BetBurger | Live and Pre-game surebets
RebelBetting - Turn betting into investing

Database of in play football (soccer!) betting odds - any ideas on how to make something of it ?

Is this the new hack of smart gambling?
polev
Gaining experience
Gaining experience
Karma: 1
Posts: 3
Joined: Sat Oct 03, 2020 7:00 pm

Database of in play football (soccer!) betting odds - any ideas on how to make something of it ?

Mon Apr 04, 2022 11:38 am

For the last couple of years, I've been working on fully-automated betting scripts using machine learning to detect misprices, make millions on them and retire at 32.

Long story short, I haven't been able to make it work so I'm throwing in the towel. But in the process I've built a high-quality database of a few tens of thousands of lines (one game = one line) of in-game information (score for home team and away team, number of red cards for each, and betfair/pinnacle betting odds for home/draw/away win as well as over/under 0.5/1.5/2.5 relative to the score at the time) at different time points during the game (10', 25', 40', half time, 50',60',70' and 80'). I used this data to train my models and backtest my strategies.

As far as I've been able to make out, similar databases available online only have prematch info, but I know many bettors do inplay as well so I feel like there could be a market for this data. So... if you guys have any ideas on how to make something of it so that the work of the last 2 years hasn't been a complete waste ?
Sam5za4
Gaining experience
Gaining experience
Karma: 4
Posts: 24
Joined: Mon Mar 29, 2021 1:19 pm

Re: Database of in play football (soccer!) betting odds - any ideas on how to make something of it ?

Mon Apr 04, 2022 12:30 pm

With this database, you should try to make a Algo which will give chance for FT correct score, based on all historical matches and statistic.
Now i'm trying to do something similar for Bitcoin price movement. Based on BTC historical daily moves + other things like supply and volume . So will give you signals once is good to sell or to buy more BTC.
Chris Albrecht
Gaining experience
Gaining experience
Contact:
Karma: 0
Posts: 2
Joined: Tue Jun 23, 2020 8:41 pm

Re: Database of in play football (soccer!) betting odds - any ideas on how to make something of it ?

Mon Apr 04, 2022 8:55 pm

I'd love to have a look at it. I do not exactly know if it's useful in any way, but I would give it a try. So if you interested i a talk about just let me know.
Chris Albrecht
Gaining experience
Gaining experience
Contact:
Karma: 0
Posts: 2
Joined: Tue Jun 23, 2020 8:41 pm

Re: Database of in play football (soccer!) betting odds - any ideas on how to make something of it ?

Tue Apr 05, 2022 4:22 pm

I made a reply with an email adress for a contact possibility, but it seems like this is not allowed. What are the options to get in contact with someone through this forum?

@polev The only way I can judge if your project is helpful for me and my projects is to have a closer look at it. It sounds promising though.
User avatar
arb12
Totally Pro
Totally ProTotally ProTotally ProTotally ProTotally Pro
Karma: 23
Posts: 1008
Joined: Sun Jul 26, 2020 2:55 pm

Re: Database of in play football (soccer!) betting odds - any ideas on how to make something of it ?

Wed Apr 06, 2022 1:07 am

polev wrote:
Mon Apr 04, 2022 11:38 am
For the last couple of years, I've been working on fully-automated betting scripts using machine learning to detect misprices, make millions on them and retire at 32.

Long story short, I haven't been able to make it work so I'm throwing in the towel. But in the process I've built a high-quality database of a few tens of thousands of lines (one game = one line) of in-game information (score for home team and away team, number of red cards for each, and betfair/pinnacle betting odds for home/draw/away win as well as over/under 0.5/1.5/2.5 relative to the score at the time) at different time points during the game (10', 25', 40', half time, 50',60',70' and 80'). I used this data to train my models and backtest my strategies.

As far as I've been able to make out, similar databases available online only have prematch info, but I know many bettors do inplay as well so I feel like there could be a market for this data. So... if you guys have any ideas on how to make something of it so that the work of the last 2 years hasn't been a complete waste ?
@ Polev,
If you need some money, or if you simply technically built the data without being keen on soccer and future long-time business vision, you may sell that database. However, if that's not the case, personally strongly advise you not to sell anything, but to diggin' in that further and beyond.
What you've achieved so far if correctly fulfilled in every row and column as a rationale behind the current odds at every XXX or XXY minute, is a wonderful tool you've built, but you need the practical application further. Look at the database as it's is a derivative of odds changing according to Bayes' theorem in standard uncomplicated cases, plus the influence of various factors. The selection and estimation of these factors is your task only, aiming to get more value than simply computing the standard odds change according to Bayes' theorem.

Often you can extract additional tenths or hundredths of the odds compared to published odds, and in the long run, be ahead. Or sometimes you may spot something and wager against totally wrong odds, let's say your computed most probability AH -0,25 versus published leading probability for AH -0,75.

Do you remember, the 2018 CL event Barça vs Hotspurs, the odds one hour prior to announcing the lack of Messi in the starting team, and the odds at the moment of that announcement? The in-play odds reaction at the moment when Messi entered that match somewhere in the second half.
Or in-play odds fluctuations due to Loris Karius' performance for Liverpool.
Or in Cricket, the usually seen in-play odds, when the chasing team is far behind, but have you seen the odds reaction when the new batsman is Andre Russell, provided that there is XXX or YYY required run rate for a win? And the in-play odds in the same situation if it is not Russell? Or complex in-play situations like the IPL 2019 final, and in-play odds when Lasith Malinga bowled the final over in the Championship vs Chennai Super Kings? The top-notch bowler Malinga has achieved an awful ECON rate before he played the last title-decider over, how were the correct odds here if that situation was standard or another bowler instead?

Don't give up, make upgrades on your odds on the fly on more sports, seek advantage, and practice yourself more and further.

Return to “Value betting talk”