Bookmaking Theory

Hello All,

I have read the forum with interest for some time, but this is my first post and I hope that someone could possibly steer me in the right direction.

I work as a trader for an online bookmaker, with the majority of my work being focused on pricing up (setting the odds) for football (soccer) matches.

For the most part when it comes to pricing up these matches I essentially have a rough idea in my head of what the prices should be and then adjust those prices in accordance to what the rest of the industry is showing. However I feel that by doing this, I am not making the most of the information available to me and feel that it can be done in a far more clinical manner.

My question is simply - How would you calculate the true percentage chance of a home win, a draw and an away win?

What factors would you hold the greatest weight to and how would you utilise them?

For example, would you look at the league average for a home win, draw and away win and then combine that with each individual teams recent records, also taking into account goal scoring averages? If so how would you do this?

Would you use the respective goalscoring averages in conjunction with Poisson distribution and then take the percentage indicated by specified correct scores and build a percentage chance of a home win, draw and away win from that? (add up the % chance of 1-0, 2-0, 2-1, 3-0, 3-1 and 3-2 for each side for their % win chance and then 0-0, 1-1, 2-2 and 3-3 for the % draw chance)

Would you look at goalscoring supremacy and adjust their goal averages in accordance to whether they were playing at home or away? (Team A average 1.76 goal overall, but 1.89 when at home)

Any help that you could give me would be very much appreciated.


Many thanks

Trisk

Quote:

---

Originally Posted by Triskaedekaphobita

Hello All,

I have read the forum with interest for some time, but this is my first post and I hope that someone could possibly steer me in the right direction.

I work as a trader for an online bookmaker, with the majority of my work being focused on pricing up (setting the odds) for football (soccer) matches.

For the most part when it comes to pricing up these matches I essentially have a rough idea in my head of what the prices should be and then adjust those prices in accordance to what the rest of the industry is showing. However I feel that by doing this, I am not making the most of the information available to me and feel that it can be done in a far more clinical manner.

My question is simply - How would you calculate the true percentage chance of a home win, a draw and an away win?

---

As a bookmaker you are not necessarily interested in the "true" probabilities
of the possible outcomes, you are setting the book to maximise your take.

So you might start with estimates of the probabilities, but you are going to
adjust these as you go to reflect the way the betting is going.

(If we are talking soccer here, there are not really enough results in a
season, and the statistics are not stationary from season to season,
to develop a good model, at least not much better than gut feeling)

Oh, and if we really knew the answer to your question we would not be
telling, instead we would be taking the bookies to the cleaners every week.

RonL

Captain,

Thanks for getting back to me.

When it comes to maximising my take and trading positions I am very comfortable in that part of my job. I have a clear and thorough understanding of reacting to market forces and trading books, but I am lacking in the general knowledge of creating a "tissue" or starting point from which to trade from.

Understanding a true probability is important as far as this goes, as I don't want to trade myself into a point where I am offering a price which is out of kilter with what the true % chance of an event happening.

I just want to have a decent process that I can use to calculate a % chance for the three outcomes of a football (soccer) match. There has to be a sensible approach that I can use to finding this out, as the price for each event has to come from somewhere - they don't just magic themselves up from nowhere.

Bigger firms come out with their prices before smaller firms like the one I work for, and I form my opinion around what their prices are. But I don't want to act like a sheep and follow the other firms, and would much prefer to have an educated argument behind being the top price for a particular event that can be backed up with something more than "Cos I think the price should be a lot bigger than it is".

Having an overround (the % we take for our advantage) protects us from any glaring errors. It doesn't have to be 100% accurate, but I just need a process that I can use to say that Team A has roughly a 40% chance, the draw is roughly a 30% chance and Team B is roughly a 30% chance.


Any help would be much appreciated.


Trisk

Quote:

---

Originally Posted by Triskaedekaphobita

Captain,

Thanks for getting back to me.

When it comes to maximising my take and trading positions I am very comfortable in that part of my job. I have a clear and thorough understanding of reacting to market forces and trading books, but I am lacking in the general knowledge of creating a "tissue" or starting point from which to trade from.

Understanding a true probability is important as far as this goes, as I don't want to trade myself into a point where I am offering a price which is out of kilter with what the true % chance of an event happening.

I just want to have a decent process that I can use to calculate a % chance for the three outcomes of a football (soccer) match. There has to be a sensible approach that I can use to finding this out, as the price for each event has to come from somewhere - they don't just magic themselves up from nowhere.

Bigger firms come out with their prices before smaller firms like the one I work for, and I form my opinion around what their prices are. But I don't want to act like a sheep and follow the other firms, and would much prefer to have an educated argument behind being the top price for a particular event that can be backed up with something more than "Cos I think the price should be a lot bigger than it is".

Having an overround (the % we take for our advantage) protects us from any glaring errors. It doesn't have to be 100% accurate, but I just need a process that I can use to say that Team A has roughly a 40% chance, the draw is roughly a 30% chance and Team B is roughly a 30% chance.


Any help would be much appreciated.


Trisk

---

Choose some measurable property of the teams (position in league or some
other measure of form) Then for some set of bins [a_i,b_i]x[a_j,b_j] of this
property compile historical data on the results of matches in these bins, and
if your match falls in a given bin use the historical proportions for results
from that bin as your prediction of the result of the current match.

RonL

Quote:

---

Originally Posted by Triskaedekaphobita

Captain,

Thanks for getting back to me.

When it comes to maximising my take and trading positions I am very comfortable in that part of my job. I have a clear and thorough understanding of reacting to market forces and trading books, but I am lacking in the general knowledge of creating a "tissue" or starting point from which to trade from.

Understanding a true probability is important as far as this goes, as I don't want to trade myself into a point where I am offering a price which is out of kilter with what the true % chance of an event happening.

I just want to have a decent process that I can use to calculate a % chance for the three outcomes of a football (soccer) match. There has to be a sensible approach that I can use to finding this out, as the price for each event has to come from somewhere - they don't just magic themselves up from nowhere.

Bigger firms come out with their prices before smaller firms like the one I work for, and I form my opinion around what their prices are. But I don't want to act like a sheep and follow the other firms, and would much prefer to have an educated argument behind being the top price for a particular event that can be backed up with something more than "Cos I think the price should be a lot bigger than it is".

Having an overround (the % we take for our advantage) protects us from any glaring errors. It doesn't have to be 100% accurate, but I just need a process that I can use to say that Team A has roughly a 40% chance, the draw is roughly a 30% chance and Team B is roughly a 30% chance.


Any help would be much appreciated.


Trisk

---

The sensible approach you're for asking starts with power ratings. See how betting lines and point spreads are made and making power ratings as examples. How to make and use power ratings is an art where you can use as much statistical technique, historical data, and knowledge of the sport and the betting on it as you want. There may be books to help you get started with this. I have only read such books on horse racing. You might call the Gamblers Book Shop in Las Vegas and ask what books they have on making power ratings.