Open Side Menu Go to the Top
Register
Math/tech knowledge necessary to build a worthwhile predictive model Math/tech knowledge necessary to build a worthwhile predictive model

06-14-2019 , 01:27 PM
Quote:
Originally Posted by DefNotRsigley
Are you sure this person isn't spending sleepless nights modeling all the sports Pinnacle offers?


No, her job is to cut limits 95% and raise vig 200%. All well adding pointless amounts of markets so pinny becomes bet365.

meanwhile you can't access copa america futures from the webpage and their CS can't understand that and i just wasted my whole morning on this bs.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 01:42 PM
I was actually surprised at the number of props Pinnacle is offering now. Before she became CEO they only carried ML/Spread/Total and maybe RSW or Team to win conference/championship.

Now they offer, does player hit a HR, total bases o/u 0.5, lame winning margin, etc. with 7%-10% hold
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 02:21 PM
Quote:
Originally Posted by DefNotRsigley
I was actually surprised at the number of props Pinnacle is offering now. Before she became CEO they only carried ML/Spread/Total and maybe RSW or Team to win conference/championship.

Now they offer, does player hit a HR, total bases o/u 0.5, lame winning margin, etc. with 7%-10% hold
They are trying to turn into Bet365
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 02:27 PM
All those props and they can't offer Total Bases though.

It really is amazing that TheGreek dying killed the whole market. Bookmaker is trying to keep it going but they know what they're doing and it's hard.

RIP Spiro, a real bookmaker.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 03:48 PM
lol Paris Smith ,it doesnt look like an alias at all..

****ing criminal ****s
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 04:00 PM
Another measure they took is seaten the belt in asian brokers,now only IBC allows for lifetime 100% PIN limits regardless if you are a winner or not,but evidently their fees are just ridiculous and trustability non existent

Asian connect etc after a small sum of winning they cut you oiff to 50% pin max bet
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 04:19 PM
Sabaneta PS3838 has them,copa america futures


https://ibb.co/s1g3MRb
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 04:30 PM
so much great info in this thread. if you aren't motivated to go build a winning sports model after reading this thread well that's on you. or perhaps you are waiting for the release of Statistical Sports Models in Excel Kindle Edition? in any event, it really shows the way we limit ourselves. do you think james holzhauer said i can't go on jeopardy watson already solved the game? no. he built a better jeopardy model, got on the show, and crushed it. let's do it guys
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 04:33 PM
Quote:
Originally Posted by MisterRodriguez
Sabaneta PS3838 has them,copa america futures


https://ibb.co/s1g3MRb

they only show up on pinny's site if you use their atrocious future view
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 10:15 PM
Quote:
Originally Posted by Like
or, find a few sharps online. check them out as best as you can. get them PPH accounts through friends/networking. Get 25% of their weekly action, up or down. Get 25% of their losses from the bookie. Make money every week no matter if they win or lose. After two bad weeks in a row and after you pay the bookie no problem, ask for more credit and a % off their losses. He agrees to 5k more and 10% a week off their losses since he think he has a whale. (10% off means if they lose 1000 they only owe 900 but if they win 1000 they get 1000). Keep the 10% yourself or split it with your sharps. So now youre printing money every week. Have multiple deals like this with different people. Have sharps sharing accounts and get as much credit on each account as possible. Everytime its maxed out ask for more. Tie up money in futures and ask for it back in credit. If you still have it when the season ends youll get "free credit."

When the agent finally shuts you down, you throw one last hail mary. This can be dicey and is situation dependent, but you tell the agent the truth. Say its been professionals betting the whole time and do you see how good they can do? These guys are part of a syndicate (or some other bs.)You get the agent to turn dirty. If you can pull this off youre a true hustler. (I did this with the head of security for Godsmack a long time ago who looks exactly howd you expect him to look. That was one of the only times I was nervous going to meet someone). The agent can get almost unlimited accounts...he gets you a bunch and you cut him in. He'll want a ridiculous percentage cuz they always have ridiculous deals with their bosses (thats how bad most people are at betting). Youll have to explain to them that theyre making money on both sides and the sharps need a big piece for their "group" etc and you work out a deal. Now hes doing a little bit of a dangerous game...hes actually hoping his book loses money. If his boss finds out he could be in some trouble.

Now you have 10 plus accounts with sometimes up to 4 or 5 people with their hands in each one. Always be on the look out for anyone taking action and put out the word to anyone you know that youll give them cash for a bookie referral. Play lots of poker, especially home games, and ask around. Create spreadsheets for easy weekly tracking and be good with meeting up. This with a fun little part time job could make you good six figures a year.
I spent a year hanging around pool halls trying to get this down and the only thing I got out of it was a couple cool pool trophies in my office. It's not like I'm some sort of antisocial weirdo either, I think the region I'm in somehow has like no bookies. The only ones I've got locally were either insanely sore losers (one told the referral to "find another victim" lmao) and the rest were like 500 credit line or something. Sad to see.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-14-2019 , 10:17 PM
Quote:
Originally Posted by DefNotRsigley
I was actually surprised at the number of props Pinnacle is offering now. Before she became CEO they only carried ML/Spread/Total and maybe RSW or Team to win conference/championship.

Now they offer, does player hit a HR, total bases o/u 0.5, lame winning margin, etc. with 7%-10% hold
Last year they jumped into the NBA player props market for a hot minute but it didn't last
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-15-2019 , 11:45 AM
Quote:
Originally Posted by Fubster
I spent a year hanging around pool halls trying to get this down and the only thing I got out of it was a couple cool pool trophies in my office. It's not like I'm some sort of antisocial weirdo either, I think the region I'm in somehow has like no bookies. The only ones I've got locally were either insanely sore losers (one told the referral to "find another victim" lmao) and the rest were like 500 credit line or something. Sad to see.
I actually wouldnt really advise doing it there and I never did. Its not quite the right vibe, though you think it would be. Its mostly old guys and guys who dont want to be bothered. I see it as somewhat of a darker vibe...little too seedy. Something about poker games, home games especially, is just like a breeding ground for good young hungry agents and even some older guys who almost never stiff.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-15-2019 , 12:15 PM
Some really good stuff here, been interesting just reading along.

@OP I have a liberal arts education in the social sciences. Haven't taken a math course since APs in highschool and taught myself some basic coding to develop my own predictive models for DFS.

A lot of it is much less reinventing the wheel and rather using what others built already and adding in your own secret sauce. If you look at fangraphs, there's very, very little variation in the various projections systems, and a big part of that is that they use each other as inputs so it's a big hall of mirrors. Derek Carty, creator of The Bat speaks about this in detail.

Creating an imaginary universe, applying the laws of physics and simulating a fictional world of baseball would require intense math and coding background. Using all the resources out there to extrapolate and modify is rather simple. It's much more about the idea and the discipline to be honest with yourself in analysis.

We all think we are infallible. Few people will ever bother to try to create their own projections. Fewer still can say "man this sucks" and start over. Most people with failed models failed because they weren't objective enough and just decided that something like BvP or xFIP determined all outcomes so just went about that and considered it bad luck when it failed.

A lot of people contact me regularly for model building advice for DFS, very few actually take it. They think things like checking average scores for college baseball games at Omaha will tell them if the game played there yesterday would be a hitters or pitchers park - the truth is it didn't matter - just one game so even if it is a pitchers or hitters park, the sample of 1 isn't large enough to have any reasonable expectation. Yet many DFS guys agonized for hours over that, unable to see they were wasting time on **** that didn't matter.

Please message me if you are looking for people to collaborate with, I posted earlier about that here. I've done my own models for DFS and looking to transition to regular sports betting.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-15-2019 , 01:16 PM
tango wrote a pretty good basic mlb simulator that could be modified and improved upon

http://www.tangotiger.net/markov.html

so much info out there to get people started
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-15-2019 , 10:28 PM
Great advice in this thread, thank you!
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-16-2019 , 07:18 AM
Yeah this was a good one,hopefully many more to come at this lvl similar to the ones i read from the old days both here and Httop in SBR,although SBR is gone now,its a forum mainly infested by clowns and trolls
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-17-2019 , 12:37 AM
yeah just to add some points

a simple solution is often deceptively effective

a big reason for a lot of the bad reg money in DFS is people trying to build models to determine sleepers - this is kind of dumb. I'm not talking the obvious sleepers - like the talented player in a positional log jam or the starter with stronger peripherals than the stats that matter for scoring - I'm talking about people literally creating models to try to determine which random receiver on the Saints is going to play 5 snaps and score a TD.

Many people actually dedicate countless hours of their lives trying to figure out which TE who averages 2 targets per game is going to have a game where they get 3 targets for 60 yards and 2 TDs. This happens every year and if you somehow lucked into it, you'd have crushed it - so people try to find it relentlessly. The end result is a year of ****ty lineups and even in the weeks where they somehow found that guy, not everything else clicked so they didn't even cash anyway.

My advice is to take what is given, don't try to invent a magic formula. You aren't trying to predict who will have the 4 hr game, that's a pointless exercise. You're more just trying to eliminate some of the bad choices. Like if you were playing the lottery, you know that numbers 0 & 1 won't ever pop up. You still won't be expected to win, but simply choosing number combos that don't include 0 & 1, you'll be able to far exceed expected outcomes. That's really what it's all about - at least for my own predictive model journey.

I keep it really simple, rely heavily on the works of others, and don't have any lock plays. It's much more about narrowing down my choices and hopefully the process of doing that eliminates a lot of the bad ones.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-17-2019 , 05:14 AM
there is more than one market lol

too many people come into this with an ego wanting to crush nba lines and nothing else

most likely where all the pessimism comes from
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-17-2019 , 01:06 PM
I'm a data scientist and I'm happy to be apart of any study group that gets formed here. I mainly work in R, so can teach some R programming too if desired.

Hastie and Tibs are GOAT, good reads there guys. I will say that if Elements is a bit tough to digest for someone, take a look at Introduction to Statistical Learning instead.

My BA and MA are in Sociology and while there is some intense data analysis course work as part of the MA, I'm mostly self taught, so lol at the need a math degree talk.

I've thought about getting into sports betting, but the edges seem so thin that between work, family, and lol live poker, I've just never seen it as a worthwhile time investment. But if I can help this community and learn a bit myself, then maybe I'll come along after all.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-17-2019 , 02:28 PM
Quote:
Originally Posted by cannabusto
I've thought about getting into sports betting, but the edges seem so thin that between work, family, and lol live poker, I've just never seen it as a worthwhile time investment. But if I can help this community and learn a bit myself, then maybe I'll come along after all.
Well it's like starting your own business. There's a lot of up-front capital costs mainly in terms of time spent learning, developing methods, building stuff, etc. If you're a good programmer you can build stuff that's flexible to being automated in the future. The good news is that the up-front dollars required is actually very low since you can get staking/financing pretty easily once you can demonstrate actual subject matter proficiency and an ability to win. And at least compared to other forms of advantage gambling sports betting scales relatively well.

Also like most small businesses, most fail in the early stages.

I'd say unless you absolutely love grinding live poker I think it's a worthwhile long term time investment to at least spend some time looking into sports betting.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-17-2019 , 04:24 PM
Thanks, Tom. Happy to get involved and share some techniques and knowledge of tools with anyone interested.
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-18-2019 , 03:31 AM
enjoying the discussion and would love to get involved in any sort of study group
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-18-2019 , 06:04 AM
https://www.youtube.com/watch?v=1LeDa9RYv_s

Paris Smith Pinnacle CEO interview

wtf,she wants to enhance Casino part of Pinnacle,i didnt even knew it existed
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-18-2019 , 10:02 AM
Quote:
Originally Posted by MisterRodriguez
https://www.youtube.com/watch?v=1LeDa9RYv_s

Paris Smith Pinnacle CEO interview

wtf,she wants to enhance Casino part of Pinnacle,i didnt even knew it existed
"I have been in the industry for 25 years now, um, I started out just...taking taking bets..."

more like taking dicks amirite?
Math/tech knowledge necessary to build a worthwhile predictive model Quote
06-18-2019 , 10:25 AM
OR

find agent, tell him you've got a sweet plan, you want 60% refund on losses but MAN are you going to lose, get poogs in the middle and tell him you need a people person, a true gambler living on the edge, someone to manage this big thing and tell him he'll be able to start threads on REDDIT DOT COM about this ****, make him feel like he's part of the team and give him FIFTY PERCENT of the ACTION.

max and relax random ****, give poogs only a sniff of your obscure and enigmatic #angles because we must be secretive in this business. make the whole thing sound cool and mysterious and tell him the variance in this industry can get you down but keep your head up poogsy one day you too will be a chatroom millionaire, they're going to call you the wolfman. compliment his truck and say you loved that bluff he made on the river there, how ballsy and raw man, you're the wolfman.

lose your ass off, agent profits, the book profits, everybody is making money and ask the agent if he has some buddies in the same business, get poogsy inolved 4 or 5 times over, get your mates to throw in some drunken max bets on the ****tiest odds imaginable, baseball players to hit over 2.5 home runs at -115 where you can only bet on the YES and ****, you and your friends are keeping the accounts all in at all times. you win, the agents win, the books win, your friends win.

poogs loses the heap and has to sell off his dodge ram. man, i don't know. i don't think i wanna be the wolfman, anymore, he tells you. you slip him a $20 and say don't cry wolfman. you're my best boy. one time i lost every day for 45 days straight. i lost for an entire year. they're going to make you a chatroom millionaire for this. think of the reddit post, wolfman.
Math/tech knowledge necessary to build a worthwhile predictive model Quote

      
m