[probettor1]

People who fail trying to making a living out of betting and dont want to quit nor being broke end up writing a book or selling picks. Go to ESPN, you will find enough numbers by team and by player to apply your knowledge. Every number there has some value. Then check a few ATS databases in the web to get more stats. Then find websites with data about referees. Then read all the info available about each game, all that is said by coaches, players, owners. Take time to read as many writes ups as you can. Make sure to get the writes ups for free and read the best only. Find a simple formula that tells you exactly how many points should get each team. This is a little complicated but I did it with zero backgroud in stats. If you cant pass that test in a couple of days then just let go, this is not for you. That formula will give you a very close result to the one used by the oddsmaker if no injuries are reported in both teams. There is no magic, simple math. Once you do all that make sure to bet as little as possible. The oddsmakers have been doing the same for 30 years.

[zorba74]

look into the basics of predicting/forecasting things other than sports. Look into the various ways that is done.

I find this very interesting. Can you provide an example or two?

[sayhey69]

I find this very interesting. Can you provide an example or two?

watch the weather channel. they forecast interesting stuff everyday. very helpful

[zorba74]

watch the weather channel. they forecast interesting stuff everyday. very helpful

Yes, if you are betting totals. Have a nice day and think about being a part of the solution rather than the problem. You are no Simon Birch.

[MonkeyF0cker]

Yes, if you are betting totals. Have a nice day and think about being a part of the solution rather than the problem. You are no Simon Birch.

There's a bit of irony here.

How do you think the weather is forecast exactly? Random? Or based on historical, scientific data?

[lecubs28]

mcduggly,

let me just save you the time and tell you that you will never model anything in sports for profit.

[HUY]

There's a bit of irony here.

How do you think the weather is forecast exactly? Random? Or based on historical, scientific data?

Do you believe that the Weather Channel would be able to make a living solely out of betting on future weather, assuming weather betting had sufficient liquidity?

Just curious what you think.

[probettor1]

mcduggly,

let me just save you the time and tell you that you will never model anything in sports for profit.

you are 99% right here

[sayhey69]

Do you believe that the Weather Channel would be able to make a living solely out of betting on future weather, assuming weather betting had sufficient liquidity?

Just curious what you think.

there are papers on how oj futures markets are better predictors of weather than any weather service. not sure how old they are or if they are dated now but im sure you could find it.

[zorba74]

There's a bit of irony here.

How do you think the weather is forecast exactly? Random? Or based on historical, scientific data?

I do not understand how forecasting the weather will help improve ones sports modeling. Perhaps I am speaking from a place of ignorance as I do not model nor do I intend to but I was interested in Miz drilling down on that point. I am legitimately interested in the cross-pollination aspect of his statement as I migrate a lot of financial trading principles into sports investing.

[allin1]

I migrate a lot of financial trading principles into sports investing.

sounds interesting

[probettor1]

What I know is that if you trust the ESPN projected era and you trust the weather forecast you will lose a lot of mlb bets and you will ruin a lot of planned barbecue days in the beach. To predict the weather I use my grandmother joint pain, if it gets worse I stay at home. For the projected era....sorry something came up, I got to go, I'll be back.

[MonkeyF0cker]

Do you believe that the Weather Channel would be able to make a living solely out of betting on future weather, assuming weather betting had sufficient liquidity?

Just curious what you think.

LOL. Not sure. I don't have sufficient data to make an analysis.

I believe a vast majority of their data comes from local/government shared sources though. So I doubt it.

[MonkeyF0cker]

I do not understand how forecasting the weather will help improve ones sports modeling. Perhaps I am speaking from a place of ignorance as I do not model nor do I intend to but I was interested in Miz drilling down on that point. I am legitimately interested in the cross-pollination aspect of his statement as I migrate a lot of financial trading principles into sports investing.

They are both forecasting/prediction models. Many econometrics principles can be employed in sports wagering.

[byronbb]

http://www.amazon.com/Calculated-Bet...ews/0521009626

I came across this book at my library. It's pretty interesting even from a layman's perspective, he can write well. He breaks down how he formulated a model to beat Jai-Lai betting and how he went about it. Maybe look at it if you can find a copy.

[Jubeibeats]

OP

What sport do you want to model, and what output do you want it to have? i.e. What lines do you want it to make?

Have you looked at least squares methods? I use a sort like this to look at spread lines for Basketball, heavily adapted, but the underlying formula for the whole thing is just a lot of equations, and then using a least squares method to create a standardised set of ratings to create lines from. The underlying equation that the model is trying to solve is
(Pts Differential I predict) - (Actual Pts Differential) = 0
Using the solver and trying to solve that equation, by using a set of ratings for the teams you can create some semblence of a ratings system for teams.
That alone is not good enough to beat the line, but with some creative adjustments you can adapt it to serve a purpose. For a Non US Market its is going to be more beneficial to do this sort of work, as the markets are not as mature, and the sports are not as stable as in the US.

Having worked on these methods for a bunch of years, with lots of practical application making virgin numbers on a wide range of Basketball for a book, modelling helps, or it helped me, doing some basic analysis of modelling scores, but it was only part of the process.

Modelling will always miss something out, so i would suggest creating a few models, using them together, but the final ratings you make should be decided by yourself and adjusted manually.
Modelling alone wont get you there, try to find the best use for your models. What do they tell you? How does that influence how strong said team is.

For ideas, think of some real easy ways to come up with a line. Top of my head, look at the standings, and look at a teams pts differential. Say Miami are +7.5 and San Antonio are +3.5, looking at that basic stat we could say Miami are 4 points better. Make an adjustment for Home Field and you have a line. What other ways can you quantify the difference between teams? Once you have a bunch, look at them, why are they wrong? Do they have certain tendencies to the dog or fav? Can you make any adjustments to them to improve them?

Once you get going, you will make a bunch that will be able to tell you somethings, but not good enough to make a line.

Keep asking questions, try to connect with people a bit more, involve yourself in other peoples threads, try and build relationships more with people here and you might get some more input privately. Although I am slightly different in having worked in books a while, and seeing the process that way also, most of the techniques I use now where what I took from that side, and added to with my own investigations in Forums like this, old school bettorsworld and MW when it first opened, and the friendshipsI made and fostered theregave me pretty much everything I needed to then go on and figure out the rest. Read what you can, realise who talks sense and who doesnt, you sound intelligent so shouldnt be too hard to find the real info on here. Absorb that info, think about it, question it.

Good luck with it all, dont expect to be making a living any time soon, but treat it as a project to improve upon, and you might get something worth exploring more, at the very least try to make models that give you some info you can at least use in your analysis of games.

[Jubeibeats]

What I know is that if you trust the ESPN projected era and you trust the weather forecast you will lose a lot of mlb bets and you will ruin a lot of planned barbecue days in the beach. To predict the weather I use my grandmother joint pain, if it gets worse I stay at home. For the projected era....sorry something came up, I got to go, I'll be back.

I would say ERA is probably not the best to look at for predicting a game, just runs average, earned and unearned, then its about the context you look at those stats.

I would say that is a big problem for bettors is taking stats out of context! Stats are just numbers, facts, its how you choose to interpret them.

[Miz]

my main point is that maybe you should aspire to improve in prediction. what you predict isn't all that critical... in the short term. in fact, the more things you try to predict, the more diverse your approaches will become. Sometimes, least squares regression is all you need, but most often not. For many things, your predictive accuracy will be best using approaches that might not be familiar to you.... yet. it just depends on what you are looking for. if you're preoccupied with the shortest path to a dollar, it will probably hinder your learning.

I like to start with a hypothesis like... angry people drive fast on the highway.

How can I determine if my assumption is valid? How can I quantify it? How can I measure the accuracy of my assessment? If I am objective I may uncover things I don't expect. The most interesting studies are ones where unexpected results occur.

[Dark Horse]

I have the knowledge of knowing what a Poission distribution is and what it represents, as well as other statistical concepts. I just don't know which information (offensive efficiency, defensice efficiency, plays per game, etc) to implement into the model and and how to backtest it. I feel like EVERY stat should matter, but I know some don't, so I'm confused as to which ones are needed. I guess that is the hard part though.

You're underestimating yourself. You have the schooling, but not the experience to follow your own questions to their answers. Once you clear that hurdle, you'll be on your way.

Start with a simple hypothesis. Example. Find a site that shows weather and football stadiums. See if you can find a relationship between wind (velocity and direction) and totals. Determine the causal relationship you expect to see beforehand. If the field is N/S and the wind blows E/W at hurricane force, there will be a lot more running than passing. Would that favor overs or unders? Once you find an advantage you can dig a little deeper. Wind direction (to field) in degree intervals, wind strength in intervals, etc. When you start to define an edge try to take it as far and deep as possible. Expertise in one field is very valuable, because it gives you a major edge over the public.

[A4K]

You're underestimating yourself. You have the schooling, but not the experience to follow your own questions to their answers. Once you clear that hurdle, you'll be on your way.

Start with a simple hypothesis. Example. Find a site that shows weather and football stadiums. See if you can find a relationship between wind (velocity and direction) and totals. Determine the causal relationship you expect to see beforehand. If the field is N/S and the wind blows E/W at hurricane force, there will be a lot more running than passing. Would that favor overs or unders? Once you find an advantage you can dig a little deeper. Wind direction (to field) in degree intervals, wind strength in intervals, etc. When you start to define an edge try to take it as far and deep as possible. Expertise in one field is very valuable, because it gives you a major edge over the public.

BINGO! It all starts with correlation. You need to keep searching for that magical connection. Teamrankings.com has their own model, and over the lifetime of their project they are slightly above 50%.

[A4K]

FYI...... for the NFL, I'd start your hypothesis with Average Starting Field Position for Opponent.

[DirkDiggs]

I love models. I really do. They are incredibly insightful for complex processes as they provide insight into how to think about the process, how processes affect each other, and allow one to refine ones thinking. Betting sports is by no means easy, neither is building an appropriate model for it. In my experience modeling sports has led me to essentially battle variance. I've found it much more efficient to identify spots where variance is your friend.

[Romanov]

Start with soccer or hockey totals if you know poisson

[antifoil]

i am reading nate silver's new book called "the signal and the noise". it deals with the subject of prediction. he developed the PECOTA forecast for baseball. he also runs a blog for the new york times running a prediction model for the elections.

i am not too far into the book but it is pretty interesting. it may not be for advanced predictors. the first half of the book deals with different situations people use models to predict, and the second half discusses the bayes theorem and application. it could be something the OP is looking for.

[Dark Horse]

I love models. I really do. They are incredibly insightful for complex processes as they provide insight into how to think about the process, how processes affect each other, and allow one to refine ones thinking. Betting sports is by no means easy, neither is building an appropriate model for it. In my experience modeling sports has led me to essentially battle variance. I've found it much more efficient to identify spots where variance is your friend.

It's an amazing learning experience, especially if you keep at it long enough. I look back at my earliest 'models', and even though they may have worked they now seem like kindergarten stick figures. The model I'm currently completing has been a year-and-a-half in the making. It's art. The ability to distill something complex into something extremely concise and precise develops with practice. It wouldn't surprise me if history would decide that serious sports bettors did predictive and probability theory an enormous favor.

[accuscoresucks]

the only diffrence between a model/simulator-vs-manually is kind of like this
ex./there are 12 mlb games being played today
1.use a pen and paper and manually handicap=1hour of your time
2.hit the print button and whalla =15seconds of time
conclusion=printing takes far less time

[HUY]

the only diffrence between a model/simulator-vs-manually is kind of like this
ex./there are 12 mlb games being played today
1.use a pen and paper and manually handicap=1hour of your time
2.hit the print button and whalla =15seconds of time
conclusion=printing takes far less time

Who would have thought?

[Justin7]

i am reading nate silver's new book called "the signal and the noise". it deals with the subject of prediction. he developed the PECOTA forecast for baseball. he also runs a blog for the new york times running a prediction model for the elections.

i am not too far into the book but it is pretty interesting. it may not be for advanced predictors. the first half of the book deals with different situations people use models to predict, and the second half discusses the bayes theorem and application. it could be something the OP is looking for.

I thought this book was very, good, and wrote a review of it here:
http://www.amazon.com/review/R38NF4N5GJAIP9

One curious thing about the reviews for this book: nearly every favorable review quickly received six negative votes. A bit odd to say the least.

[SBReinvented]

Whoever talked about applying financial/business concepts to sports betting absolutely has the right mindset. I've been doing this for a few years now and haven't had a losing year yet. I've always been profiting from betting on sports (NHL & MLB). It's not focusing as much on the year-to-year stats that'll make you win, but more on the historical stats and the dynamics of a long season with 30 different teams.

[HUY]

Whoever talked about applying financial/business concepts to sports betting absolutely has the right mindset. I've been doing this for a few years now and haven't had a losing year yet. I've always been profiting from betting on sports (NHL & MLB). It's not focusing as much on the year-to-year stats that'll make you win, but more on the historical stats and the dynamics of a long season with 30 different teams.

And what's the "financial/business" concept that you are applying?

[uvarunthetable]

Whoever talked about applying financial/business concepts to sports betting absolutely has the right mindset. I've been doing this for a few years now and haven't had a losing year yet. I've always been profiting from betting on sports (NHL & MLB). It's not focusing as much on the year-to-year stats that'll make you win, but more on the historical stats and the dynamics of a long season with 30 different teams.

wtf does any of this even mean? the "dynamics of a long season"? please start a thread for the year and post picks, i'll wager you lose money overall

[Mens et Manus]

I have found in my experience that models are only successful if you adapt them to trends in the respective sport you are capping.

Once you win consistently you keep your mouth shut. Don't take offence to it. It can take years to find a successful consistent model.

There is however, nothing wrong with pointing you in the right direction. The books listed, especially by Justin7 are great.

Here is my advice, try quantifying game conditions that you can't find on the stat sheets, then you will find an edge. There are things/stats you can use to cap games that LVSC hasn't figured out yet believe it or not, you just have to find them.