This forum is read-only now. Please use Forum 2 for new posts

xml No replies possible in the archive
About football and statistical predictions
Author: cska
Date: 11-04-2006, 12:08
Dear fans,
This topis is created, because I feel that some fans, like Aramand, posted statistical predictions in other threads that were not exactly dedicated for this.
Let me put a remark (you may see also my post in thread Chelsea & Bayern, appears nearly at the bottom) that if you want probabilities about game outcomes, then consult the statisticians who work with the bookmakers. Have you seen a bookie gone bankrupt ? Have you seen a person, who bets, become a millionaire ?
Bookmakers experts really can count and apply statistics to assign probabilities. That's their job. I advise Aramand to get in contact with them.
About probabilities and predictions, I can only say that statistics requires that all possible events should be assigned probabilities, which add up to 100%. So, Aramand, if you just say that team X will be champion of A with 95% probability, then it is out of thin air, if you don't explain which events you assign the rest 5% and how exactly.
And remember, football is fun, not maths ! By the way, I won a silver medal on a mathematical olympiad for university students. So, I really know what probabilities mean.

Re: About football and statistical predictions
Author: Edgar
Date: 11-04-2006, 12:20
I expect Aramand to tell us he won a gold medal

Re: About football and statistical predictions
Author: ferdi
Date: 11-04-2006, 12:38
Bookmakers are not interested in probabilities of match outcomes, but in the betting behaviour of those who bet. If for example two of three betters bet on team A, whereas one of three bets on team B, then a bookmaker would pay less then 15:10 if team A wins, and less then 30:10 if team B wins. The probabilities of outcome of that match do not affect the bookmakers calculations at all.

Re: About football and statistical predictions
Author: dawgs
Date: 11-04-2006, 12:42
The probabilities of match outcomes in the heads of football fans, however, do influence bookmakers' odds. Exactly the point cska is trying to make, as much as I can follow it.

Re: About football and statistical predictions
Author: Ricardo
Date: 11-04-2006, 12:52
Ferdi is having a point there. If you bet in England on a match of England versus .... The expectance of a English win is always higher than other reasons will give. That higher expectance of the betters is taking into account by the bookmakers. But this result in an incorrect result regarding real chances of being victorious.

Re: About football and statistical predictions
Author: ferdi
Date: 11-04-2006, 12:56
Edited by: ferdi
at: 11-04-2006, 13:10
There is no mechanism which ensures, given for example that the probability of outcome would be 67%:33%, that in this case 67% of the betters will bet on team A and 33% will bet on team B. It could as well be that 75% decide to bet on the favorite team and 25% decide to bet on the outsider. In this case the relevant infomation for the bookmaker is 75:25, and not 67:33. Otherwise the bookmaker himself would start to bet.

Re: About football and statistical predictions
Author: dawgs
Date: 11-04-2006, 13:10
I fear that I'm way over my head posting in this thread to begin with... But I thought Aramand's procedure did involve him assigning arbitrary (not "absolute") numbers into his calculations. Gold, silver, bronze, etc are examples of things that he has in his head, but many others disagree with, therefore even if his testing is not flawed, it is not the Gospel either.

All fans across the world do similar, if not as complex calculations before they bet, therefore he is just yet another one, if more pretencious wanna-be prophet of football games outcome.

Bah, I should stick to topics I feel more comfortable with.

Re: About football and statistical predictions
Author: MalcolmW
Date: 11-04-2006, 14:52
I rarely bet at all, though when Bolton reached the last 32 of the UEFA Cup I decided to place a bet that would cover the cost of a trip to the final.
This was immediately after the last 32 had been decided, and before the draw was made, I backed Bolton at 25:1. Looking through the odds from UK bookmakers I saw that Steaua was 50:1. That was just too good to miss, so I placed a further half of the original stake on them. Looking further down - Rapid 100:1(!!), so a further quarter of the original stake went on.
I figured that, whatever Bolton's real chances, Steaua's were better and Rapid's no less than Bolton's. Perhaps the true odds at that stage were 10:1 against Steaua and 40:1 against Bolton and Rapid. But the bookmakers odds reflected the UK market expectations, so Middlesbrough and Bolton were over rated and the Romanians, amongst others, under rated.
Of course I was disappointed when the Bucharest teams were drawn to meet in a quarter final. Now that we are down to 4 I was amazed to find that Steaua are regarded as the rank outsiders! I can understand the Romanian frustrations at the disdain with which their progress is regarded, but my advice is simply to be patient (or bet with UK bookmakers).

Re: About football and statistical predictions
Author: cska
Date: 11-04-2006, 19:54
Dear all,
Some of you have the point to say that bookies try to reflect market expectations. However, nobody can predict what exactly the betting proportions would be. On the other hand, bookies really calculate probabilities and hire mathematicians (mainly from insurance companies, because the principles of betting are similar to insurance - in fact when you bet you buy a policy that a match will not finish as you predict and if it does, then the bookie "compensates" you against the "disaster").
Just an example, there's a match soon: Middlesborough-Charlton (FA Cup) and one of the bookies gives 2,10-3,10-3,30 (1-X-2). You can calculate how much you must bet to have 100% chance to receive 1 unit (pound, dollar, etc):
You place 1 / 2,10 = 0,48 for Home (1); 1 / 3,10 = 0,32 for Draw (X) and 1 / 3,30 = 0,30 for Away (2). So, no matter how the match finishes you get 1 unit in return. (multiply 0,32*3,10 = 1 for example). However, 0,48 + 0,32 + 0,30 = 1,10 , not 1. This means that you must bet totally 1,10 to get 1 at 100% probability. These 10% are the margin for the bookie. However, you can use this margin to calculate what is the probability that the bookie expexts:
0,48 * 1 / 1,10 = 43,6% Home ; 0,32 * 1 / 1,10 = 29,1% Draw ; 0,30 * 1 / 1,10 = 27,3% Away. See that they ad up to exactly 100%.
By the way, if some English Conference or some Maltese teams play, how can the bookie predict how many fans they have to predict "market expectations". Bookies always use statistical tools to assign probabilities, but they "skew" the balance if one of the teams is a popular giant. For example, Manchester Utd odds are always biased, because Manchester Utd has many fans, who normally bet on it. So, the bookie lowers the coefficient for Man Utd to increase the margin on account of fans' emotions.