Poisson Distribution - This Can Help You

wonderwall
Group 2 Class
Tips
Posts: 107
Joined: Sat Aug 06, 2016 12:01 am

Re: Poisson Distribution - This Can Help You

Postby wonderwall » Sun Oct 29, 2017 11:25 pm

When you're able to, it be good if you post your actual spreadsheet on here as there's definitely some bugs in it. I just saw that Away teams winning to nil.
37.56 Man City 32.12 Chelsea 29.08 West Ham that's pretty odd as West Ham have played 5 matches away (at the point of your post), lost 3, drawn 2, not coming close to a Win. I watch them closely, so a Win to nil seems unlikely. Without delving under the hood of your spreadsheet, no one I can help you further.

davidg3907
At Stud
User avatar
Betting Blog Tips
Posts: 4853
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Mon Oct 30, 2017 3:07 am

Hi Wonderwall,

I can try to post the spreadsheet after the next round of matches.

It is still pretty much WIP, but to quickly respond to your point about West Ham.

The Win to nil would have been heavily influenced by the lack of goals scored by Palace in their 4 previous home games this season, and the current weighting taking last season into account. The ratings (very experimental) had West Ham @ 2.20 favs, so again bias towards them winning. West Ham's "AWAY FOR" rating looks as though it may be too generous.

The other anomaly is when 2 very strong teams play each other, there seems to be insufficient bias to the defensive side of the tallies.

Wins to nil (particularly Homes) are possibly the most consistent figures on the limited data to hand for this season. This is heartening as it is what I thought would be the most accurate aspect before the season started.

Despite a slight rise in the number of goals scored, they are still some way below par. There is also the problem that too many have been scored by a small number of teams, leaving Over 2.5 goals and BTTS to suffer. Believing that this can't continue, I am hesitant to make widespread changes at this point, but am keen to receive any input that helps get round the major issues.

Something I have been intending to look at, but keep forgetting, is to check how closely the bookies Wto0 odds match my Wto0 figures.
LAYING Taking it further
LAYING Delving deeper - NEW

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

davidg3907
At Stud
User avatar
Betting Blog Tips
Posts: 4853
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Thu Nov 02, 2017 9:33 pm

Retaining the same format for ease of reference, the top percentages in the various categories this week are listed below. Matches involving Huddersfield, Brighton, and to a lesser extent Newcastle are based on a reduced amount of available data, and then coupled with estimates.

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

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

davidg3907
At Stud
User avatar
Betting Blog Tips
Posts: 4853
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Thu Nov 02, 2017 10:02 pm

Retaining the same format for ease of reference, the top percentages in the various categories this week are listed below. Matches involving Huddersfield, Brighton, and to a lesser extent Newcastle are based on a reduced amount of available data, and then coupled with estimates.

Average Goals top.
4.45 Man City 3.46 Chelsea 3.34 Tottenham 2.86 Everton 2.77 West Ham
Average Goals bottom.
1.74 Huddersfield 2.01 Swansea 2.11 Southampton 2.32 Stoke 2.56 Newcastle
Over 2.5 Goals.
76.69 Man City 63.73 Tottenham 62.10 Chelsea 53.66 Everton 51.58 West Ham
Under 2.5 Goals. (The quoted figures are for OVER so deduct from 100)
25.53 Huddersfield 32.31 Swansea 34.84 Southampton 40.47 Stoke 45.92 Newcastle
Home teams winning to nil.
44.49 Tottenham 27.87 Southampton 27.67 Newcastle 26.84 Chelsea 25.71 Man City 23.70 Everton 22.67 Swansea 20.68 Stoke 20.36 Huddersfield 15.09 West Ham
Away teams winning to nil.
29.51 West Brom 24.60 Brighton 23.92 Liverpool 22.97 Leicester 18.78 Burnley 15.38 Bournemouth 14.29 Watford 13.55 Man Utd 7.17 Arsenal 4.09 Crystal Palace
BTTS Yes.
64.24 Man City 55.96 Everton 54.35 West Ham 52.71 Chelsea 48.11 Newcastle
BTTS No. (The quoted figures are for YES so deduct from 100)
30.34 Huddersfield 38.08 Swansea 40.40 Southampton 46.03 Stoke 47.59 Tottenham
LAYING Taking it further
LAYING Delving deeper - NEW

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

nors
Legend
User avatar
Betting Blog Tips
Posts: 14789
Joined: Mon Jul 31, 2006 11:00 pm
Location: Coventry

Re: Poisson Distribution - This Can Help You

Postby nors » Tue Nov 07, 2017 10:30 am

More brilliant work David, do you think with all the focus on the EPL that there may be some mileage in poisson for "smaller leagues"

davidg3907
At Stud
User avatar
Betting Blog Tips
Posts: 4853
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Tue Nov 07, 2017 11:43 am

I can see no reason why not in principle.

I am currently updating the records for the first 11 rounds of PL action and making a few minor amendments (plus a fairly radical one) to the spreadsheet.

After that, there would be the small matter of collecting the same information in the same format for the lower leagues, checking the similarities and differences in the data, and modifying the spreadsheet to accommodate 12 matches rather than 10.

However, my priority is to create the best working model on PL that I can. Knowing what the issues are is one thing, finding remedies without adversely affecting anything else is another matter. The way the system is set up, I could back fit all results to the Correct Score, but it would achieve nothing (however good it would look! :D ).

After Wonderwall's comment, I sent him a full copy but as yet have heard nothing back. In the meantime, if anyone else would like a copy to assess, and possibly help with any suggestions or improvements, please let me know.
LAYING Taking it further
LAYING Delving deeper - NEW

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

davidg3907
At Stud
User avatar
Betting Blog Tips
Posts: 4853
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Wed Nov 08, 2017 4:22 pm

Retaining the same format for ease of reference, the top percentages in the various categories this week are listed below. Matches involving Huddersfield, Brighton, and to a lesser extent Newcastle are based on a reduced amount of available data, and then coupled with estimates.

The strange fact about this week's results is that no games finished with either 2 or 3 goals

Average Goals top. 4 1 1 4 5
Average Goals bottom. 1 1 1 4 1
Over 2.5 Goals. Yes No No Yes Yes
Under 2.5 Goals. (The quoted figures are for OVER so deduct from 100) No No No Yes No
Home teams winning to nil. Yes Lost Lost Yes Won Won Lost Drew Yes Lost
Away teams winning to nil. Lost Yes Won Drew Yes Yes Lost Lost Lost Lost
BTTS Yes. Won Won Won Lost Lost
BTTS No. (The quoted figures are for YES so deduct from 100) Won Won Won Lost Won
LAYING Taking it further
LAYING Delving deeper - NEW

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

davidg3907
At Stud
User avatar
Betting Blog Tips
Posts: 4853
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Wed Nov 08, 2017 5:25 pm

Poisson - updated figures.

Up to November 5th, there have been 11 complete rounds of matches, meaning 110 games played.
Of those, 51 have seen 3 or more goals. Splitting the matches into 5 groups of 22 gives the following results (Rank, average goals prediction range, total goals, matches 3+ goals)

1-22 3.34 to 4.45 67 11
23-44 2.82 to 3.27 62 12
45-66 2.47 to 2.77 65 12
67-88 2.11 to 2.46 42 8
89-110 1.66 to 2.10 47 8

Over & Under 2.5 goals follows a pattern that is similar enough not to warrant repetition.

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

1-22 34.92 to 60.91 13 4 3 2
23-44 28.57 to 34.85 7 4 6 5
45-66 23.70 to 28.37 9 3 5 5
67-88 16.29 to 23.09 3 2 8 9
89-110 6.59 to 16.25 2 3 3 14

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

1-22 29.08 to 53.08 9 3 4 6
23-44 19.30 to 23.38 4 4 8 6
45-66 15.38 to 19.18 7 2 7 6
67-88 7.33 to 14.69 3 1 3 15
89-110 2.68 to 7.24 0 1 4 17

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

1-22 52.78 to 64.24 9
23-44 49.19 to 52.71 8
45-66 44.65 to 48.48 8
67-88 39.28 to 44.64 12
89-110 23.63 to 38.86 7

The format was designed around an expectancy of 2.70 goals per game. The actual figure after 110 games is 283 which is 14 short of the 297 benchmark.

The 10 year averages based on 110 matches (excluding this season). This season.

Goals 298.2 283
Over 2.5 goals 56.8 51
Home W-0 28.8 34
Away W-0 16.3 23
BTTS 56.0 44
0-0 Draw 8.9 9

The frequency of matches with just one goal (27) is almost 50% higher than expectancy (18.9).

Although the overall sample size is still small, most sets of figures are beginning to 'feel' closer than in the early stages. In fact, with all except the BTTS figures (which are the furthest from expectation as far as totals are concerned anyway), there is a big difference between the top 60% and bottom 40% of the rankings in each category. If this continues, then it is possible that a clear dividing line in %s will also appear.
LAYING Taking it further
LAYING Delving deeper - NEW

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

davidg3907
At Stud
User avatar
Betting Blog Tips
Posts: 4853
Joined: Thu Jun 25, 2009 11:00 pm
Location: Oadby , Leicestershire

Re: Poisson Distribution - This Can Help You

Postby davidg3907 » Sun Nov 12, 2017 10:30 pm

A few tweaks in some areas and a major overhaul in another haven't made as much difference as expected. The season weightings have been changed to 2 4 8 9, but the FOR/AG ratio remains at 60/40 and the ratings (although some have been updated) keep their 85% value. The ratings possibly need a slightly wider range, but simply amending the weighting has a dreadful effect on the total goals, so all 80 ratings would need adjusting manually in a way that keeps the goals steady - something that may not prove easy.

The major restructuring revolved around amending the distribution, particularly in the event of two free-scoring sides meeting. This does seem to have gone fairly smoothly, having the desired effect on the matches that needed it while having a minimal effect on the rest, but results will be the judge.

Even with the adjusted seasonal weightings, the system is having difficulty keeping up with the improvement in Manchester City this season, so they may stay underrated for another few matches. Whether Leicester's home ratings are too high or not, I think this is less of a problem overall. West Ham's away ratings have been downgraded but may still be a shade high.

Retaining the same format for ease of reference, the top percentages in the various categories this week are listed below. Matches involving Huddersfield, Brighton, and to a lesser extent Newcastle are based on a reduced amount of available data, and then coupled with estimates.

Average Goals top.
3.34 Arsenal 2.95 Leicester 2.81 Manchester United 2.79 Liverpool 2.69 watford
Average Goals bottom.
1.86 Burnley 2.27 Bournemouth 2.35 Crystal Palace 2.39 West Brom 2.43 Brighton
Over 2.5 Goals.
59.17 Arsenal 54.29 Leicester 51.58 Manchester United 51.10 Liverpool 49.55 Watford
Under 2.5 Goals. (The quoted figures are for OVER so deduct from 100)
28.48 Burnley 39.05 Bournemouth 40.94 Crystal Palace 42.53 West Brom 43.13 Brighton
Home teams winning to nil.
56.82 Manchester United 38.65 Liverpool 33.64 Burnley 31.12 Bournemouth 31.08 Brighton 23.30 Arsenal 22.29 Crystal Palace 18.93 Watford 14.68 Leicester 13.42 West Brom
Away teams winning to nil.
32.68 Chelsea 27.32 Manchester City 21.66 Everton 20.68 West Ham 20.05 Tottenham 16.22 Swansea 15.08 Huddersfield13.78 Stoke 9.39 Southampton 4.19 Newcastle
BTTS Yes.
53.01 Watford 49.99 Leicester 46.63 Arsenal 45.80 Brighton 45.42 Crystal Palace
BTTS No. (The quoted figures are for YES so deduct from 100)
31.40 Manchester United 34.19 Burnley 42.85 Bournemouth 43.63 Liverpool 43.91 West Brom

During the week, I shall attempt to upload a copy of the main spreadsheet for perusal and comment.
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