Free £25 Bet!
Free £25 Bet!
| Free £25 Bet! | | Free £25 Bet! |
In association with Sports-Punter Free Bets Odds Comparison BetHelp Limso
We are the Official Forum of FreeBetting.net & FCBet.com
Sports News  | Sports Stats | Live Scores | OddsChecker | Place Bets | Suggest a Site |
| |||||||
| Systems & Strategy Forum Discuss all your strategies, systems, selection methods and staking plans here. Try and keep your match selections to the other forums. |
| |  |
 |
| | Thread Tools | Display Modes |
27-10-2006, 22:23
| #1 (permalink) | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| aka plopplop   Join Date: 08 Aug 2006 Location: Nottingham Age: 29
Posts: 618
|  Predicting The English Premier League-using stats-Strike Rate 57.2% - Yield 16.9% Hello all, I have kind of been lurking around for a few months after finding this site through my research for my own model development. I've got to a position now where I feel happy to share where I'm at and draw on the massive experience that clearly resides on this forum. Where to start...........I'm gonna lay it out piece by piece as I genuinely want some feedback and help on how to improve it. I started with the prem but I have the ability to do something very similar for 21 different leagues around europe.........sadly - i have a day job. I've got stats etc for 2000-01 --> 2005-06 & now into 06/07 and they are the following:
First off, have I missed anything on the form side????? There are a few things I'd like to include - stats relating to resolve - performance vs top/bottom half - stuff you can find on football365. I've included the bookmakers odds as inputs to the model as I feel that by doing this I incorporate their vastly superior systems which derive the %age chance of an outcome for a given football match - these must be good predictors of an outcome.......and they are - its knowing when not to bet thats key......i.e when Norwich will beat Man U. So i've got this dataset for 6 seasons......... I use a stats package to run a predictive model for each of the three outcomes, HW/D/AW. I build my model on 4 seasons of data - 00/01 - 03/04 I validate my model on 1 season of data - 04/05 The stats package balances the weights applied to each of the variables by using the validation data to ensure that the model not only explains as much variation in what makes up a HW/D/AW but remains robust when deployed across a unseen set of data. I test my model on 1 season of data - 05/06 This is in effect an unknown sample of data that the model has not seen before and therefore is a true test of the models ability to predict the matches. Selection Criteria: When scoring upcoming matches or backtesting I am presented with a score for each of HW / D / AW which can be transformed into a percentage via a simple formula. The package offers optimal cutoffs for each of the three scores for a given match - when only one outcome is highlighted - i place a bet. The cutoffs are determined by assigning costs to certain outcomes. If I place a bet and I am correct I gain on average 1.8 points, whereas if I place a bet and am wrong - I lose 1 point. Not placing a bet regardless of the outcome costs/makes me 0pts. The package calls this a confusion matrix. The same is done for Draw and AW. Staking Plan: Currently I've got everything set at level stakes, 1pt per bet. Be it home, draw, away and differing odds, I stake only 1pt. I'm a bit of a novice at staking plans, i've read a fair bit on here about it and a lot of people seem to be suggesting that you can't go far wrong with level stakes - this is a section I definately need help on......!!!!!! What does it all mean: As I've mentioned - this model was scored across the 2005-2006 english premiership season using a level staking plan of 1pt for bet and using the optimised cutoffs suggested by the package. Here are a few bits of info about how it is getting on....
So as you can see, I only place a bet on 193 of the 380 matches in the EPL last season. 3 matches didn't manage to get over any of the hurdles. Here is how the bets break down:
The model tends to find HW bets in two-thirds of cases, of which it gets 71% (90/127) correct. The D and AW strike rate isn't as good but the odds are far greater, on average, so that mitigates this. Staking at 1pt per bet returned the following..... Overall Strike Rate: 65% (H 71%, D 38%, A 57%) Staked: 193pts (H 127, D 8, A 58) Returned: 211.84pts (H 10.67, D 1.4, A 6.77) Yield: 9.76% NB: The above is calculated using William Hill odds only - using the best available across the bookmakers mentioned in the big list above would boost the yield to ~15%. Here is how the 18.84 pts were accumulated across the season:  Thats it - my cards are on the table. I would love to hear from any of you that do something similar or have any suggestions / experiences to share / data to share/exchange. Cheers Matt I originally posted this in Punter's Tools & General Betting Help but am now in a position to post my selections etc for upcoming matches. Original thread: http://www.punterslounge.com/forum/s...d.php?p=593410 2006/2007 performance to date is: 0.84 pts profit using william hill odds 2.45 pts profit using the best odds across gamebookers, interwetten, ladbrokes, sportingbet, william hill. Last edited by Matthew; 14-05-2007 at 22:09. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|  |
|
27-10-2006, 22:41
| #2 (permalink) |
| aka plopplop Join Date: 08 Aug 2006 Location: Nottingham Age: 29
Posts: 618
| Re: Predicting The EPL - via statistical modelling - finished but seeking help... Selections for this weekend: 28/10/2006 Arsenal Everton - HW - 1.44 - premierbet/skybet 28/10/2006 Bolton Man Utd - AW - 1.91 - totalbet/betfred/betdirect/ukbetting 28/10/2006 Fulham Wigan - No Draw - 1.23 - ladbrokes 28/10/2006 Liverpool Aston Villa - HW - 1.62 - totesport/bluesq/betdirect/stanjames/ladbrokes 28/10/2006 Newcastle Charlton - HW - 1.83 - PremierBet 28/10/2006 Portsmouth Reading - No Away Win - 1.12 - ladbrokes 28/10/2006 Sheffield Utd Chelsea - AW - 1.33 - Betfred 28/10/2006 Watford Tottenham - No Draw - 1.23 - ladbrokes I'm not entirely sure about the value available in these double chance Ladbrokes bets. |
|
| |
|
28-10-2006, 13:03
| #3 (permalink) |
| Shrewdie Punter Join Date: 24 Jul 2005
Posts: 674
| Re: Predicting The EPL - via statistical modelling Looks very impressive to me. I'm not the best person to give advice (there are many who are better qualified than me on this forum) but you are right about the poor value on the Ladbrokes prices. You would defininately do better to lay the outcome at Betfair. 28/10/2006 Fulham Wigan - No Draw - 1.23 - ladbrokes Lay the draw @ 3.35 with Betfair = odds of approx 1.43 before commission. 28/10/2006 Portsmouth Reading - No Away Win - 1.12 - ladbrokes Lay Reading @ 4.5 = approx odds 1.29 before commission. 28/10/2006 Watford Tottenham - No Draw - 1.23 - ladbrokes Lay the draw @ 3.4 with Betfair = odds of approx 1.42 before commission. Good luck. |
|
| |
|
29-10-2006, 13:27
| #4 (permalink) |
| aka plopplop Join Date: 08 Aug 2006 Location: Nottingham Age: 29
Posts: 618
| Re: Predicting The EPL - via statistical modelling - finished but seeking help to imp Thanks for pointing out the LAY @ betfair option. I'll probably do that from now on. Am I right in assuming that to calculate the back odds from the lay you do the following? lay odds of 4.5 = 1/4.5 = 22% chance. The flipside is 1/??? = 1-22% chance as 1/1.29 = approx. 78% chance There is certainly more value in that than the 1.12 I got from ladbrokes. Last edited by Matthew; 29-10-2006 at 15:16. |
|
| |
|
29-10-2006, 15:16
| #5 (permalink) |
| aka plopplop Join Date: 08 Aug 2006 Location: Nottingham Age: 29
Posts: 618
| Re: Predicting The EPL - via statistical modelling - finished but seeking help to imp 29/10/2006 West Ham v Blackburn (Lay Draw) - 3.45 All selections are currently being based on 1pt level staking plan. I'd like to hear from anyone who can suggest an alternative staking plan. |
|
| |
|
29-10-2006, 16:02
| #6 (permalink) |
| Senior Punter Join Date: 03 Feb 2006 Age: 37
Posts: 120
| Re: Predicting The EPL I typed out a few issues/points for discussion, and it is really, really frustrating indeed not being able to paste on this message board. Anyone got any suggestions about that ? plopplop - good luck. |
|
| |
|
29-10-2006, 16:32
| #7 (permalink) |
| aka plopplop Join Date: 08 Aug 2006 Location: Nottingham Age: 29
Posts: 618
| Re: Predicting The EPL - via statistical modelling - finished but seeking help to imp I am able to paste on this message board. When the pointer is inside the reply box: Right-click - paste is below Select All |
|
| |
|
29-10-2006, 17:37
|
