An idea that seemed brilliant but fizzled spectacularly.
There are a lot of bookmakers around, like 60 decent ones and hundreds of trashy. I wrote a lot of scrapers collecting prices from them, invented an algorithm for automatic name matching (different bookies spell names of same players differently, like R. Federer is same as Roger Federer, or even Russian or Polish spelling of same name etc), so i can compare prices in different books.
Idea was that: betting on the line of one bookie using inverse prices of another bookie as a source of probabilities, run statistically over a long period of time (i.e. virtually betting on all events where price in one booke multiplied by inverse price from another one yields a profit, then recording which bets 'lost' or 'won' and aggeregating profit/loss, doing so for each vs each of the books), finding out as a result, which book 'knows' a particular sport best and worst of all (if i bet in the best of them using a line of worst for probabilities, i get maximum loss, and maximum profit vice versa). Then i wrote a script that does the execution, placing the bets.
Surprisingly, i found an investor willing to back all this in half a day, after pitching just a few of my former clients (i do custom development).
Estimates indicated a lot of profit, but it took just a week to know that there is no such thing as a 'betting market'. All of my accounts have been suspended: they seemingly track events which come into arbitrage with other books, and show down people who bet on them, which i did (except i didn't bet on both sides of the arbitrage, only one that was more likely to win).
I was quite disappointed and depressed for a month.
Fortunately, the investor didn't lose anything: he needs the resulting code for a totally different thing (no execution of bets), and quite happy with what he got, so it ended up being just one more custom development project - very fun to do.
Yup betting is "for fun" unless you use A+ bookies like Pinnacle Sports. Also, some bookies have to have a particular price not because of their knowledge of the market, but because of the people that placed bets with them.
In the end what you describe is basically playing an arbitrage (for which there are sites), but these kind of plays get you suspended, which they did. Nice fore trying it out and writing all the parsers. I know how time consuming it is to get all the synonyms, dates and times right to aggregate the information.
Sorry for not clarifying, i used opening lines. Sites - whole point was not arbitrage but picking only one side of arbitrage which is expected to be more profitable. Also, it was a cloud solution which resulted in very low scraping delays.
Anyways, it didn't work, and i learned it the hard way.
There are a lot of bookmakers around, like 60 decent ones and hundreds of trashy. I wrote a lot of scrapers collecting prices from them, invented an algorithm for automatic name matching (different bookies spell names of same players differently, like R. Federer is same as Roger Federer, or even Russian or Polish spelling of same name etc), so i can compare prices in different books.
Idea was that: betting on the line of one bookie using inverse prices of another bookie as a source of probabilities, run statistically over a long period of time (i.e. virtually betting on all events where price in one booke multiplied by inverse price from another one yields a profit, then recording which bets 'lost' or 'won' and aggeregating profit/loss, doing so for each vs each of the books), finding out as a result, which book 'knows' a particular sport best and worst of all (if i bet in the best of them using a line of worst for probabilities, i get maximum loss, and maximum profit vice versa). Then i wrote a script that does the execution, placing the bets.
Surprisingly, i found an investor willing to back all this in half a day, after pitching just a few of my former clients (i do custom development).
Estimates indicated a lot of profit, but it took just a week to know that there is no such thing as a 'betting market'. All of my accounts have been suspended: they seemingly track events which come into arbitrage with other books, and show down people who bet on them, which i did (except i didn't bet on both sides of the arbitrage, only one that was more likely to win).
I was quite disappointed and depressed for a month.
Fortunately, the investor didn't lose anything: he needs the resulting code for a totally different thing (no execution of bets), and quite happy with what he got, so it ended up being just one more custom development project - very fun to do.