Predicting soccer score

bagaspuds

Brainteaser as follows;

I want to try and predict the score in a soccer match.
I've used Poisson's Distribution which assumes equal chance of scoring at any time in the match based on the scoring rate of the team, goals per hour. I've done this - ok.
Now I want to incorporate the goal conceding rate into this also.
How do I do this?

Is it OK to take the average of both, i.e Team A scoring probability & Team B conceding probabilty.:rolleyes:

Any suggestions welcome !!

Fremen

Use a Skellam distribution, maybe?
It's defined as the difference between two Poisson RVs which aren't necessarily independent. The Wiki article specifically mentions that it's a good model for soccer scores.
http://en.wikipedia.org/wiki/Skellam_distribution

bagaspuds

Thanks Freman,

The first paragraph does say only independent variables are considered -

"Only the case of uncorrelated variables will be considered in this article. See Karlis & Ntzoufras, 2003 for the use of the Skellam distribution to describe the difference of correlated Poisson-distributed variables."

I guess I need a correlation factor between the scoring and conceding rates.

Anyone ????

dreamers75

Impossible tbh


The bounce of a ball can not be predicted regardless of how many times bounced.

you can get lucky tho and if you do paddypower.com tbh

Fremen

dreamers75 wrote: »

Impossible tbh


The bounce of a ball can not be predicted regardless of how many times bounced.

you can get lucky tho and if you do paddypower.com tbh

Yes, but the statistics of ball-bouncing, on the other hand...

OP, you'll need a spreadsheet and a load of effort to work that out. I doubt anyone knows offhand.

LeixlipRed

I think if something like this could work properly the bookies would be broke.*Insert bookie cliche here*

bagaspuds

It works quite well actually.
Heres the predicted score odds for the Aston Villa v Bolton match at the weekend. I used last years scoring rates and I compared them to the current odds on Betfair.:D

LeixlipRed

You'll eventually make massive losses. Don't worry

blackbetty69

doesnt seem possible but interesting notion all the same:)

cavedave

There are some great articles on the maths of soccer in plus. One here, here and here.

Sheeps

dreamers75 wrote: »

Impossible tbh


The bounce of a ball can not be predicted regardless of how many times bounced.

you can get lucky tho and if you do paddypower.com tbh

What the **** are you talking about? How can the bounce of a ball not be predicted?

bagaspuds

Thanks for all comments,

Firstly I do believe that the betting exchanges are highly effecient but I don't believe that they are anywhere near 100%. That is the odds at all times reflect the true propbability. People tend to bet intuitatively.

This for example is the probability of a 1-0 score versus time left in the match (Arsenal v's Bolton again). I don't have typical data from the market but I doubt it if if follows this curve closely.

I believe also that bookies shift their odds depending on the flow of money, and that statistics are only used as a starting pt.

LeixlipRed

Sheeps wrote: »

What the **** are you talking about? How can the bounce of a ball not be predicted?

Relax a little. What sort of conditions are we talking about here? Take football for example, a sport which I play competitively. There are a million variables interacting in terms of how a ball will bounce when kicked. For example, how hard it's kicked, what part of the ball is struck, does the player follow through or not, how heavy is the ball, how well made is the ball, what type of weather, what type of surface you're playing on. And many, many more. The more variables the more difficult to predict what's going to happen exactly.

However, if you're just talking about dropping a ball on the ground from a stationary postion indoors or something then it's easy enough. I believe the posters point was that football is an unpredictable game, you can't factor in human error. And average goals scored/conceded only gives a small portion of the picture.

cavedave

I presume the low scoring nature of soccer makes it less predictable. One bounce off a beach ball and you lose a game. Whereas in Hurling, rugby or other higher scoring sports the better team should win more often?

Macca07

LeixlipRed wrote: »

I think if something like this could work properly the bookies would be broke.*Insert bookie cliche here*

How do u think the bookies work it out??? Do you think they just give random odds

bagaspuds

Macca07 wrote: »

How do u think the bookies work it out??? Do you think they just give random odds

I think they adjust the odds so that they make money regardless of the outcome. They use statistics alright but that is just their starting point.

How else do you explain the odds moving on a horse race before the race starts!

Sheeps

LeixlipRed wrote: »

Relax a little. What sort of conditions are we talking about here? Take football for example, a sport which I play competitively. There are a million variables interacting in terms of how a ball will bounce when kicked. For example, how hard it's kicked, what part of the ball is struck, does the player follow through or not, how heavy is the ball, how well made is the ball, what type of weather, what type of surface you're playing on. And many, many more. The more variables the more difficult to predict what's going to happen exactly.

However, if you're just talking about dropping a ball on the ground from a stationary postion indoors or something then it's easy enough. I believe the posters point was that football is an unpredictable game, you can't factor in human error. And average goals scored/conceded only gives a small portion of the picture.

What ever about not being able to factor in complex variables in your prediction of the results of football, the bit about not being able to predict the trajectory of the ball is complete non-sense. You can do it with rather accuracy with just a few rather elementary level physics formulas.

LeixlipRed

So you're telling me bookies can accurately predict the final scores of football games are you?? They just calculate the probabilities using statistics initially (like our OP is doing). Then they give odds usually much lower than the probability. Also, a huge chunk of their money is made on things like accumulators where the actual probabilities are astronomical yet the odds are small enough to make it look like it's possible.

LeixlipRed

Sheeps wrote: »

What ever about not being able to factor in complex variables in your prediction of the results of football, the bit about not being able to predict the trajectory of the ball is complete non-sense. You can do it with rather accuracy with just a few rather elementary level physics formulas.

Can you do it accurately enough to within a small enough margin or error so that you can predict whether and offensive or defensive player will reach the ball first? Of course you can't. You're talking about perfect experimental conditions. Not a real life match.

Sheeps

LeixlipRed wrote: »

Can you do it accurately enough to within a small enough margin or error so that you can predict whether and offensive or defensive player will reach the ball first? Of course you can't. You're talking about perfect experimental conditions. Not a real life match.

Like I said, I'm not talking about being able to predict the score. I'm talking about predicting the trajectory of a sphere that has a force applied to it.

I was simply correcting you when you posted "The bounce of a ball can not be predicted regardless of how many times bounced.", that's all.

LeixlipRed

I didn't say that. And a ball is not a sphere

Sheeps

LeixlipRed wrote: »

I didn't say that. And a ball is not a sphere

Woops, well who ever it was who said it. I was correcting them, and it doesn't matter whether it's a solid sphere or not, it can still be modeled and calculated.

lintdrummer

I work in a bookies. They obviously use this kind of maths to set odds but they definitely take the amount of money being put on an outcome into account as well. They tend to revise the odds regularly.

bagaspuds

Ok thanks 'emmettogara' that clarifies it.

...anyway back to my original post (not the side show with the 2 boys) - is it ok to take the average

LeixlipRed

Sheeps wrote: »

Woops, well who ever it was who said it. I was correcting them, and it doesn't matter whether it's a solid sphere or not, it can still be modeled and calculated.

Nothing to do with it's solidity. Footballs simply aren't perfectly curved like a sphere. Just listen to the Nike people every year when they launch a new ball. "Rounder than ever" is a phrase they constantly use.

Sheeps

LeixlipRed wrote: »

Nothing to do with it's solidity. Footballs simply aren't perfectly curved like a sphere. Just listen to the Nike people every year when they launch a new ball. "Rounder than ever" is a phrase they constantly use.

Jesus why am I still discussing this with you? Who cares, the difference it makes is minuscule. You can still predict where the thing is going to go. Ever hear of a soft body physics engine? It can deal with imperfect spheres, hallow spheres, spheres that change shape, it doesn't matter. It's an approximation and it's accurate enough to be able to say that you can predict where a football will end up if you kick it.

LeixlipRed

In perfect experimental conditions only. Anyway, end of the debate.

Pherekydes

I just calculated the result of Chelsea-Man Utd using the aforementioned method. It works out at Chelsea 95 Man Utd 0. Sounds about right.

LeixlipRed

Shels v KC tonight. Kildare have gone bust just today so we could have a scoreline like that. Goalfests ftw

Azelfafage

Even Ada Lovelace thought she had figured out the odds.

Ada was the mathematical genius daughter of the poet Byron .

Ada was a true genius.

She was 70% responsible for Babbage's "Analytical Engine".

She was a gambler.

She died young:

http://en.wikipedia.org/wiki/Ada_Lovelace

.