Announcements
*** June Competition Winners: Well done to Le Male (NAPS), Coco69 (KO Cup), Saddlesore (Royal Ascot) & Like2Fish (Poker) ***
*** New Last Man Standing Competition - Win up to £1500 Annually - See Competitions Forum ***
nikoloout

Lets explain Dixon and Coles

Recommended Posts

I am studing for a while the theory of Dixon and Coles and i can't make it work.I will be glad if anyone here have the knowledge to share with us the way... Lets start....If we have only 2 teams...for example Chelsea and Arsenal ok?We have only one match ..Home team is Chelsea and away team is Arsenal. Can anyone help us with the maths..

Share this post


Link to post
Share on other sites

Re: Lets explain Dixon and Coles I also cannot contribute much, but I'd like to recommend paper "Assessing the number of goals in soccer matches", a Master’s Thesis by Rasmus B. Olesen made at Aalborg University in the Machine Intelligence department; available, for example, here: http://projekter.aau.dk/projekter/files/14466581/AssessingTheNumberOfGoalsInSoccerMatches.pdf He used 4 prediction methods aiming to

The goal of this project is to, in a number of different approaches investigate the correlation between a soccer teams history results and the probability distribution for the number of goals in a given soccer match
Hence, his investigation covers only number of goals, not actual outcome of the match, but it's worth reading anyway; his conclusions:
the test an evaluation has shown, that the Dixon-Coles model is not as well suited for predicting the number of goals as it is at predicting the actual outcome. The over-estimation of the probability of low-scoring games, caused the model to show the worst prediction scores of the four assessors. .... It can therefore not be concluded that any of the assessors proposed was better than the bookmaker, nor can it be said that they in general are significantly worse.
I might be wrong, but, being a long-term fan of systems that rely on statistics, I believe that statistics solely cannot yield profitable results, other factors must be taken into consideration as well.

Share this post


Link to post
Share on other sites

Re: Lets explain Dixon and Coles The poission distribution is: ab327bb7e6cf7f72071e92b9e3ba42ee.png Suppose you have Team Chelsea which scores 1.4 goals average and Arsenal 1.2. Then lambda_Arsenal = 1.2 and lambda_Chelsea= 1.4. So now you can calculate f(0;lambda_Arsenal)*f(0;lambda_Chelsea) to score 0:0, f(1;lambda_Arsenal)*f(0;lambda_Chelsea) to socre 1:0 etc. This will now lead to slightly wrong estimation of low results 0:0, 1:0, 0:1, 1:1. Higher results are fairly ok. Now Dixon and Coles invented a tweak parameter to shift results, I think its from 0:1, 1:0 towards to 0:0, 1:1. The whole process still remains poisson distributed, see (4.2) in their paper. This parameter is rho, it cannot be estimated by a single match, thus need more. I also need a boundary between, so then whole process still remains poisson. The whole optimization is done with a mathematical routine. We can use a EM algorithmn, or better since we could estimate the derivatives some inner point algorithm. We don't need to understand these algorithms, we just feed them. Its like a compiler, you don't need to know whats going on inside to use one.

Share this post


Link to post
Share on other sites

hi. 

This is from wikipedia 

"In statisticsSpearman's rank correlation coefficient or Spearman's rho, named after Charles Spearman and often denoted by the Greek letter \rho (rho) or as r_s, is a nonparametric measure of statistical dependence between two variables. It assesses how well the relationship between two variables can be described using a monotonic function. If there are no repeated data values, a perfect Spearman correlation of +1 or −1 occurs when each of the variables is a perfect monotone function of the other."

You can see here, https://en.wikipedia.org/wiki/Spearman's_rank_correlation_coefficient, more.

 

Share this post


Link to post
Share on other sites
On 28/05/2015, 13:32:16, froment said:

Re: Lets explain Dixon and Coles I also cannot contribute much, but I'd like to recommend paper "Assessing the number of goals in soccer matches", a Master’s Thesis by Rasmus B. Olesen made at Aalborg University in the Machine Intelligence department; available, for example, here: http://projekter.aau.dk/projekter/files/14466581/AssessingTheNumberOfGoalsInSoccerMatches.pdf He used 4 prediction methods aiming to

Hence, his investigation covers only number of goals, not actual outcome of the match, but it's worth reading anyway; his conclusions: I might be wrong, but, being a long-term fan of systems that rely on statistics, I believe that statistics solely cannot yield profitable results, other factors must be taken into consideration as well.

"Statistics alone cannot yield profitable results....." !!?

Golly. This is a gem of a quote. 

If this is true then why bother with statistics at all?

Why not just use the other factors?

What are these other factors?

Share this post


Link to post
Share on other sites
On 4/25/2017 at 11:54 AM, moggis said:

"Statistics alone cannot yield profitable results....." !!?

Golly. This is a gem of a quote. 

If this is true then why bother with statistics at all?

Why not just use the other factors?

What are these other factors?

If it is not true, then why there are odds at all?!

You just have a look at the league table, the team that has better statistics (whatever you define it is) always wins, and that's it.

Share this post


Link to post
Share on other sites

Dixon and Coles assumed the home goals expected was( home attacking ability x away defensive weakness x home advantage factor) home advantage can be seen from the more goals the teams score at home.the expected away number of goals was the away teams attacking ability multiplied by the home team defensive weakness.no extra advantage for away team.

Share this post


Link to post
Share on other sites
On 07/05/2017 at 12:59 PM, froment said:

If it is not true, then why there are odds at all?!

You just have a look at the league table, the team that has better statistics (whatever you define it is) always wins, and that's it.

Its probably because I've been up all night but I can't get my head round what you're saying. It's the use of the negative that's confusing me even though it shouldn't of course.There are odds because without odds the bookies would lose money .

As far as league tables are concerned yes obviously using ststistics that everyone can see and knows doesn't make a profit.

So I will just repeat, if it's impossible to make a profit from betting from statistics alone ,what's the purpose of using statistics at all ? Why not just use the other factors ? 

Im genuinely simply asking what the purpose is of statistics in gambling if they don't make you a profit ?

And what are the other factors ?

These may be thought stupid questions but I'd have thought that any beginner reading this might like to know the answer. 

 

Because after all finding relevant statistics is quite tedious work and some people don't want to do it and do if they perhaps aren't necessary thst would be a big help.

 

 

 

Share this post


Link to post
Share on other sites
On 1/23/2018 at 3:18 PM, moggis said:

Its probably because I've been up all night but I can't get my head round what you're saying. It's the use of the negative that's confusing me even though it shouldn't of course.There are odds because without odds the bookies would lose money .

As far as league tables are concerned yes obviously using ststistics that everyone can see and knows doesn't make a profit.

So I will just repeat, if it's impossible to make a profit from betting from statistics alone ,what's the purpose of using statistics at all ? Why not just use the other factors ? 

Im genuinely simply asking what the purpose is of statistics in gambling if they don't make you a profit ?

And what are the other factors ?

These may be thought stupid questions but I'd have thought that any beginner reading this might like to know the answer. 

 

Because after all finding relevant statistics is quite tedious work and some people don't want to do it and do if they perhaps aren't necessary thst would be a big help.

 

 

 

This is old thread, so I had to remind myself of what is the subject of argument between you and me... :)

It seems it started when you questioned my statement "Statistics alone cannot yield profitable results". I still hold on to that statement. If predictions that are based purely on the stats could be profitable, everyone would make profit from betting - you just analyse previous seasons to find a profitable pattern, apply it to next season, and that's it, you earn money! :)

But it's not the case, so you must include those "other factors" you insist on, and they are highly subjective - referee, weather conditions, absent players, distance between cities, fatigue of one of the teams due to tight schedule, pitch conditions, etc etc...

So I don't really get what you want to say "Why not just use the other factors?" - you are free to use whatever you want.

Share this post


Link to post
Share on other sites

Interesting, please allow me to jump in :)

First of all, I must say I have the same position as Froment. Statistics is the backbone of the entire betting world but unfortuntelly it is not enough for us to win on long periods. You can not win against statistics using ...statistics.

Bookies use statistics - how many players, in percents, will back, for a game like Real - Vigo, 1,x,2,...etc in order to offer the "correct odds" for they? They will look back on their registers for this info. This means statistics.

Players use statistics - 95% of the palyers, including me, never have been closer than 1000 km to Madrid. Most of us knows nothing about 90% of the Madrid players.Their life, their interests, their problems. We just look at the table from time to time. Real won,used to :)  at least, over 95% of the games at home and have players that worth 1 billion. It seems a very good reason to win again ...... This is based on pure statistics only.

Like Froment said, this is not enough because everyoane see the same picture. It is like a running contest where all runners are holding hands. We just must have some info to decide if the odds provided are valuable or not.  Here may be a long discussion and it is not the current subject.

Regarding the question " if it's impossible to make a profit from betting from statistics alone ,what's the purpose of using statistics at all ? Why not just use the other factors ? " , my answer would be

because it is the base of everything else.

Let-s say you have to find the correct probability for Real to win the game.   First of all we have to look at tables (statistics) to make an aproximation (all begginers do that so no advantage here) then to "fine tune" the results based on other information: good players status, mood..etc, refree, pitch...etc  After you process these info and apply them over plane statistics will know if the odds offered are good to play or not.

If you use just information you will not be able to answer to question: Why 1.23 is good to play and 1.15  not? 

 

 

 

 

 

 

 

Edited by giraldi

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.