you!