Blackbox modeling (Video)

[Wrecktangle]

Yep, I'd do bases if: there were two of me or there were at least 36 hrs in each day.

I noticed that the guy who does Baseball Insight is giving it up, and he only does lines, scores, starting pitchers, innings pitched, BIP, and a few other cat and dog stats. Not nearly enough to run a sim. So unless you build some screen scrapers, where do you get the data?

On the other hand, the NFL got harder when the NFL started posting Gamebooks and every other guy who could throw a line of code started writing football simulations.

[Justin7]

Football is still easy to crush.

MLB- I used to download rosters and player data from USA Today daily (and before that, I hand-typed it weekly).

[Wrecktangle]

Justin7, I'd love to know what you consider a "crush" is in the NFL. Frankly, if I'm under +10 flat betting units per year, I consider it a blown season. Barely made it this year in sides; 2005 & 2004 were the last bad years before this.

[NeedProtection]

Football is still easy to crush.

lol

[statsguy41]

justin,

do you have any more information on a formula a person can start out with for their blackbox model? i have a lot of weekly nfl data i have scraped and i need a starting point for a formula but i dont know where to start.

thanks

[Justin7]

justin,

do you have any more information on a formula a person can start out with for their blackbox model? i have a lot of weekly nfl data i have scraped and i need a starting point for a formula but i dont know where to start.

thanks

Come upwith your own formula to predict their scoring based on the info you have. Then backtest and see how close you were.

You could start with just average points per game, adjusted by the opponent's defense's average points allowed.

Or you could estimate how many yards you think they'll get, and convert that to points.

[VBOMBER]

Does anyone have any good recommendations on books about modeling?

[MrLuckyPants]

Great info man, thanks for this.

[Justin7]

I don't think there are any good books explaining how to make a model in sports betting. I've actually started writing one, though I doubt more than 20 people would pay for it.

[MrLuckyPants]

I'd pay for that lol - it's somewhere to start for someone not very proficient at excel like me.

[VBOMBER]

I don't think there are any good books explaining how to make a model in sports betting. I've actually started writing one, though I doubt more than 20 people would pay for it.

Not specifically for sportsbetting, just in general. There are so many books and textbooks out there I was wondering if anyone had any experience (from college, an MBA program, or just self-teaching) with one that taught them a number of skills they still use when actually creating models for sports.

[Kyleben]

Not specifically for sportsbetting, just in general. There are so many books and textbooks out there I was wondering if anyone had any experience (from college, an MBA program, or just self-teaching) with one that taught them a number of skills they still use when actually creating models for sports.

I have been using some various modeling methods for sports betting for about two years( i have been posting my system outcomes in the nba thread on this forum for a couple weeks.). I am in school right now getting my masters in economics, and i can say that i got almost all of my ideas and knowledge from a combination of textbooks and sports knowledge. There is a great textbook out there called, "Principles of Macroeconometric Modeling", that you may want to check out, it should be on amazon or somewhere else.

[Kyleben]

also:Modeling Dynamic Economic Systems (Modeling Dynamic Systems) by Matthias Ruth, Bruce Hannon, and Jay W. Forrester

[Wrecktangle]

Statsguy: You can use Elo ratings (from the chess world) which only uses win -loss records. You can find the formula by google search. Or you can use a simple predictor that uses scores only and adjustes for playing schedule. One place that has a decent explanation is:

http://www.sportsbookreviewsonline.com/sportsbettingarticles/sportsbetting11.htm

but there are other places that have similar. As I recall, the book "The Hidden Game of Football" had a description of this also.

[LT Profits]

lol

What is so funny?

Justin is one of the best football cappers I have ever seen. Too bad there is no Arena League this season.

[Wrecktangle]

I don't think there are any good books explaining how to make a model in sports betting. I've actually started writing one, though I doubt more than 20 people would pay for it.

Justin, I'd buy it. Shoot, I bought Dan Gordon's piece of crap, I'd surely buy yours...

[Neil Nollidge]

Since books have much more resources than bettors why is it that a statistical approach can beat them?
1. Because they intentionally skew the line to take advantage of public perception.
2. There's an inherent advantage one has betting into a line, over the one posting the line
3. something else.

I am not sure about 1 & 3, but 2 looks real to me - the effective line of the bettor is a combination of the poster line and the raw bettor line.

[MadTiger]

Just a minor. I was 4 hours short of getting a second major in it, but the senior spring bug got me.

Minor here as well.

Thank you for your efforts with the video.

[durito]

.

[mathdotcom]

I don't think there are any good books explaining how to make a model in sports betting. I've actually started writing one, though I doubt more than 20 people would pay for it.

[mathdotcom]

That was my approach - simulation, play a matchup 10 million times and set prices based on the results. Over 1000 lines of code.

Sadly, I haven't had time to fix it up. I know where I went wrong... The definition of depression: reading someone's Ph.D. thesis explaining EXACTLY where my model went wrong *sigh*

lol somehow I doubt this sufficed for a guy's PhD thesis. Where can I find the thesis, Justin? Everyone would have a PhD if all it took was fixing all your mistakes. Unfortunately, they have this "original contribution to knowledge" clause that makes it kinda tricky.

[tacomax]

I don't think there are any good books explaining how to make a model in sports betting.

You've not seen this gem?

[mathdotcom]

also:Modeling Dynamic Economic Systems (Modeling Dynamic Systems) by Matthias Ruth, Bruce Hannon, and Jay W. Forrester

How does that help you with sports betting? If it's anything like other dynamic economic systems books, it is mostly concerned with maximizing certain functions over finite/infinite time horizons (dynamic programming).

This is all 2nd year econometrics/regression analysis

[Justin7]

lol somehow I doubt this sufficed for a guy's PhD thesis. Where can I find the thesis, Justin? Everyone would have a PhD if all it took was fixing all your mistakes. Unfortunately, they have this "original contribution to knowledge" clause that makes it kinda tricky.

The thesis wasn't about my model, but an invalid assumption that modelers/analysts make.

[IrishTim]

Justin, I don't know how you put up with this mathdotcom guy following you around and making stupid little comments.

[dodger33]

Yep, I'd do bases if: there were two of me or there were at least 36 hrs in each day. I noticed that the guy who does Baseball Insight is giving it up, and he only does lines, scores, starting pitchers, innings pitched, BIP, and a few other cat and dog stats. Not nearly enough to run a sim. So unless you build some screen scrapers, where do you get the data?...

Justin and/or wrecktangle are you guys aware of any group of people or a company working together to build a mlb model?

[Justin7]

Justin and/or wrecktangle are you guys aware of any group of people or a company working together to build a mlb model?

Yes. Both of us (I have several, he's probably developing one this spring).

[mathdotcom]

The thesis wasn't about my model, but an invalid assumption that modelers/analysts make.

Which is what? That once you throw in a few important dummy variables you are drastically lowering your degrees of freedom?

Is your MLB model slow coming because you can't figure out logit/probit?

[fiveteamer]

All this shit and Justin is still 50%

[durito]

Which is what? That once you throw in a few important dummy variables you are drastically lowering your degrees of freedom?

Is your MLB model slow coming because you can't figure out logit/probit?

How are your sports betting models coming along?

Cleaning up at SBG Global I would hope.

[fiveteamer]

Bottom line Justin is a failed lawyer and a 50% coin flipper when it comes to gambling.

The difference between Justin and the average square is vocabulary.

[Justin7]

Bottom line Justin is a failed lawyer and a 50% coin flipper when it comes to gambling.

The difference between Justin and the average square is vocabulary.

Thank goodness for my adequate vocabulary. Everyone has to have something.

[mathdotcom]

How are your sports betting models coming along?

Cleaning up at SBG Global I would hope.

lol, do not need a model to beat SBG. It is easy money. But you can keep enjoying your A+ books that you can't beat, but would definitely payout if you did.

[durito]

lol, do not need a model to beat SBG. It is easy money. But you can keep enjoying your A+ books that you can't beat, but would definitely payout if you did.

gee really

[IrishTim]

All this shit and Justin is still 50%

Last I checked his spreadsheet, he was hitting over 56% and up 17+ units.