Poisson Distribution - This Can Help You

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Tue Jan 16, 2018 10:56 am

We move on to round 24, after which there is another FA Cup weekend so slightly longer until round 25 at the end of the month. Again, views on progress (past and going forward) would be appreciated.

Image
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Mon Jan 22, 2018 11:32 pm

The results from week 24 saw 25 goals with 3 over 2.5 goals, 5 BTTS Yes, and 5 Wins to nil (3 home 2 away).

Average Goals top.
4 5 1 2 2
Average Goals bottom.
1 4 2 2 2
Over 2.5 Goals.
4 5 1 2 2
Under 2.5 Goals. (The quoted figures are for OVER so deduct from 100)
1 4 2 2 2
Home teams winning to nil.
Won Won Draw Yes Yes Draw Lost Draw Lost Yes
Away teams winning to nil.
Yes Draw Lost Yes Draw Lost Lost Draw Lost Lost
BTTS Yes.
Lost Lost Won Won Won
BTTS No. (The quoted figures are for YES so deduct from 100)
Won Won Lost Won Lost

The Home/Away goal ratio was adjusted slightly for week 23. This also led to a small increase in the expected goals total. The average goals per match is now 2.675, still a shade below overall expectations.

This is another decent set of results so the format may be settling down. All teams that played at home in the previous set of fixtures were away in this set, so that should give a much better picture over the two sets of results.
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Tue Jan 23, 2018 6:18 pm

Poisson - updated figures.

Up to November 5th, there had been 11 complete rounds of matches, meaning 110 games played. Splitting the matches into 5 groups of 22 gave the results on page 10 of this thread. This table shows a comparison done the same way after 240 matches. (Rank, average goals prediction range, total goals, matches 3+ goals)

1-48 3.12 to 4.45 153 28
49-96 2.73 to 3.11 138 27
97-144 2.53 to 2.73 124 24
145-192 2.31 to 2.53 118 22
193-240 1.66 to 2.30 109 20

With 240 results in, that is now representing the correct pattern.

Over & Under 2.5 goals is in effect covered in the final column above so is not repeated here.

66 Home teams won to nil (Rank, % chance, W-0, Won BTTS, Drew, Lost).

1-48 37.03 to 60.91 26 11 7 4
49-96 27.78 to 37.02 17 9 14 8
97-144 22.71 to 27.77 16 6 15 11
145-192 16.09 to 22.70 4 7 18 19
193-240 5.24 to 16.08 3 7 11 27

That looks very good in view of the still relatively small sample when you bear in mind that it was not as close as this on the first post.

40 Away teams won to nil (Rank, % chance, W-0, Won BTTS, Drew, Lost).

1-48 27.32 to 53.08 19 9 19 9
49-96 19.98 to 26.72 6 9 8 17
97-144 15.83 to 19.87 9 5 20 14
145-192 7.46 to 15.40 6 4 9 29
193-240 2.68 to 7.33 0 1 10 37

Apart from the second row, this again looks fairly good. Only one away team with a W-0 expectancy less than 8% has won (52 matches). That was Burnley on the opening day at Chelsea, and there is no successful W-0 in that range.


111 matches have ended with BTTS Yes (Rank, % chance, BTTS).

1-48 52.71 to 64.24 22
49-96 48.85 to 52.49 26
97-144 45.38 to 48.63 25
145-192 41.28 to 45.33 21
193-240 23.63 to 41.25 17

A combination of more data, more matches with BTTS, and a couple tweaks have all helped make this look better than it was before.

The format was designed around an expectancy of 2.70 goals per game. The actual figure after 240 games is 642 which is now just 6 short of the 648 benchmark.

The 10 year averages based on 240 matches (excluding this season). This season.

Goals 650.6 642
Over 2.5 goals 123.9 121
Home W-0 62.8 66
Away W-0 35.6 40
BTTS 122.2 111
0-0 Draw 19.4 22

Although the overall sample size is still small, most sets of figures are beginning to 'feel' much closer than in the very early stages, and even substantially closer than in November. Consequently, I will not change a lot between now and the end of the season. Any changes that are made will be limited to team ratings and the weightings for previous seasons.

All figures will be reviewed at the end of the season. At that point any suggestions received (either then or earlier) may be amalgamated into what should become the working model for next season.
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Thu Jan 25, 2018 1:45 pm

I am in the process of writing a post wanting ideas for next season. What to include, omit, amend etc.

Please include a brief description of any change you would like and (if possible) the effect it may/should have. This would make planning any change easier to plan and implement.

Image
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

eduardoneco
Handicapper
Tips
Posts: 15
Joined: Sat Jun 25, 2011 11:00 pm

Re: Poisson Distribution - This Can Help You

Postby eduardoneco » Thu Jan 25, 2018 2:44 pm

Hello guys....

I have a question/suggestion/opinion about this kind of calculations. You make the calculations about the goals beens score and conceeded in the all matches and do the Poisson Distribuition. But this kind of calculations doesn´t take into account the strength of the opponent.

Example: Chelsea Vs Man City is completly different of Chelsea Vs Watford. And you use the same values on Chelsea to calculate the outcome. Do you see what I mean?

I think to a better calculations we need to divide the teams by classes. Teams to win Premier League - Class A; Teams to go in Europe - Class B; Teams to relegate - Class D and the rest Class C. And then, see the results against those teams and take those results into calculations.

Why do this? When you incorporate all the results you are adding deviations in the final result and making wrong decisions.

I see these kind of calculations like this. Sorry if I made a mistake somewhere.

Best regards.

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Thu Jan 25, 2018 4:34 pm

Hello guys....

I have a question/suggestion/opinion about this kind of calculations. You make the calculations about the goals beens score and conceeded in the all matches and do the Poisson Distribuition. But this kind of calculations doesn´t take into account the strength of the opponent.

This is hidden in the spreadsheet. In short, it compares Home For/ Away Against to Away For / Home Against and adjusts them via a Home team home rating v an Away team away rating. That is not ideal, but the closest I could conveniently program.

The estimated odds indicate that the overall program is not too far out, and it would be easy to force it to copy the bookies' odds. However, that would look great but achieve absolutely nothing. :D

Example: Chelsea Vs Man City is completly different of Chelsea Vs Watford. And you use the same values on Chelsea to calculate the outcome. Do you see what I mean?

Correct as stated above, but do you realise that Chelsea v Watford is different to Watford v Chelsea?

I think to a better calculations we need to divide the teams by classes. Teams to win Premier League - Class A; Teams to go in Europe - Class B; Teams to relegate - Class D and the rest Class C. And then, see the results against those teams and take those results into calculations.

How did we know at the beginning of this season that Man City would be the only team in class A?

Which of the 14 teams that have been involved in the relegation battle go into class D - and would you have included Everton at the start?

Why do this? When you incorporate all the results you are adding deviations in the final result and making wrong decisions.

I see these kind of calculations like this. Sorry if I made a mistake somewhere.

The real issue is this. All the major bookies have many highly paid people with very expensive equipment and access to a lot of data. I am doing this as a hobby and will never achieve a comparable end product. I simply want the most effective piece of software that I (along with input from others) can provide.

Best regards.
Hi,

I am actually in the process of composing a post that will give a complete 'state-of-play' update to what I have always said is a Work In Progress. All of what you say above is in effect taken into account in some form (although I couldn't tie in any current ELO figures that differentiated between home and away form - something that you have not mentioned).

By the end of the season, I want as many suggestions as possible for improvements etc, but more will be explained when that post is completed over the weekend.
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Wed Jan 31, 2018 4:39 pm

Image
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Wed Jan 31, 2018 4:47 pm

When I set out on this venture towards the end of last season, I had no real idea of what the finished article would (or indeed should) look like. Neither was I sure what aspects would be most suited to this approach even if I did have some ‘gut feelings’ about it.

A prototype was ready to go at the beginning of this season, but of course it had to rely solely on historic data at that point. The adopted solution was to use weightings for the past 3 seasons combined with data from this season, adjusting the weightings as the current figures provided a larger and possibly more reliable set of data.

At the end of this season, I want to analyse the results in phases, hopefully in line with the evolving of the program. To this end, I shall separate the first 6 weeks due to the lack of current data, and split the remaining 32 weeks into 8 batches of 4 rounds of matches.

I would appreciate any suggestions of what to include, amend, or delete in next season’s format. Please include reasoning and anticipated effect of any amendments as that would simplify both altering the program and monitoring the changes to see whether they had the desired effect.

The current formula is basically as follows.
1. Expected goals Home For and Away Against.
2. Expected goals Away For and Home Against.
3. For the home team, take an average of HF/AA and AF/HA including a variable to put more emphasis on the FOR rather than AGAINST figure – but not on a match to match or week to week basis.
4. Repeat this process for the Away team.
5. Both figures are then adjusted according to the rating given to each home team at home, and each away team away. There is a Home and Away rating for each team, so 40 in all.
6. A slight amendment to the Poisson calculations to more closely match the historic figures for goals over the past 10 seasons (approx. 2.70), and the ratio between home goals and away goals (approx. 4/3), as well as the frequency of home and away goal totals.
7. Run the modified version of Poisson.
8. Check (visually as not programmed at the moment) that the total average goals and the ratio of home/away goals is within tolerance. These will vary from week to week.

What I would like to create is a program that is well suited to a niche aspect of football betting WITHOUT being tied down to steering it in any pre-determined direction. In other words, let the program decide what it does best and then contemplate the best way to take advantage of it!

We can improve this together and I would appreciate any suggestions of what to include, amend, or delete in next season’s format.
Please remember to include reasoning and anticipated effect of any amendments as that would simplify altering the program and monitoring the changes to see whether they had the desired effect.
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Thu Feb 01, 2018 9:30 pm

The results from round 25 saw 24 goals with 5 over 2.5 goals, 5 BTTS Yes, and 4 Wins to nil (2 home 2 away).

Average Goals top.
3 3 2 3 3
Average Goals bottom.
2 2 0 2 4
Over 2.5 Goals.
3 3 2 3 3
Under 2.5 Goals. (The quoted figures are for OVER so deduct from 100)
2 2 0 2 4
Home teams winning to nil.
Yes Lost Drew Yes Drew Drew Won Drew Lost Won
Away teams winning to nil.
Lost Yes Drew Drew Drew Lost Lost Drew Lost Yes
BTTS Yes.
Won Lost Lost Won Won
BTTS No. (The quoted figures are for YES so deduct from 100)
Lost Won Lost Won Won

Another good set of figures, with just the 2 shock results affecting many areas. Again, all the teams that were at home midweek are away this weekend.
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Tue Feb 06, 2018 12:14 pm

The results from round 26 saw 35 goals with 6 over 2.5 goals, 9 BTTS Yes, and 1 Win to nil (1 home 0 away).

Average Goals top.
6 4 2 3 4
Average Goals bottom.
2 5 2 5 2
Over 2.5 Goals.
6 2 4 3 4
Under 2.5 Goals. (The quoted figures are for OVER so deduct from 100)
2 5 2 2 5
Home teams winning to nil.
Yes Won Drew Drew Won Drew Lost Drew Drew Won
Away teams winning to nil.
Drew Lost Lost Won Drew Drew Lost Drew Lost Lost
BTTS Yes.
Won Won Won Won Won
BTTS No. (The quoted figures are for YES so deduct from 100)
Won Lost Lost Lost Lost

I have recently been working on a few patterns. There will be more news on these when more data is available to confirm my findings as this will continue to make 'back fitting' less likely.

The one market that I can't really do much about at the moment is the Correct Scores as I have not stored the weekly analysis for this part of the program. However, it is coming up with figures something closer to realistic odds than ever before at the front end of the market so it is hopeful that may be carried through the whole range. One amendment for next season (it would be a headache to try it mid-term) is to increase the CS grid to separate all scores involving at least one team scoring 5, 6, or 7+ goals. These are currently banded together in 5+ goals, something that was done to make the grid fit into the space available on the original sheet.

Goals for the season now stand at 701 heading for a season total of 1024.54 (2.697 per game) compared to the setting of 2.70 per game (1026 for the season). This should help the analysis to be more accurate overall. The Home/Away ratio is 56.21/43.79 (394/307).

The next set of fixtures also looks ripe pickings for goals - so no doubt there will be three 0-0 draws. :D

I have not amended Chelsea's Home or Away rating yet as I don't want this program to become a weekly prediction sheet. The main purpose is to give solid background data which can be modified in view of current form, injuries, turmoil, strife, illness, suspensions, and likely refereeing decisions.
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Tue Feb 06, 2018 2:52 pm

It would seem that Mr. Poisson isn't over-keen on Arsenal's Win-0 chances at Wembley. :D

Image
LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

davidg3907
Stallion
User avatar
Betting Blog Tips
Posts: 5094
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Tue Feb 13, 2018 12:53 pm

The results from round 27 saw 35 goals with 4 over 2.5 goals, 4 BTTS Yes, and 6 Win to nil (5 home 1 away).

Average Goals top.
1 6 3 4 2
Average Goals bottom.
5 2 1 1 2
Over 2.5 Goals.
1 6 3 4 2
Under 2.5 Goals. (The quoted figures are for OVER so deduct from 100)
5 2 1 1 2
Home teams winning to nil.
Yes Won Yes Drew Won Yes Yes Won Lost Yes
Away teams winning to nil.
Lost Yes Lost Lost Lost Drew Lost Lost Lost Lost
BTTS Yes.
Lost Won Lost Won Lost
BTTS No. (The quoted figures are for YES so deduct from 100)
Lost Won Won Won Lost

LAYING Taking it further
LAYING Delving deeper - NEW

FOOTBALL blogs under Blogs/Football
New RELEGATION diary in Blogs/systems

Who is online

Users browsing this forum: No registered users