Forums

General Betting

There is currently 1 person viewing this thread.
These 832 comments are related to the topic:
Correct Score Torture

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
Page 7 of 21  •  Previous | 1 | ... | 5 | 6 | 7 | 8 | 9 | ... | 21 | Next
sort by:
Show
per page
Replies: 832
By:
FatalStatistics
When: 14 Jun 09 22:12
Racing Club v Boca Juniors 3-0***WON***
3-0 @ 40 ( 34.24 )


.. lovely !
By:
crescit
When: 14 Jun 09 23:53
nice one Fatal!
By:
FatalStatistics
When: 14 Jun 09 23:54
Aalesunds v Viking 0-1***WON***
0-1 @ 12.50 ( 11.52 )
Lahti v IFK Mariehamn 1-1***WON***
1-1 @ 8.40 ( 7.73 )
Valerenga v Stabaek 0-1***WON***
0-1 @ 15 ( 12.98 )


50 from 596 ( @ 13.95 ) ( 17.03% ROI )
By:
FatalStatistics
When: 14 Jun 09 23:54
thanks cres, lookin good..
By:
gleeman
When: 15 Jun 09 08:33
Nice work, Fatal.

You need great patience for something like this, but it seems to be working.

Is this just a paper exercise, or are you making real money?
By:
buzzer
When: 15 Jun 09 09:25
well done fatal, that 90 th minute equaliser in the Lhati game made me some cash too ;)
By:
buzzer
When: 15 Jun 09 09:25
Lahti sorry
By:
ebasson61
When: 15 Jun 09 09:42
Hi FS

I just wondered if you have analysed the poisson success rate for each individual score? The reason I ask is that Dixon and Coles (1997) presented evidence that the Poisson model has a tendency to under-forecast 0-0 and 1-1 draws and over-forecast 1-0 and 0-1 score lines. Have you experienced this yourself

Their method for dealing with this was to increase the model predictive probability of 0-0 and 1-1 and reduce the model predictive probability of 1-0 and 0-1. This was done by defining a new parameter in the model (phi), such that the model probabilities for the four score outcomes in question were multiplied by the following factors:

0-0: Multiply this probability by 1 + (Phihome x Phi away x Phi)
1-1: Multiply this probability by 1 + Phi
1-0: Multiply this probability by 1 - (Phi away x Phi)
0-1: Multiply this probability by 1 - (Phi home x Phi)

Calculation of Phi seems to vary between 0.05 and 0.1.

Anyway, continued good luck to you.
By:
subversion
When: 15 Jun 09 09:58
i use a similar method to highlight 'out-of-sync' prices

i've never tried betting purely based on this method though, i only use this information alongside other data i have, so this thread is a very interesting read

i construct a full probability tree rather than using any closed-form solution for the distribution... makes it easier for me to tweak the the individual tree nodes for any kind of major events such as a sending off at some point in the match, as well as tweaks to the implied goal rate based on the current score (which fixes the draw prediction problem)

anyway, this thread has definitely given me food for thought
By:
FatalStatistics
When: 15 Jun 09 11:39
50 from 597 ( @ 13.95 ) ( 16.83% ROI )

Nice to see i've generated some 'serious' interest !.

Gleeman - slow and steady wins the race, just betting small stakes whilst in this 'testing' period.
- promised myself not to crank up the stakes until conclusively profitable beyond 1000 bets ! - (Discipline, discipline !)
Thanks Buzz -
ebasson61 - i have not kept stats on each individual scores success rate ( yet ), never seen the Dixon & Coles evidence myself ( yet ) But i can tell you i have made vaquely similar modifications myself by -
*increasing the 0-0 1-1 2-2 3-3 draw result by varying percentages and
*decreasing all other outcomes by the same total percentage ( including 1-0 & 0-1 )
But I must say, poisson, with tweaked draws probably will not be really profitable on its own..
The inevitable losing streak will kill your bank. The incorporation of a 'likelihood factor' is what makes this method work more 'consistently' for me. This 'factor' i'll keep to my self for now.
Subversion - sounds like you bet in-play with a similar method ?? I do not because of too many quick change variables etc.. but very good luck to you and i'm very interested in any further input from you or anyone else experienced in these areas..
Good luck All.
By:
FatalStatistics
When: 15 Jun 09 12:28
Brazil v Egypt
0-1 @ 55 ( 37.71 )
England U21 v Finland U21
1-0 @ 7.80 ( 7.38 )
Start v Molde
0-1 @ 15.50 ( 13.18 )
USA v Italy
0-3 @ 16.50 ( 14.77 )
Spain U21 v Germany U21
0-0 @ 11.50 ( 11.00 )
&
3-0 @ 34 ( 29.58 )
By:
subversion
When: 15 Jun 09 14:12
Fatal - yes i make most of my money in-running, but not from this method

the problem is - building the tree/distribution means assuming a 'goals per game' average figure for each team

and i seem to be at my best when assuming these 'goals per game' figures are incorrect, rather than just assuming they are correct and looking for individual scores which don't conform

if that makes sense...

but i do find that being able to view where the tree is 'out of sync' with itself helps me to extract that little extra bit of value
By:
Cosmic Horizon
When: 15 Jun 09 14:53
As a matter of possible interest I give my correct score odds in double brackets (worked out via complex formulas derived from match odds and under/over 2.5 goals odds). These figures are after taking 5% commission into account and are to 1 significant decimal place.

England U21 v Finland U21
1-0 @ 7.80 ( 7.38 ) ((7.3))
Start v Molde
0-1 @ 15.50 ( 13.18 ) ((17.0))
Spain U21 v Germany U21
0-0 @ 11.50 ( 11.00 ) ((11.1))
&
3-0 @ 34 ( 29.58 ) ((28.1))
By:
Cosmic Horizon
When: 15 Jun 09 14:54
They are for backing which obviously will be higher than the laying figures due to commission.
By:
subversion
When: 15 Jun 09 15:12
well just for a laugh here are mine for a comparison


England U21 v Finland U21
1-0 @ 7.80 ( 7.38 ) SUB - 7.9 NO BET
Start v Molde
0-1 @ 15.50 ( 13.18 ) SUB - 14.3 NO BET
Spain U21 v Germany U21
0-0 @ 11.50 ( 11.00 ) SUB - 10.2 BET
&
3-0 @ 34 ( 29.58 ) SUB - 27 BET


the real outliers i'm finding are

England U21 v Finland U21
lays of 0-3, 2-3, 3-3 at odds of over 200-1... yikes

Start v Molde
Back 0-3 (BF 55, Me 45)
Lay 1-1 (BF 7.4, Me 8.3)
Back 3-0 (BF 34, Me 29)
Back 3-3 (BF 75, Me 65)

Spain U21 v Germany U21 (2 backs, same as FinalStats ones)
Lay 2-2 (BF 18, Me 21)
Lay 2-3 (BF 48, Me 60)
Lay 3-2 (BF 40, Me 47)
Lay 3-3 (BF 90, Me 123)


funny how our Spain v Germany ones match closely, but the others not so much
By:
Cosmic Horizon
When: 15 Jun 09 15:47
Mine in brackets

Start v Molde
Back 0-3 (BF 55, Me 45) (would back at 56.8 would Lay at 51.5))
Lay 1-1 (BF 7.4, Me 8.3) (B8.8 L8.0)
Back 3-0 (BF 34, Me 29) (B28.0 L25.4)
Back 3-3 (BF 75, Me 65) (B61.2 L55.4)

Spain U21 v Germany U21 (2 backs, same as FinalStats ones)
Lay 2-2 (BF 18, Me 21) (B19.3 L17.5)
Lay 2-3 (BF 48, Me 60) (B62.6 L56.7)
Lay 3-2 (BF 40, Me 47) (B45.3 L41.1)
Lay 3-3 (BF 90, Me 123) (B119.8 L108.4)
By:
subversion
When: 15 Jun 09 15:59
looks like we're agreeing more than we're disagreeing on those cosmic

2-2 is starting to look an interesting back in the England game at odds close to 50
By:
subversion
When: 15 Jun 09 16:00
i've now got this in the England v Finland

Back 2-2 (BF 50 Me 41)
By:
subversion
When: 15 Jun 09 16:01
and now

Lay 1-0 (BF 7.4 Me 8.5)
By:
Cosmic Horizon
When: 15 Jun 09 16:03
2-2 in Eng match I have 38.4 to back!
By:
Cosmic Horizon
When: 15 Jun 09 16:03
I have lay at 6.9 for eng to win 1-0
By:
subversion
When: 15 Jun 09 16:11
you using any adjustments for draw probabilities etc?
By:
Cosmic Horizon
When: 15 Jun 09 16:14
No.
By:
FatalStatistics
When: 15 Jun 09 22:51
Subversion & Cosmic, very interesting and well done .
close with the england one and we all agreed on this winner...

Spain U21 v Germany U21 0-0***WON***
0-0 @ 11.50 ( 11.00 )


51 from 603 ( @ 13.90 ) ( 17.57% ROI )
By:
FatalStatistics
When: 15 Jun 09 23:03
And Sub, your statement..

the problem is - building the tree/distribution means assuming a 'goals per game' average figure for each team

is the absolute heart of my method and any method that uses the poisson formula.
By:
subversion
When: 16 Jun 09 06:52
yep, sorry my wording wasnt clear - that isn't the 'problem'... the 'problem' is that my edge and best profits come from using the probability model to figure out what the market is using for those goals per game figures, and then assuming that those goals per game figures are wrong, and working from there

those goals per game figures are assumptions that you have to assume are correct for your method to work... and it isn't something i've really looked into much personally, so i honestly have no idea if i'd be able to find an edge

but this is why i find your thread interesting
By:
subversion
When: 16 Jun 09 06:58
so we're doing the same thing up to the point where we generate those goal per game figures

at that point, we take opposing assumptions, and work from there

but that isn't to say the approaches cant both work, since we're both using the figures to look for good value

and certainly, if i come to the conclusion that the goal per game figures are actually pretty accurate after watching the match for a bit, then our approach becomes pretty similar
By:
FatalStatistics
When: 16 Jun 09 11:57
so you'd agree sub , that the key is finding the cs's that are substantially 'out of sync' with the rest of the same market .. and letting probability do the rest.
We're just doing it slightly differently.

Sweden U21 v Belarus U21
3-0 @ 17.50 ( 16.29 )
Italy U21 v Serbia U21
3-0 @ 30 ( 26.09 )

GL.
By:
crescit
When: 16 Jun 09 12:18
I'm getting confused chaps. You both seem to use your stats, goal expectancy etc to price the cs runners in a market and then look for the value selections. Hopefully right so far. Sub you then use this for your picks but Fatal you then also look at distribution and factor in a likelihood on this result is due? otherwise, I'm a little confused on where poission comes in?

And Fatal, great results so far esp given the size of your sample. I was wondering what sort of ROI you might achieve if at the other end of your scale you layed the scores you believed to be poor value

GL
By:
subversion
When: 16 Jun 09 13:57
cresc - the difference seems to be that Fatal is trying a strategy based entirely on this single approach

whereas i use this approach alongside several other indicators to make my decisions

i've never really focused on placing large numbers of bets using this single approach all by itself, which is why i'm finding this thread interesting, and why i offered some of my numbers for comparison

ultimately, we're both looking for value... it seems that Fatal is focusing more on individual scores being out-of-sync, whereas i tend to focus first on whether or not i believe the entire market is out-of-sync, before looking at individual scores
By:
subversion
When: 16 Jun 09 14:10
here are some of my outliers for tonight btw (excluding lays of 50+)


Sweden v Belarus
Back 0-0 (BF 12 Me 10)
Back 3-0 (BF 17.5 Me 15)
Lay 1-1 (BF 7.6 Me 8.5)
Lay 2-1 (BF 9.8 Me 11)
Lay 3-2 (BF 40 Me 50)
Lay Over 4.5 (BF 8.8 Me 10.3)

Italy v Serbia
Back 3-0 (BF 30 Me 25)
Lay 2-2 (BF 19.5 Me 22)
By:
subversion
When: 16 Jun 09 14:11
Fatal - looks like we agree on those 3-0s
By:
crescit
When: 16 Jun 09 15:32
ok cheers sub, GL both of you..
By:
blackjackking
When: 16 Jun 09 15:48
Fatal statistics Ive read and watched your efforts so far and i have been very impressed!

If only i knew how you achieved your expected odds (using your poisson system - no idea hwhat this is or how it works) i think i would try doing this myself. Could you point me in the right direction to doing this myself rather than watching every post you have made.

Im just very curious and wanna work things out and see how things go before placing bets myself. watching you and the workings has been good fun , just wanna understand a little more really :)

cheers
By:
Cosmic Horizon
When: 16 Jun 09 16:27
Game already started but here is my prior to match odds

Sweden v Belarus
Back 0-0 (BF 12 Me 10) (back at 9.8 Lay at 9.0))
Back 3-0 (BF 17.5 Me 15) (b19.3 L17.6)
Lay 1-1 (BF 7.6 Me 8.5) (b8.0 L7.3)
Lay 2-1 (BF 9.8 Me 11) (b11.0 L10.0)
Lay 3-2 (BF 40 Me 50) (b 51.7 L46.9)
By:
FatalStatistics
When: 17 Jun 09 03:44
51 from 605 ( @ 13.90 )( 17.19% ROI )

\ Cres, we're using the poisson distribution to give each correct score's %chance (&therefore odds), if you do a poisson 'forum search' you might get more info and a clearer picture.
And yes i thought about the lay side of things , but monitoring everything becomes very time consuming and tiresome, so i'll continue as is for now but will def check out the lay end at some stage. I expect it would also be profitable, but with much 'more' liability !

\ blackjack, thanks, but i cant do much for you other than to suggest bucketloads of research on poisson and excel spreadsheets etc and maybe do some forum searches on the topics also, as i found them helpful initially. If you have more specific questions, i'm happy to help if i can though !.

http://fred77.co.uk/betfair-forum-archive.php

\ sub, hope you didnt lay the over 4.5 SwedenV Belarus !

\ Cosmic, are you actually backing/laying what your posting??
By:
subversion
When: 17 Jun 09 04:40
Fatal - lol nope, i don't trade based on these numbers alone, and certainly not pre-match... i'm an in-running player, just posting my pre-match model outputs for comparison

but that was the only lay so far to come in, so wouldnt have done much harm

btw i configure my probability tree based on the match odds and over/under 2.5s... so that lay showed that the 2.5s and 4.5s markets were totally out of sync with each other... i dont think i've seen a divergence that big before based purely on the goal-per-game probabilities, but clearly there was a good reason for this

it was a very weird game, barely any chances for 30 minutes, then the floodgates opened, with the sweden team very strong on paper but not having played much together... so no wonder the markets were so strange...

it looks likely there was uncertainty about whether the sweden team would 'gel', but when they did - carnage
By:
FatalStatistics
When: 17 Jun 09 09:50
South Korea v Iran
0-0 @ 8.80 ( 8.27 )
Australia v Japan
AnyUnq @ 24 ( 20.57 )
By:
FatalStatistics
When: 17 Jun 09 12:16
51 from 607 ( @ 13.90 )( 16.80% ROI )

Spain v Iraq
1-0 @ 14.50 ( 11.46 )
IFK Mariehamn v FC Haka
2-0 @ 15.50 ( 14.54 )
Tampere United v RoPS
1-2 @ 38 ( 29.85 )
VPS v KuPS
0-1 @ 16.50 ( 14.76 )
Bryne v Viking
1-1 @ 9 ( 8.49 )
Bahrain v Uzbekistan
3-0 @ 30 ( 25.55 )
JJK v Lahti
1-0 @ 14.50 ( 13.42 )
Saudi Arabia v North Korea
1-0 @ 6.80 ( 6.42 )
South Africa v New Zealand
1-0 @ 8.60 ( 7.85 )

GL All.
By:
FatalStatistics
When: 17 Jun 09 12:27
2 in the Tampere United v RoPS game...

Tampere United v RoPS
((1-2 @ 38 ( 29.85 ) ))
&
1-0 @ 8.60 ( 8.08 )
Page 7 of 21  •  Previous | 1 | ... | 5 | 6 | 7 | 8 | 9 | ... | 21 | Next
sort by:
Show
per page

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
‹ back to topics
www.betfair.com