Thank you Hedgie and I think you are right. I would like to take this opportunity to apologize to anyone that felt disrespected by my post and would like to specifically apologize to Rsigley for calling him an "idiot". That was hurtful and unbecoming and it's not who I want to be in this world. I hope that we can all go forward in a productive manner.

What I find interesting is that you seem to be under the assumption that everyone in this thread is refuting the idea that betting off market lines is profitable. Literally no one has said that.

This thread is titled "Math/tech knowledge necessary to build a worthwhile predictive model," thus people are discussing building models here. No one has said that is the only way to win at this game.

This thread is not titled "The easiest way to make money betting sports." That is a different discussion that is not relevant in this thread.

Also amusing how you storm into this thread acting like a know it all jerkoff and mock/bully several posters here (all of whom have been more successful betting sports than you will ever be) then turn around and play the victim card, complaining about bullying after you're exposed for being a brokedick busto playing needledick stakes.

Johnny that statement is hurtful and inaccurate. We are turning a new leaf on the sports betting forum and the new frame is one of collaboration and coherence. I look forward to sharing the fruits of our labor with each other.

cody,**** off


go chase steam and get your 3 percent edges in euro books for peasant stakes whilst in your other monitor you crush pokerstars freerolls

i think that really sums up the heart of the matter - no more pokerstars freerolls so he now has more time to come here and sh*tpost

REKT

Would you actually be satisfied with being able to project K prop lines? Even closing lines on that prop aren't exactly super sharp IME and I'm a little surprised that you'd be okay with using the prop line as a dependent variable here.

but if you're predicting 9 K's and the o/u is 6 consistently, you are better off starting a website based on your gambling knowledge and try to sell your subscription to people than use it to make money better

shout out to action network/fantasy labs

being off by 1/2 K is possible, 3-4 K's nah

Yeah I hear ya, I've actually been using your projections to retool my model, some of them are smack dab on the money and some are upwards of 1.5-2 Ks off. I figure you're probably better than market closing line. But if you do preseason projections, why wouldn't you try to use those rather than market rates? Lineups and compound errors I guess would be something I'd worry about.

There's some insanely infuriating bug in my code where I can't seem to figure out why I can't refactor log5 into a more general form to add explanatory variables and it's driving me nuts. Just got back from a brief vacation which should help, I'm a terrible coder so often times it's a technical glitch that I miss due to tunnel vision.

Today's DNRSDs (DefNotRsigleyDiscrepancies):

Sabathia dead on
Sanchez 0.5 K under DNRSP (DefNotRsigleyProjection)
Eflon dead on
Matz dead on
Rodriguez 0.5 K over DNRSP
Giolito 1.0 K over DNRSP
Plutko within +/- 0.1 K
Keller dead on
Lester 0.8 K over DNRSP
Teheran 0.2 over DNRSP
Greinke 0.3 under DNRSP
Kershaw dead on
Pomeranz 0.4 under DNRSP
Jon Gray 1.0 K over DNRSP

Had to make a few assumptions to come up with an easy number, but looks decent. Back from vacation now so I'm going to add in a couple new factors to get it in line with The Market so I can get better DNRSPCLV. Just eyeballing the ones where I'm off by ~1 K I can see a couple issues already that should be flattened out with the v1.1 update.

As expected, the error was an embarrassingly simple one. Just a couple logistic regressions away from a model I'll be proud to Pastebin.

screw building models, let's just operate out of Germany https://forumserver.twoplustwo.com/2...ayers-1746446/

for the steam chaser but not chasing steam guy

how do i know which is "good" or bad steam

there's a Euro basketball W's game that starts at 9:45 (4 minutes ago) and tons of money came in on over 130.5 pushing it to -149, and belgium -2 pushing it to -135

betting should have closed at 9:45, but is still open now and suddenly it looks like tons of money is coming in on under 130.5 as it moved a lot and is currently -160

money must have come in as russia +2 also since belgium is back to -115

which steam do i trust? if i picked off a rogue number at 9:44 it would have been different than a rogue number at 9:50. so what is a "right" play? last steam wins?

Due to your post and others I started modeling this week. I read through a handful of your posts and I realized that the requisite skill you demonstrated was quite a bit higher than I realized.

As for that match in particular, I am not familiar with the teams or leagues in question. I do concede the larger point though.

<-- MS statistics, R programmer, DFS rake generator

OP, that salad model thing is certainly jargon-dense, but it isn't all that deep conceptually. It's a very similar idea to using wisdom of crowds, where the n=11 different models or whatever are the people in the crowd giving their opinions. And then the ensemble method is basically just averaging them in some way. We're picking ones that are the least correlated because it doesn't help us much averaging 11 touts from the Action Network who all get their information from the same source and tell us to pound the over. We want opinions from experts that draw on different features to form their opinions, and who are right and wrong about different things.

There's no reason why you'd need to grind through thousands of hours of theory and proofs to implement that in practice (or want to, for that matter). The heaviest lifting is going to be the programming part, especially just wrangling up all of the data, formatting it, merging it, etc. Like RSigley said, the best way to do this is to just start doing something.

I can try to answer theoretical, philosophical, or R questions if people have them.

dfs refugees are welcome here.

the talent level of our members here is truly amazing. i'd put our education, skills, and experience here against the department of any major tech company.

how is everyone's progress coming along? i'm about 100 pages into The Elements of Statistical Learning. thanks for the recommendation. it's full of great, practical information without being overly technical in terms of proofs and derivations (no disrespect to lagrange and his multiplier delta but we have computers now that solve optimization problems).

Thanks for the share. The biggest leak in my game is I never sit down to learn things properly because the Google + Stack Overflow + Duct Tape solution is usually all I have time to do as I'm usually working on something a few hours before contest begins.

I just downloaded it and in my initial skim, it seems like this is ideal for me. My main issues is that I have a lot of trouble getting what I specifically need. My foundational academic knowledge is non-existent but my practical usage somewhat advanced so I understand and utilize fairly advanced stuff that a layman wouldn't be doing and yet when I try to read mathematic notation, it looks like hieroglyphics to me.

I can only seem to find "If Johnny has 4 apples and 3 oranges" type learning material which doesn't help anything I need or something that assumes I've already decoded the Rosetta Stone and can see advanced notation and immediately know what it all means. Here they seem to introduce it all and break it down in the preceding paragraphs before just dumping the notation out there, which means this is something I that won't just help me formally learn many things but going forward, it'll be easier googling stuff as a refresher since hopefully I'll be capable of understanding the notation.

Been meaning to find and read a book like this for some time. Thank you for the advice.

@Lawnmower Man, I too am a DFS rake payer. Still pretty active but don't see much future in it. If interested in joining a study group on slack, send me a pm.

I'm just here for the models. I'm trying to spend less time on DFS, not more. Simultaneous move games with a ton of analytical people crowdsourcing the answer aren't really gold mines at 15% takeout. It reminds me a lot of turbo SNGs only more simplistic and with higher rake. That isn't stopping people from selling packages at 1.25 though.

ugh, i misrote, i meant for sportsbetting

Can we all take a second to just marvel at the absolute beating that cody american eagle guy took in this thread?

It was an unprecedented thrashing, the likes of have which rarely been seen on 2+2. I too would revel in the joy of this unparalleled beatdown were I not at the receiving end.

...and all without Thremp chiming in...

i think youll come out the other side a better man

You know what I think the quote "Better to remain silent and be thought a fool than to speak out and remove all doubt" is dead wrong particularly over the internet. If people want to think you're a fool that's their prerogative. It doesn't matter what they think and in the meantime, we are processing information and learning and what better way to learn then to hear differing opinions? Poogs learned years ago that's the best way to get that information over the internet so who is really the fool here? Don't believe me, how about this quote. "If any of you think you are wise by the standards of this age, you should become 'fools' so that you may become wise." BOOM. St. Paul ***** your **** up.