Poisson calculation

Pages: [1]
Author Topic: Poisson calculation  (Read 648 times)
AceJack
Has some experience
**

Karma: -12
Posts: 153



View Profile

AceJack

Poisson calculation
« on: April 10, 2017, 07:15:39 PM »

Hello :)

Recently Pinnacle sport released a pretty interesting article about Poisson calculation for soccer , finding this pretty interesting, -  I have spent the better part of a day building this Poisson calculator up:

Match for Testing is: English Premier: Crystal Palace vs Arsenal (for my model i used stats from 2015/2016 table)

My model produced these Odds: H:5,51 D:4,36 A:1,7   which isnt that far from the actual odds which are  H:4.9 D:4,2 A:1,78

(price for arsenal went up from 1,73 to 1,78 last hour)


But a closer Look at my model I can see that it has it way wrong with 0-0, (14,44) 1-0 (16,41) compered to real odds of 23 and 24 respectively . After reading about this it has come to my attation that the standard  Poisson calc is off and a Zero-inflated model should be used.

 Do anyone here have some links so I can continue reading or maybe some pointers so I can try to update this formula more accurately ?
Logged
maletaja
Probably a Pro
****

Karma: -23
Posts: 431


View Profile

maletaja

Re: Poisson calculation
« Reply #1 on: April 10, 2017, 08:11:36 PM »

Recently i participated some premier league competition.Played many games. And for me it was suprise that market offered 0-0 or under 0.5 goal  at odds 20+
It was suprise to me , i assumed like 8-12
Logged
AceJack
Has some experience
**

Karma: -12
Posts: 153



View Profile

AceJack

Re: Poisson calculation
« Reply #2 on: April 10, 2017, 08:30:14 PM »

Recently i participated some premier league competition.Played many games. And for me it was suprise that market offered 0-0 or under 0.5 goal  at odds 20+
It was suprise to me , i assumed like 8-12

I have been betting against the draw for quite some time and actually 8-12 for 0-0 is close to avrg I would think. 

You will see last season Arsenal on average concedes 1,32 goal(s) in away matches and Crystal palace conceded 1,21 goal(s) on home ground last season, this is probably reason for high odds for 0-0 in this match as well as Arsenal form is pretty ugly atm...
« Last Edit: April 10, 2017, 08:39:55 PM by AceJack » Logged
showMeDaMoney
Has some experience
**

Karma: 2
Posts: 158


View Profile

showMeDaMoney

Re: Poisson calculation
« Reply #3 on: April 19, 2017, 06:49:46 PM »

I've created the following spreadsheet that will take the latest 10 matches.
https://ufile.io/n799u

The concept is taken from Peter Webb youtube video
The matches are latest 10 for home side and latest 10 for away team.
I should probably include the latest 5 h2h for home/away but then I've found this mostly irrelevant in football for deciding results. I might chuck it in to see if it makes any difference to outcome

I don't want to go too far back into previous seasons because;
- Squads change, coaches, systems, feelings, money, opposition, league, position

Does your poisson reflect differently when using  latest matches.
Feel free to take a look at the spreadsheet linked, I'm just getting started out, but I'm sure I've messed it up somehow.
« Last Edit: April 19, 2017, 07:08:36 PM by showMeDaMoney » Logged
showMeDaMoney
Has some experience
**

Karma: 2
Posts: 158


View Profile

showMeDaMoney

Re: Poisson calculation
« Reply #4 on: April 20, 2017, 10:30:37 PM »

I tweaked the spreadsheet algorithm and found the odds are being reflected based on recent history.

So far I managed to pick 10/10 O/U based on 10 selected matches and for a stake of $100.
I would have made $462
But ofcourse I'm sure, once I put real money, it won't work! and I would have lost $400+

Next test is to put some small money and see what happens.
If using Poisson, it could be a good method place this algorithm against O/U value bets.
Logged
Pages: [1]
Print