Laying 0-0 and UQ C/S's

[SpodoKomodo]

Hi all, Long time lurker, first time poster. I am by no means an expert statistician, but I have been looking a lot into Poisson theorems, and despite the negative reports I have read on here regarding its use in CS betting, I am convinced there is some value to be found somewhere. From what I have seen, most Poisson theorems seem to overstate the chances of mid-range correct scores. So, to begin with, I am looking at the 2 extremes, laying no goals, and laying UQ. I'd like people to share their views if anyone has any on Poisson. I'll start posting bets soon!

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's I'd also like to state, that due to my lack of skills with Excel, I do not have an automatic "machine" in my spreadsheet which will calculate all the odds if I import the data of a match from a website or something. So I have to do everything manually. Too simplify things, I'm going to trial just betting (paper trialing) the CS which my system gives the most value to. All bets 1pt. Bet1: CS 0-0: Bury v Shrewsbury @ 9/1

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's Also throws up this as a value bet... I wonder how long it will be before the first winner ! :rollin Bet 2: CS 0-1: Rotherham v Torquay @ 14/1

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's Sorry, it was late and I was tired when I wrote this. To evaluate, my purpose in creating this thread was to: a) provoke debate regarding to the use of Poisson to find value in the CS market b) post my predictions using various Poisson systems, to try and see if there is value to be found System A (my two earlier predictions) is based upon previous resukts of both teams solely. I know this is basic, but it is not just both teams past 5 games taken into consideration or something like that - I've gone into it a bit more intricately and only taken data from certain results, to try and remove anomalies.

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's Hmm, trying a couple more for today. Spreadsheet throws up massive values on this: Bet 3: CS 2-3: Hoffenheim v Monchengladbach @ 50/1

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's Last one for today, let's see if I get close in any of them. Bet 4: CS 2-3: Roda v Heracles. I don't like the two 2-3 predictions - the Roda one at least has been skewed by the fact the top 3 in the league appear in the data used more than once. Ajax, PSV, Twente all won convincingly in previous games against these 2.

[clay747]

Re: Laying 0-0 and UQ C/S's

Hi all, Long time lurker, first time poster. I am by no means an expert statistician, but I have been looking a lot into Poisson theorems, and despite the negative reports I have read on here regarding its use in CS betting, I am convinced there is some value to be found somewhere. From what I have seen, most Poisson theorems seem to overstate the chances of mid-range correct scores. So, to begin with, I am looking at the 2 extremes, laying no goals, and laying UQ. I'd like people to share their views if anyone has any on Poisson. I'll start posting bets soon!

Hi SpodoKomodo! How do you calculate the lambda value(expected number of occurrences) for each team? The problem with the basic Poisson model is that it assumes independence between events(goals). But in football matches teams will make tactical changes when a goal is scored, such as committing more men to attack, or even replacing defensive players with attacking ones. Some teams will play for the draw and only start attacking when they go a goal behind. Bivariate Poisson seems to be a better solution to the problem. I also read one paper that tried to correct the basic Poisson model by calculating a correction value for each score using historical data. Anyway, good luck, i will be following your picks with interest.

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

Hi SpodoKomodo! How do you calculate the lambda value(expected number of occurrences) for each team? The problem with the basic Poisson model is that it assumes independence between events(goals). But in football matches teams will make tactical changes when a goal is scored, such as committing more men to attack, or even replacing defensive players with attacking ones. Some teams will play for the draw and only start attacking when they go a goal behind. Bivariate Poisson seems to be a better solution to the problem. I also read one paper that tried to correct the basic Poisson model by calculating a correction value for each score using historical data. Anyway, good luck, i will be following your picks with interest.

Hi Clay, My lambda value is based upon goals scored and conceded in previous matches for each team. I won't go completely into it, but it is based on home and away previous results for the home team, and just previous away results for the away team. I'll give an example. Chelsea v Arsenal Chelsea average goals scored (AGS): 1.5 Chelsea average goals conceded (AGC): 1 Arsenal AGS: 1.6 Arsenal AGC: 1.8 Chelsea goals = (1.5+1.8)/2 = 1.6 Arsenal goals = (1.6+1)/2=1.3 These 2 values go in tho my Poisson formula, to calculate the probability of how many goals each team will score in the match. I then multiply the chances to create odds for each scoreline (e.g. Chelsea chance to score 1 goal mutiplied by Arsenal chance to score 2 goals = odds for 1-2 CS). I realise I have just explained the formula (I think) but hopefully you can confirm that I have the right end of the stick here? :tongue2 Note: I do not just use goals scored and conceded in previous matches for each team, as I realise this would be too simple. I'm just treading water looking for possible value. I understand what you mean about independence...that is one of the reason I am looking for ways to interpret the data in a different way. I'm not really familiar with Bivariate Poisson, do you have any good links to any articles or tutorials? If not I'll just Google it! And I'm not too sure what you meabt by your last statement (in bold). Could you possibly expand? ;) I started looking at this after following the Poisson CS predictions of "fish" on the betfair forum. He had some astounding winners, and a very good strike rate. He stopped posting after a while because he went through a dry patch and people started giving him stick for it.

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's After reading my post through, I realise that most CS Poisson predictions are based on previous goal stats, but what I meant to emphasise is that with the current model I am using, I am overstating the importance of the away team's results, in comparison with the home team's. It's just an experimental way of approaching it.

[clay747]

Re: Laying 0-0 and UQ C/S's

Hi Clay, My lambda value is based upon goals scored and conceded in previous matches for each team. I won't go completely into it, but it is based on home and away previous results for the home team, and just previous away results for the away team. I'll give an example. Chelsea v Arsenal Chelsea average goals scored (AGS): 1.5 Chelsea average goals conceded (AGC): 1 Arsenal AGS: 1.6 Arsenal AGC: 1.8 Chelsea goals = (1.5+1.8)/2 = 1.6 Arsenal goals = (1.6+1)/2=1.3 These 2 values go in tho my Poisson formula, to calculate the probability of how many goals each team will score in the match. I then multiply the chances to create odds for each scoreline (e.g. Chelsea chance to score 1 goal mutiplied by Arsenal chance to score 2 goals = odds for 1-2 CS). I realise I have just explained the formula (I think) but hopefully you can confirm that I have the right end of the stick here? :tongue2 Note: I do not just use goals scored and conceded in previous matches for each team, as I realise this would be too simple. I'm just treading water looking for possible value. I understand what you mean about independence...that is one of the reason I am looking for ways to interpret the data in a different way. I'm not really familiar with Bivariate Poisson, do you have any good links to any articles or tutorials? If not I'll just Google it! And I'm not too sure what you meabt by your last statement (in bold). Could you possibly expand? ;) I started looking at this after following the Poisson CS predictions of "fish" on the betfair forum. He had some astounding winners, and a very good strike rate. He stopped posting after a while because he went through a dry patch and people started giving him stick for it.

There are many things you can try when calculating the rate values, how many previous games to include, time decay on those results(more resent results count), head to head results and so on. You can also look at spread betting prices to see what values they are using. for example on Sporting Index, Everton v Man utd. Man Utd/Everton(h) 0.7 - 0.9 -- centre point (0.7+0.9)/2 = 0.8 Total Goals 2.6 - 2.8 -- centre point (2.6+2.8)/2 = 2.7 2.7 - 0.8 = 1.9 2.7 - 1.9 = 0.8 Man Utd Goal Rate = 1.9 Everton Goal Rate = 0.8 Put those values into your correct score Poission model and you will get something close to what they think the correct scores will be. Don't have any useful links in Bivariate Poission, I'm currently working my way through an Open University course on the applications of probabilty, which covers among other things Bivariate Poission. But its a third year maths course and not, what i would call easy reading! Lots of this :wall and after a two pages i want to do this :zzz!! The correction method is something like this, 1) Calculate the mean Home and Away goals from your historical data(one season of prem. games would 380 matches) 2) Uses these values in your poission model to estimate the probability of each score happening in the 380 matches. 3) calculate the actual probability of each score from your historical data. 4) your correction factor will be actual probability divided by the estimate probability. So if the data had more 1-1 scores then our estimates of the same data, we would have a correction value > 1, = 1 if they are the same, and

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's Getting a bit confused about what you mean about the correction data, clay. I'll have a re-read of what you said, and do some Googling, over the weekend. Tonight threw up no correct score predictions, but some interesting results nonetheless. Bury v Shrewbury 0-0 @ 9/1 - CS: 1-0 Rotherham v Torquay 0-1 @ 14/1 - CS: 1-1 Hoff v Moenchen 2-3 @ 50/1 - CS: 2-2 Roda v Heracles 2-3 @ 40/1 - CS: 2-1 So the 2 under predictions finished unders. And the 2 over predictions finished overs. That's not a bad sign! I could have also traded out on all of these for profit. 1 question I do have, is there any way anyone can suggest extracting the data I need, say from soccerway.com or similar, into my spreadsheet, so I do not have to enter the figures for each separate match manually? Thanks

[lunatism]

Re: Laying 0-0 and UQ C/S's

Getting a bit confused about what you mean about the correction data, clay. I'll have a re-read of what you said, and do some Googling, over the weekend. Tonight threw up no correct score predictions, but some interesting results nonetheless. Bury v Shrewbury 0-0 @ 9/1 - CS: 1-0 Rotherham v Torquay 0-1 @ 14/1 - CS: 1-1 Hoff v Moenchen 2-3 @ 50/1 - CS: 2-2 Roda v Heracles 2-3 @ 40/1 - CS: 2-1 So the 2 under predictions finished unders. And the 2 over predictions finished overs. That's not a bad sign! I could have also traded out on all of these for profit. 1 question I do have, is there any way anyone can suggest extracting the data I need, say from soccerway.com or similar, into my spreadsheet, so I do not have to enter the figures for each separate match manually? Thanks

:loon excel web queries - you're aware of the function, right?!!

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

:loon excel web queries - you're aware of the function, right?!!

Yeh I know the query, but I cannot find a website that imports the data in a neat, simple way.

[lunatism]

Re: Laying 0-0 and UQ C/S's

Yeh I know the query' date=' but I cannot find a website that imports the data in a neat, simple way.[/quote'] :rollin come on.....that would be toooooooo easy! you have to do a bit of fiddling around, and then be prepared to wait while you let excel do its stuff. Web queries in excel are limited only by the imagination of the user, you need help?

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

:rollin come on.....that would be toooooooo easy! you have to do a bit of fiddling around' date= and then be prepared to wait while you let excel do its stuff. Web queries in excel are limited only by the imagination of the user, you need help?

Maybe :eyes

[lunatism]

Re: Laying 0-0 and UQ C/S's

Maybe :eyes

hmmm...ok....what is your level of experience with Excel then?

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

hmmm...ok....what is your level of experience with Excel then?

Low to medium I'd say.... I know how to do a web query, I just don't know what website I can use to extract the data - all I really want is past 20 or so results of each team, split into home and away. What I want to do is update the web query with whatever match I want to study, which will then update everything in my spreadsheet to give me the data I want for that match. For example, Arsenal are playing Chelsea, so I update the web query to show the stats for these 2 teams, then that updates my spreadsheet to give me the correct odds (the odds which I have determined anyway). Does this make sense? :unsure

[lunatism]

Re: Laying 0-0 and UQ C/S's

Low to medium I'd say.... I know how to do a web query, I just don't know what website I can use to extract the data - all I really want is past 20 or so results of each team, split into home and away. What I want to do is update the web query with whatever match I want to study, which will then update everything in my spreadsheet to give me the data I want for that match. For example, Arsenal are playing Chelsea, so I update the web query to show the stats for these 2 teams, then that updates my spreadsheet to give me the correct odds (the odds which I have determined anyway). Does this make sense? :unsure

Ok, you know how to do a web query, i'm guessing you may have not played around with VBA specifics of web queries or XMLHTTP scrapes (which can be FAR more efficient if looking for specific data)? If you want to learn, a quick search on google for XMLHTTP and parsing data into excel would provide a lot of useful links. To cycle from one match to the next, it's a good idea to set up some vba script that repeats a certain action (that would be the scrape) a certain number of times (that would be the number of fixtures that you need the data for), and then each time you download the data, you relist with just the fixtures lacking the scraped info.....makes sense? ;)

[kmabet]

Re: Laying 0-0 and UQ C/S's

Ok' date= you know how to do a web query, i'm guessing you may have not played around with VBA specifics of web queries or XMLHTTP scrapes (which can be FAR more efficient if looking for specific data)? If you want to learn, a quick search on google for XMLHTTP and parsing data into excel would provide a lot of useful links. To cycle from one match to the next, it's a good idea to set up some vba script that repeats a certain action (that would be the scrape) a certain number of times (that would be the number of fixtures that you need the data for), and then each time you download the data, you relist with just the fixtures lacking the scraped info.....makes sense? ;)

I there anything that this man (lunatism) doesn't understand???!!! Spodo, good idea but you're not covering all angles with this. Pay attention to the details and have a closer look at what you are doing? A good idea for a close may be, Place 0-0 HT lay on HT correct score market. Place 0-0 FT back on FT correct score market. Dutch accross 2 markets. If you lose the half time (still 0-0), you have more options from that position. Theoretically, you are in a winning situation with purchased equity.

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

Ok, you know how to do a web query, i'm guessing you may have not played around with VBA specifics of web queries or XMLHTTP scrapes (which can be FAR more efficient if looking for specific data)? If you want to learn, a quick search on google for XMLHTTP and parsing data into excel would provide a lot of useful links. To cycle from one match to the next, it's a good idea to set up some vba script that repeats a certain action (that would be the scrape) a certain number of times (that would be the number of fixtures that you need the data for), and then each time you download the data, you relist with just the fixtures lacking the scraped info.....makes sense? ;)

Thanks for the info lunatic, I'm going to do some googling on those topics now :ok and kmabet, thank you for your advice, at the moment I'm just trying to understand some different statistical modelling to be honest. Don't know why I made that title to be honest, I should have just called the thread "Experimenting using Poisson and other statistical models", or something :tongue2 lunatism, you may be interested in this, it's a very good read and something I personally will have to read a few more times before i get my head round it: http://www.ylikerroin.com/file/Complete.pdf[url=http://www.ylikerroin.com/file/Complete.pdf]

[lunatism]

Re: Laying 0-0 and UQ C/S's

lunatism' date= you may be interested in this, it's a very good read and something I personally will have to read a few more times before i get my head round it: http://www.ylikerroin.com/file/Complete.pdf

That's an interesting looking book - but 95 pages!! :loon will have to put aside some time to take a look through it

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

Ok, you know how to do a web query, i'm guessing you may have not played around with VBA specifics of web queries or XMLHTTP scrapes (which can be FAR more efficient if looking for specific data)? If you want to learn, a quick search on google for XMLHTTP and parsing data into excel would provide a lot of useful links. To cycle from one match to the next, it's a good idea to set up some vba script that repeats a certain action (that would be the scrape) a certain number of times (that would be the number of fixtures that you need the data for), and then each time you download the data, you relist with just the fixtures lacking the scraped info.....makes sense? ;)

Hi again lunatism, I have not given up or anything, but I have decided that for now, learning vba script is something I don't have time to do. I have done research into it, and it does seem a great time-saver in the long run, but I'm going to concentrate on the actual system for now. I'm going to create a new thread because this has gone off-track and I am really getting into Poisson - I think we should have an official Poisson discussion thread of some kind :ok

[lunatism]

Re: Laying 0-0 and UQ C/S's

Hi again lunatism, I have not given up or anything, but I have decided that for now, learning vba script is something I don't have time to do. I have done research into it, and it does seem a great time-saver in the long run, but I'm going to concentrate on the actual system for now. I'm going to create a new thread because this has gone off-track and I am really getting into Poisson - I think we should have an official Poisson discussion thread of some kind :ok

:tongue2 you don't actually have to learn it, all you need to do is have access to the scripts, and i could easily send you some in text files. Since...this is more than just a timesaver (well, actually i guess it isn't), but saving you lots of free time to do ANYthing...such as spending more time working on betting theories is a much better use of time than doing something manually that needn't be :loon I could take a look at it if you want....

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

:tongue2 you don't actually have to learn it, all you need to do is have access to the scripts, and i could easily send you some in text files. Since...this is more than just a timesaver (well, actually i guess it isn't), but saving you lots of free time to do ANYthing...such as spending more time working on betting theories is a much better use of time than doing something manually that needn't be :loon I could take a look at it if you want....

Thanks lunatism! :loon So let me know if I've got this straight - you could send me the script that I can import into my spreadsheet, which will detect, for example, the match in the column to the left, and give me the odds or stats?! :eek

[lunatism]

Re: Laying 0-0 and UQ C/S's

Thanks lunatism! :loon So let me know if I've got this straight - you could send me the script that I can import into my spreadsheet, which will detect, for example, the match in the column to the left, and give me the odds or stats?! :eek

Well, there's a couple of options - the first is that i send you a couple of text files, but these will only be templates, you'd have to modify them yourself in order for it to work with your spreadsheet.... Second option being you send me the spreadsheet and i do the modifying myself. Second option easier, but first option means you get to play around with stuff

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's

Well, there's a couple of options - the first is that i send you a couple of text files, but these will only be templates, you'd have to modify them yourself in order for it to work with your spreadsheet.... Second option being you send me the spreadsheet and i do the modifying myself. Second option easier, but first option means you get to play around with stuff

Great, thanks! :loon So these text files - will I understand them? Can you expand at all? I they just text instructions on how to code the VBA? I understand XML if that makes it any easier to understand the conversiont that will be required? I'm at work at the moment so can't send my spreadsheet - I'll do that when I get in.

[lunatism]

Re: Laying 0-0 and UQ C/S's

Great, thanks! :loon So these text files - will I understand them? Can you expand at all? I they just text instructions on how to code the VBA? I understand XML if that makes it any easier to understand the conversiont that will be required? I'm at work at the moment so can't send my spreadsheet - I'll do that when I get in.

Hmm....that's a point - for people "in the know", i guess it would be fairly simple to understand, can't be sure for others. Maybe the second option would be better, and then we could open up the code together afterwards and i'd explain it, so that you'd know for future reference?

[SpodoKomodo]

Re: Laying 0-0 and UQ C/S's Ok sounds good! My email is [email protected] If you could send me the txt files, that would be great! I'll then have a read of them on the off-chance I completely understand them! Then I can forward my spreadsheet to you? Cheers :ok