mathematics behind sports betting

what kind of logic/branch of mathematics are used when people make betting models on sports like boxing/ufc?seems like there would be some guess work or subjective inputs.

what about for a game like chess given rankings?

i made a model a few years back on surfing. the logic was to find the probability distributions of scores for each surfer based on different types of waves(some waves require different skill sets). once you have the the distributions for 2 different people you can find the probaility of 1 beating another. the model turned out to be too close to the market price to make any major +ev bets. and the sample size was too small.

some other limitations to my model were that it didnt take into account form of surfers which would be important because i was treating scores that were from 3yrs ago as important as scores that were happening nowc

From what I understand, for things like point spreads and over/unders that pay out -110 (you have to risk 110 to win 100), odds makers try to set the lines to a point where they can get 50-50 action. If 50% of people pick one side and 50% pick the other then they are basically freerolling and collect the extra edge while paying out one side with the other sides bets. It's actually incredibly sick how accurate some vegas lines can be regarding these things.

For + - lines, you try to determine the probability of outcome and then set the line to give you some edge. For instance, say a ufc fight had a fighter expected to win 80% of the time and another to win 20%. If the population bet 50-50 on each side, the breakeven line would be -400 for the favorite and +400 for the underdog. An odds maker is going to make the line something like -500 for the favorite and +350 for the underdog, since the payoff won't match the expected return on the outcome. You are essentially getting inferior odds on the payoff no matter which side you choose.

This is why it is incredibly hard to beat sports betting long term. Even if you can develop a model to predict favorites, chances are the odds you are getting betting the favorites isn't going to be good enough. It's like betting odds on the don't pass line in craps. You are gonna win more often but you will have to risk more to win less (technically speaking that bet is 50/50 but I'm just trying to illustrate a point). Optimally, I'd assume it's probably best to find instances where underdogs are potentially not underdogs so you would be getting better than a 1 to 1 payoff on a hidden favorite. Even then, the best odds makers are extremely smart people and the sportsbooks wouldn't be in business if these guys sucked at their jobs.

I'm not an expert on this kind of stuff so anyone feel free to correct me if I'm mistaken.

No you are spot on sir. Incredibly difficult to beat sports especially just betting a site like bovada straight up.

If you are trustworthy and bet enough there are some 2+2 groups that bet with each other on vig free lines

for fighting maybe you could get some ideas from this: https://getd.libs.uga.edu/pdfs/johnson_jeremiah_d_201208_ms.pdf

disclaimer: haven't read most of it

On April 26 2015 06:26 K40Cheddar wrote: From what I understand, for things like point spreads and over/unders that pay out -110 (you have to risk 110 to win 100), odds makers try to set the lines to a point where they can get 50-50 action. If 50% of people pick one side and 50% pick the other then they are basically freerolling and collect the extra edge while paying out one side with the other sides bets. It's actually incredibly sick how accurate some vegas lines can be regarding these things. For + - lines, you try to determine the probability of outcome and then set the line to give you some edge. For instance, say a ufc fight had a fighter expected to win 80% of the time and another to win 20%. If the population bet 50-50 on each side, the breakeven line would be -400 for the favorite and +400 for the underdog. An odds maker is going to make the line something like -500 for the favorite and +350 for the underdog, since the payoff won't match the expected return on the outcome. You are essentially getting inferior odds on the payoff no matter which side you choose. This is why it is incredibly hard to beat sports betting long term. Even if you can develop a model to predict favorites, chances are the odds you are getting betting the favorites isn't going to be good enough. It's like betting odds on the don't pass line in craps. You are gonna win more often but you will have to risk more to win less (technically speaking that bet is 50/50 but I'm just trying to illustrate a point). Optimally, I'd assume it's probably best to find instances where underdogs are potentially not underdogs so you would be getting better than a 1 to 1 payoff on a hidden favorite. Even then, the best odds makers are extremely smart people and the sportsbooks wouldn't be in business if these guys sucked at their jobs. I'm not an expert on this kind of stuff so anyone feel free to correct me if I'm mistaken.

addressed this in another thread, but this is a common misconception. sure vegas likes to have a freeroll, but they set the lines in the most accurate way they can first and having 50/50 action is not a priority. the reason being that if the line is clearly wrong, sharks will come in and pound the line all at once and punish them for having a bad line. incentive is there to never have very exploitable lines.


for boxing/UFC, models dont really work. sample sizes are too small, fighters are always changing, styles often make fights and level of competition is always different from future opponents. lines are made off of the eye test and intuition.

other sports, models make sense:
-in baseball, i think most models are a markov chain. since there are statistics on everything, we can simulate every situation thousands of times and come up with results. for instance, one pitcher has certain stats vs right handed hitters, the right handed hitter at the plate has certain stats vs left handed pitchers, we can figure out what their likely stats are to be against each other, figure out how likely the player is to hit a ball to every location, how often each position player will make an error or get to a ball, etc. just model the whole thing out and run it tens of thousands of times in a monte carlo simulation.

can do that type of stuff for tennis as well.

for basketball, you can either model based off of each possession or you can base it off likely minutes and how players do with certain lineups adjusting for defense.

you can imagine how models work in other sports too

ok thanks guys.

i came across this article whilst browsing the net

http://www.ualberta.ca/~bhumphre/class/humphreys_paul_weinbach_wp.pdf

and thought it may apply to things im working on at my job (financial trading)

then i got curious because of the big fight next we a eekend

another thought was if all information is available on something like basketball, then how does haralbos volgaris has a huge edge in betting? he must be using a different betting system to the bookmakers, or only bets when he thinks the line is way out

There do exist some statistics about mma which are fairly reliable. Examples:

Someone 10 years younger or more will generally win.
Someone with 6 inch reach advantage will generally win if its a standing fight (fuck you Shane Campbell)
Home advantage is usually big in Brazil and Canada (for examples, judges yesterday sometimes made it 30-27 instead of a clear 29-28).
Missing weight usually is a bad sign.
Ring rust of 2 years or more
etc.

As an example, in yesterday's fight, Clements vs Taleb, i wasn't sure who would win looking at both fighters's recent fights, but statistics suggested STRONGLY that taleb had this, so i went with him.

On April 26 2015 06:58 traxamillion wrote: No you are spot on sir. Incredibly difficult to beat sports especially just betting a site like bovada straight up. If you are trustworthy and bet enough there are some 2+2 groups that bet with each other on vig free lines

I honneslty think its possible to make money on mma, especially if you use multiple sites. 5dimes has really good line the day of the events, Pinnacle has good line when they open.
I know of some sites where some guys sell their advices and they seem to make money in the long term.

Their strategy seems to only pick a FEW fights where they are pretty confident in their choice, instead of betting on fights which are toss ups.

I also think that someone who knows all fighters very very well can make money betting on opening lines. Sometimes, you just know one line will drop. For example, Alexandra Albu that opened at -125. You can then freeroll with a second bet on the day of the event.

So Floofy do you pick fighters based on looks?

On April 26 2015 16:28 bigredhoss wrote: So Floofy do you pick fighters based on looks?

no i don't?? why are you asking this? Are you trying to imply the stats i said above are wrong/stupid?

On April 26 2015 15:39 Floofy wrote: Show nested quote + On April 26 2015 06:58 traxamillion wrote: No you are spot on sir. Incredibly difficult to beat sports especially just betting a site like bovada straight up. If you are trustworthy and bet enough there are some 2+2 groups that bet with each other on vig free lines I honneslty think its possible to make money on mma, especially if you use multiple sites. 5dimes has really good line the day of the events, Pinnacle has good line when they open. I know of some sites where some guys sell their advices and they seem to make money in the long term. Their strategy seems to only pick a FEW fights where they are pretty confident in their choice, instead of betting on fights which are toss ups. I also think that someone who knows all fighters very very well can make money betting on opening lines. Sometimes, you just know one line will drop. For example, Alexandra Albu that opened at -125. You can then freeroll with a second bet on the day of the event.

books set betting limits for early lines very low and try to get the line corrected from people betting them. you cant really make very big money like this because you cant get much down on bad lines.

On April 26 2015 16:40 Floofy wrote: Show nested quote + On April 26 2015 16:28 bigredhoss wrote: So Floofy do you pick fighters based on looks? no i don't?? why are you asking this? Are you trying to imply the stats i said above are wrong/stupid?

i'm just trying to verify the most important factor in a fighter's success. it sounds like you think it's looks. or is it something else?

On April 26 2015 17:04 bigredhoss wrote: Show nested quote + On April 26 2015 16:40 Floofy wrote:   On April 26 2015 16:28 bigredhoss wrote: So Floofy do you pick fighters based on looks? no i don't?? why are you asking this? Are you trying to imply the stats i said above are wrong/stupid? i'm just trying to verify the most important factor in a fighter's success. it sounds like you think it's looks. or is it something else?

Reread my post:

Someone 10 years younger or more will generally win. Someone with 6 inch reach advantage will generally win if its a standing fight (fuck you Shane Campbell) Home advantage is usually big in Brazil and Canada (for examples, judges yesterday sometimes made it 30-27 instead of a clear 29-28). Missing weight usually is a bad sign. Ring rust of 2 years or more"

THis does not have anything to do with looks. As an example, i think Swanson looked much more muscular than Holloway, and based purely on looks, i would have picked Swanson, but according to those stats, Holloway was the must pick.
But even then, i'm not advising anyone to bet based purely off those stats. But OP asked about stats for mma, so i'm answering. Those stats can help a bit, but they're not everything.

EDIT: This isn't proven stats, but i would even go as far as to say that if someone has "bad looks", his line will generally go under real value and its a good bet. But there isn't any real stats to back this up.

so you don't think looks are important?

On April 26 2015 17:38 bigredhoss wrote: so you don't think looks are important?

looks > game when it comes to fighters

lol

I suck at math.

When it comes to betting, I usually just point and click.

ty daut for verifying the importance of looks

On April 26 2015 14:38 julep wrote: another thought was if all information is available on something like basketball, then how does haralbos volgaris has a huge edge in betting? he must be using a different betting system to the bookmakers, or only bets when he thinks the line is way out

few things:

1. i don't think your premise is true about all the data being available. there's a ton of data available to the public, but even more collected data that's proprietary. Voulgaris has spent millions on data and programmers/analysts who work for him, and a lot of NBA teams spend millions on analytics these days.

2. maybe this is splitting hairs and it's all relative, but i don't think he has a 'huge edge', more like he has a small edge in a huge market, like a 0.5bb/100 winner at 2k/4k NL or something.

3. he's talked about his previous edges a bit (ones that obviously no longer exist). probably the most dubious one was that apparently bookmakers used to get their 2nd-half total #s by literally taking half of the full-game totals, for ex. if the o/u for the game was 204 they would set the o/u of the 2nd half for 102, not accounting for all the fouling and clock stoppages in the 4th quarter that often lengthens the game. he's also known for being one of the first guys to realize the value of the corner 3 (closer distance than other 3-pointers, and the second most efficient shot type after dunks/layups). there was also something about understanding some coach's tendencies that the books didn't account for. all these edges are long gone but still kind of interesting to read about, you can probably find more if you search some of his interviews.

4. books have gotten much sharper in recent years and Voulgaris was recently on a podcast where he said his results this year weren't that great (though it's not clear exactly what that means by his standards).

On April 26 2015 16:28 bigredhoss wrote: So Floofy do you pick fighters based on looks?

thats what i do

On April 27 2015 08:42 RaiNKhAN wrote: Show nested quote + On April 26 2015 16:28 bigredhoss wrote: So Floofy do you pick fighters based on looks? thats what i do

not for basketball though