Just wondering how many posters here use code in their betting process. Either for modeling/statistics, scraping, placing bets automatically, etc. My background is programming and software, so I tend to view sports betting as a math/code problem.
-
-
I'm not 'technical' enough to scrape odds. I still copy/paste odds from several sites onto Excel sheets. I'm proficient enough to write formulae into Excel so it'll reproduce the data in horizontal or box layouts rather than vertical lines of data. I can also write formulae to convert the data proportionally or as needed. For example a soccer match with a 105% book price on the 1, X, 2 outright might have a 126% book on correct scores for the 3 outcomes. Theoretically books should have the total percentage for correct score outcomes on 1 at 120% of their 105% book price. For example a home team with a 50% chance would have a 52.5% as an outright on a 105% book. Converting that to correct scores should see that 52.5% become 63% on the higher juice correct score book price. However, books always have more juice on the correct scores for the underdog so the original 50% favourite might only show at 58% on the correct score market. It can be useful to re-adjust odds back into proportional alignment for scores, totals, and handicap purposes. I've written formulae and screen layouts on Excel to do that. That's about as far as my 'coding' skills go.Comment -
We've been trying to evolve into more of a machine learning environment.
It's the next step or evolution in modeling. It's always so static though, and so massaging the stats and inputs to account for a more dynamic market environment is what I bring to the table.
Always learning though, we never know enough.
dmm, what do you think about machine learning principles?Comment -
I rely almost exclusively on data analysis. I do all the things you mentioned with the exception of placing bets automatically. I'd like to get there, but it's way down my priority list. I use a combination of programming (in Basic) and spreadsheets. Since I retired 20 years ago, I've developed systems in most of the sports, written dozens of programs, and probably 100,000 lines of code.
I've often thought about trying to put together a team of like minded people with similar skill sets. There's just too much to do alone.Comment -
Very cool, that's quite the codebase you've developed!
I don't currently do any sort of modeling myself. I've only been betting sports seriously for about a year now. Currently just using a bit of code to look for low hanging fruit -- mostly reacting quickly to market news with a few middles and arbs here and there.Comment -
VBA scripts and Excel are the way to go for meComment -
Was a VBA guy. Now I do everything in python. Easy to learn, easy to use, python is pretty great. I tend to do simulations on baseball and football. Python is able to calculate 1000 runs of 60 or so college football games in about 3 or 4 hours. VBA used to take 2 days.Comment -
Anybody here use a bot to bet with offshores? I'm looking to get one builtComment -
Any major books where this is more difficult to pull off?
I don't currently do this, but when the automated analysis and triggers light up, we're on a mad dash to get down. All manual, with a shitload of primed windows being lined up and sorted through on their own dedicated screens. Then there are the phone call bets. It's a lot to try and do quickly.
I'm interested in this aspect of automating or scripting, but books logging me out constantly is also an issue.
Do you guys try to automate the sign in process as well?Comment -
Bet365 does some detection of scraping/automation tools, so that's a bit more difficult.
How do you organize your windows/screens? By market you're monitoring? How many books still take phone betting? I've never actually placed a phone wager before.
Books logging you out can usually be solved by automating some sort of "activity" on the site. Even just refreshing a market page can keep sessions alive.
Sign-in process can be automated on most sites. The exception is those that generate CAPCHA tests. Bodog/Bovada is the only site that I've seen where this is an issue.Comment -
So it's possible. Good, that's what I wanted to hear.
I'm probly gonna put up an ad on Upwork to get something like this built.
Any special measures I need to take to prevent the sites from finding out I'm using a bot? I'm sure they wouldn't be big fans ..Comment -
I would recommend that the scraping tool use a browser object to navigate the pages. That way it will be damn near impossible for it to determine if it is a bot or not. I would recommend selenium.Comment -
Selenium is definitely a great way to go. It's also easy to detect without ongoing modifications to your code -- see https://fingerprintjs.com/ which directly sells their software to gaming companies.Comment -
You need to be aware that scraping bots usually need maintenance, so you may be continually paying someone to fix your bots when they break. Another challenge is mapping your bets to the appropriate team on the various sites. Some use different notation so you will need to keep a dictionary of the variations between books to make sure you get your bets down.Comment -
This. Talking 13000-15000 lines of code total in order to write scripts for each site and then do filtering plus ongoing maintenance. That’s just to get sorted output not including automating the next steps after that. So you need someone that’s kind of ongoing retainer after the initial build.Comment -
What places allow you to bet through API. The only one I'm aware of is Fairlay?Comment -
Pinnacle, Matchbook, Betfair plus the aggregators like Vodds.Comment -
And to contribute further, I automate everything. Including bet placement.Comment -
When I signed up for Pinnacle, they told me the API was no longer available.Comment -
It is. Need to guarantee monthly turnover of $250k iirc. Or pay a fee, maybe $5k but that I cannot rememberComment -
And yet every one of you in here are losers. Every one. Mr Chichens**t (KVB) will not even place awager with me because he KNOWS that doing so will cost him money (if he has any) or much more (if he does not). Right Mr. Chickens**t?Comment -
I've challenged you to a handicapping contest, but crickets. And I dont even have an axe to grind.Comment -
I'm a big coder, I code like 4 or so hours a day at work. I would never be comfertable with automated bet placement. God bless if you are, I would be worried about a bug some place that would end up making me bet like 10,000 when I meant to bet 100 or something like that.
Funny story, I worked with a guy, was in his 50's head of research when I was first starting out, who was a quant programmer at goldman sachs in earlier years. He built a trade processing program for the equities desk. He was demonstrating it, and he put out an order that said "Buy 100 shares of Apple at $.01". Of course, it didnt work because apple was like $200, so the order didnt get filled. So he switched it, and said "Sell 100 shares of apple at .01". It filled immediately, causing a loss of 19,999. He said all traders cracked up, and they said " Don't worry about it, we'll make it back in like an hour".
That story scared me from every building an automated betting program.Comment -
Haha that's hilarious. I know the feeling, lately I've been spending my time building stock market algos but they haven't panned out since I am unable to find a reliable enough broker (that also has a backend I can use to place orders automatically with python or matlab) that also doesn't charge commission. I've had some limited success at smaller scales and lower frequency, but I don't trust any brokers set up or my own code enough to run it with more than tiny amounts of actual money.Comment -
Anything that I do with betting I automate as much as possible.
Feel free to DM me would love to code for different sites and issues/solutions that i've found. All of my code is in Python.Comment -
I started with Excel sheets when I began betting. I wanted my models to get better all the time, so I moved to Python. Many years later I branched off into a software career. DevOps style role.
Used to automatically place bets on the Pinnacle api until early 2022?
Couple of years ago I made models that use the swar/nba_api module to apply SVR, Random Forest and Linear regression using the player and team stats. It really helps guide decisions.
One of the most helpful things I've done is gotten good at matplotlib. I have a OHLC style chart with teams open/close odds. Very informative for identifying teams that are always backed or laid heavily (Magic have drifted in every game this season bar six and in 3 of the last 4). Stuff like pie charts for % a team has covered a spread in a given situation... easier than looking at rows of csv all the time.
Pandas, numpy, scikit, matplotlib, Jupyter and databases to persist your bets/data. Steep learning curve.
As KVB says in one of the videos if you have the ability to do the quant stuff and know the market/sport well, you will win out. It has turned me from a loser to a winner.Last edited by Spankyf; 01-19-23, 08:43 PM.Comment -
All my models scrape jjgold comments as a micro indicator. been tailing the man for yearsComment -
CODE what fawking Code lol aint no soaftware to get me the right side in games in any sport my charts ats work at ana amzing 89% clip
in nfl and fading teams after scoring 40 + points on subbie losers when they meet up with an 4/6 team or betterComment
