Predicting The English Premier League-using stats

[Rushian]

Re: Predicting The English Premier League-using stats plopplop, I have been following your thread this year as well, and I wouldn't be disheartened with the bad start... Hopefully, things will change soon :hope I have a question if I may about something which I'm not too sure that I understand correctly: how come your model assigns probabilities which add to more than 1 (unless of course this is just due to rounding although I have a feeling this may not be it)? If it uses all those variables to model the probability of H, D or A simultaneously, then the constraint of unity should also be included somehow, I guess. Unless you are modelling odds (rather than probabilities) in which case the constraint would be more complex (and probably more difficult to apply). Apologies if this is something that has already been explained. Once again, good luck for the future.

[DC]

Re: Predicting The English Premier League-using stats Good luck with the system this year mate. As with any start to a new season it'll take time for trends to form and results to "settle down". Just keep going with it. Danny :hope

[Matthew]

Re: Predicting The English Premier League-using stats

plopplop, I have been following your thread this year as well, and I wouldn't be disheartened with the bad start... Hopefully, things will change soon :hope I have a question if I may about something which I'm not too sure that I understand correctly: how come your model assigns probabilities which add to more than 1 (unless of course this is just due to rounding although I have a feeling this may not be it)? If it uses all those variables to model the probability of H, D or A simultaneously, then the constraint of unity should also be included somehow, I guess. Unless you are modelling odds (rather than probabilities) in which case the constraint would be more complex (and probably more difficult to apply). Apologies if this is something that has already been explained. Once again, good luck for the future.

Another good question - you guys certainly are paying attention. I suppose "it matters more when there is money on it!" I have three separate models in action - one each predicting HW / D / AW, the %age chance that is output generally sums to around 100% but I reconcile these percentages with that of the overround I experience form the 5 bookies in my dataset. So if I have say: HW - 50% D - 25% AW - 25% With bookies odds of: HW - 2.00 (50%) D - 3.20 (31.25%) AW - 3.75 (26.67%) The overround is almost 8%. To marry up my %age's with the bookies I multiply them out to sum 108%. So I do 50% * (108%/100%). This may or may not be exactly right but it doesn't seem to have done me any harm as the odds I generally get are a cut above thanks to Betfair.

[Matthew]

Re: Predicting The English Premier League-using stats

Good luck with the system this year mate. As with any start to a new season it'll take time for trends to form and results to "settle down". Just keep going with it. Danny :hope

Cheers Danny - just like the Boro - I was happy to get my first win at the weekend. It actually took my bird to point out that your avatar is infact a bra!!!! SUPERB!!!!!

[Rushian]

Re: Predicting The English Premier League-using stats

Another good question - you guys certainly are paying attention. I suppose "it matters more when there is money on it!" I have three separate models in action - one each predicting HW / D / AW, the %age chance that is output generally sums to around 100% but I reconcile these percentages with that of the overround I experience form the 5 bookies in my dataset. So if I have say: HW - 50% D - 25% AW - 25% With bookies odds of: HW - 2.00 (50%) D - 3.20 (31.25%) AW - 3.75 (26.67%) The overround is almost 8%. To marry up my %age's with the bookies I multiply them out to sum 108%. So I do 50% * (108%/100%). This may or may not be exactly right but it doesn't seem to have done me any harm as the odds I generally get are a cut above thanks to Betfair.

However by changing your estimated probabilities to sum to e.g. 108% and transforming these back to odds, would it not mean that more bets would (wrongly) be highlighted as "value"? I would think that comparing your probability with the actual bookies odds and betting when that comparison suggests a "value bet" would be better. Just a thought.

[Matthew]

Re: Predicting The English Premier League-using stats

Date Home Away Prediction Odds Bookie Stake Correct Return 25/08/2007 Arsenal Man City D 3.80 Betfair £32.53 . . 25/08/2007 Aston Villa Fulham AW 5.30 Betfair £32.53 . . 25/08/2007 Bolton Reading D 3.45 Betfair £32.53 . . 25/08/2007 Chelsea Portsmouth HW 1.34 Betfair £32.53 . . 25/08/2007 Derby Birmingham HW 2.60 Betfair £32.53 . . 25/08/2007 Everton Blackburn D 3.35 Betfair £32.53 . . 25/08/2007 Sunderland Liverpool D 3.90 Betfair £32.53 . . 25/08/2007 West Ham Wigan D 3.45 Betfair £32.53 . .

[SecondSight]

Re: Predicting The English Premier League-using stats PlopPlop Interestly alot of draws there mate. Typically a game has approx 26.5 - 27% chance of drawing, except in big homers and derby games. Good luck

[Matthew]

Re: Predicting The English Premier League-using stats

PlopPlop Interestly alot of draws there mate. Typically a game has approx 26.5 - 27% chance of drawing, except in big homers and derby games. Good luck

I'd dispute this. Very rarely will a game have this exact chance of a draw. This is merely the proportion of games that end in a draw at the end of a season in the EPL. Each and every match has to be assessed on its merits - similarly does every day of matches. I just think its a day of draws on sat. Well I don't - thats what my laptop thinks. Fingers crossed!!!!

[Matthew]

Re: Predicting The English Premier League-using stats

However by changing your estimated probabilities to sum to e.g. 108% and transforming these back to odds' date=' would it not mean that more bets would (wrongly) be highlighted as "value"? I would think that comparing your probability with the actual bookies odds and betting when that comparison suggests a "value bet" would be better. Just a thought.[/quote'] You may be right - I did it this way when I started it out and however right or wrong it might be - my assessment of how the current will work includes this "error". Being a path of least resistance kind of guy I'm happy to leave it how it is. If I get some time - I might see how different the net result would be if I don't reconcile to the bookies overround and just leave my %ages at a total of 100.

[Matthew]

Re: Predicting The English Premier League-using stats A slight loss on the day.............

Date Home Away Prediction Odds Bookie Stake Correct Return 25/08/2007 Arsenal Man City D 3.8 Betfair £32.53 N 25/08/2007 Aston Villa Fulham AW 5.3 Betfair £32.53 N 25/08/2007 Bolton Reading D 3.45 Betfair £32.53 N 25/08/2007 Chelsea Portsmouth HW 1.34 Betfair £32.53 Y £43.03 25/08/2007 Derby Birmingham HW 2.6 Betfair £32.53 N 25/08/2007 Everton Blackburn D 3.35 Betfair £32.53 Y £105.15 25/08/2007 Sunderland Liverpool D 3.9 Betfair £32.53 N 25/08/2007 West Ham Wigan D 3.45 Betfair £32.53 Y £108.24

System Stats:

ALL HW D AW No of Bets: 22 2 14 6 Staked: £779.61 £65.06 £491.87 £222.68 Returned: £439.04 £43.03 £396.01 £0 P+L: £-340.57 £-22.03 £-95.86 £-222.68 Strike Rate: 22.7% 50% 28.6% 0% Yield: -43.7% -33.9% -19.5% -100% Avg Back Odds: 3.51 1.97 3.61 3.79 Avg Win Odds: 2.95 1.34 3.35 0 Each Person: £35.19

[Matthew]

Re: Predicting The English Premier League-using stats Take this on whatever merits you feel............but rather annoyingly, due to my camping trip to Rutland, we missed the 2 matches on Sunday that I would have predicted correctly!!!!!! Although I would argue that Man U beating Tottenham and a draw in the NE derby were far from surprising.

[Hodgey]

Re: Predicting The English Premier League-using stats

Take this on whatever merits you feel............but rather annoyingly, due to my camping trip to Rutland, we missed the 2 matches on Sunday that I would have predicted correctly!!!!!! Although I would argue that Man U beating Tottenham and a draw in the NE derby were far from surprising.

Couldn't you have placed the Sunday bets at the same time as the Saturday ones?

[barrhead73]

Re: Predicting The English Premier League-using stats Unlucky Plop. Hope the camping trip was worth it!!!!! I was online on Sunday looking for yours picks and i thought you would probably come up with a HW for Man U and a draw in the other game. But i never backed it either. Good luck for this weekend.

[Matthew]

Re: Predicting The English Premier League-using stats

Date Home Away Prediction Odds Bookie Stake Correct Return 29/08/2007 Bolton Everton D 3.40 Betfair £32.38 29/08/2007 Fulham Tottenham D 3.50 Betfair £32.38 29/08/2007 Newcastle Wigan HW 1.73 StanJames £32.38 29/08/2007 Liverpool Derby HW 1.22 StanJames £32.38 29/08/2007 Man Utd Sunderland HW 1.22 BlueSq £32.38

[Matthew]

Re: Predicting The English Premier League-using stats

Couldn't you have placed the Sunday bets at the same time as the Saturday ones?

Not really. I need all the stats Saturday's games to feed into some of the variables that I use for Sunday's games. The model was built on this basis and as I do not understand the effects of predicting Sunday's games prior to Saturday - I'll leave well alone.

Unlucky Plop. Hope the camping trip was worth it!!!!! I was online on Sunday looking for yours picks and i thought you would probably come up with a HW for Man U and a draw in the other game. But i never backed it either. Good luck for this weekend.

Cheers barrhead73 - I had a great time camping - in future i'll try and remember to say whether I'll make the Sunday games. I'll pretty much always be able to make Saturday and any midweek games.

[happygooner]

Re: Predicting The English Premier League-using stats Hey Matt I just wondered if you had opened an account (or even taken a look at the odds) at Mansion? Did it recently myself and have to say that while not many options for other division the prem is there and odds quite favourable. For example I got both Sat draws at 3.37 & 3.47 respectively. Slightly lower odds than offered at betfair yes but with the benefit of having no commission it actually comes out better. After all those 3.4 and 3.5 odds come out around 3.23 & 3.32 Tell me to can it if you like but i'm only trying to help haha! GL:ok

[Nade]

Re: Predicting The English Premier League-using stats Your predictions look like they have a good chance of coming off this week, unlike the last couple of weeks there are no funny results in there, but i guess that's the benefit of adding more info to the system with it evolving every week, anyway good luck:ok

[Matthew]

Re: Predicting The English Premier League-using stats

Hey Matt I just wondered if you had opened an account (or even taken a look at the odds) at Mansion? Did it recently myself and have to say that while not many options for other division the prem is there and odds quite favourable. For example I got both Sat draws at 3.37 & 3.47 respectively. Slightly lower odds than offered at betfair yes but with the benefit of having no commission it actually comes out better. After all those 3.4 and 3.5 odds come out around 3.23 & 3.32 Tell me to can it if you like but i'm only trying to help haha! GL:ok

Good spot. I'll look at them in the future as a final check. I use Oddschecker and bestbetting for my odds comparison needs. Do you have a better alternative - we all know how important it is to get the best possible price!!!! Cheers for the GL msgs Happygooner and Nade. I've got everything crossed - we need at least one of the draws to come in - there isn't enough cover in the other matches.

[happygooner]

Re: Predicting The English Premier League-using stats

Good spot. I'll look at them in the future as a final check. I use Oddschecker and bestbetting for my odds comparison needs. Do you have a better alternative - we all know how important it is to get the best possible price!!!! Cheers for the GL msgs Happygooner and Nade. I've got everything crossed - we need at least one of the draws to come in - there isn't enough cover in the other matches.

No I don't have a better alternative and to be honest i think they all seem to show the same books. I just use odds-comparison.co.uk which is linked to at the top of PL which has most but not all books. Heard on the grapevine Mansion didn't charge commission if the bet is matched straight away (only 1% if it isn't) and figured as a betting exchange the odds were worth checking out. Same thing happened last week also although by the time I got on to place bets odds had fallen into line.

[trollerboy]

Re: Predicting The English Premier League-using stats for oddscomparison: www.oddsoddsodds.com www.betexplorer.com www.tip-ex.com

[Hodgey]

Re: Predicting The English Premier League-using stats Late goals went your way today mate. :ok Well done, nice one. :clap

[Matthew]

Re: Predicting The English Premier League-using stats

Late goals went your way today mate. :ok Well done' date=' nice one. :clap[/quote'] I'm gonna stop watching Soccer Saturday. Newcastle had disallowed goals............everton score a late winner.........by the end of my tirade of abuse against Bolton, Fulham pop up with an equaliser. But then I guess that is why we do it.......;) £80 up for the day..........a dent in those losses.

Date Home Away Prediction Odds Bookie Stake Correct Return 01/09/2007 Bolton Everton D 3.40 Betfair £32.38 N 0 01/09/2007 Fulham Tottenham D 3.50 Betfair £32.38 Y 107.66 01/09/2007 Newcastle Wigan HW 1.73 StanJames £32.38 Y 56.02 01/09/2007 Liverpool Derby HW 1.20 StanJames £32.38 Y 38.86 01/09/2007 Man Utd Sunderland HW 1.22 BlueSq £32.38 Y 39.58

System Stats:

ALL HW D AW No of Bets: 27 5 16 6 Staked: £941.51 £162.2 £556.63 £222.68 Returned: £681.16 £177.49 £503.67 £0 P+L: £-260.35 £15.29 £-52.96 £-222.68 Strike Rate: 33.3% 80% 31.3% 0% Yield: -27.7% 9.4% -9.5% -100% Avg Back Odds: 3.27 1.62 3.59 3.79 Avg Win Odds: 2.49 1.37 3.38 0 Each Person: £38.68

[Matthew]

Re: Predicting The English Premier League-using stats Not many surprises for tomorrow. Thanks to HappyGooner for the Mansion pointer.........

Date Home Away Prediction Odds Bookie Stake Correct Return 02/09/2007 Arsenal Portsmouth D 3.98 Mansion £35.59 :hope 02/09/2007 Aston Villa Chelsea AW 1.81 Mansion £35.59 :hope 02/09/2007 Blackburn Man City D 3.43 Mansion £35.59 :hope

[Matthew]

Re: Predicting The English Premier League-using stats

Not many surprises for tomorrow. Thanks to HappyGooner for the Mansion pointer.........

Date Home Away Prediction Odds Bookie Stake Correct Return 02/09/2007 Arsenal Portsmouth D 3.98 Mansion £35.59 N 02/09/2007 Aston Villa Chelsea AW 1.81 Mansion £35.59 N 02/09/2007 Blackburn Man City D 3.43 Mansion £35.59 N

No surprises indeed.....which twonk said that!!!!! .....thats just wiped out saturdays good work.........and some. System Stats:

ALL HW D AW No of Bets: 30 5 18 7           Staked: £1048.28 £162.2 £627.81 £258.27 Returned: £681.16 £177.49 £503.67 £0 P+L: £-367.12 £15.29 £-124.14 £-258.27 Strike Rate: 30% 80% 27.8% 0% Yield: -35% 9.4% -19.8% -100%           Avg Back Odds: 3.25 1.62 3.6 3.5 Avg Win Odds: 2.49 1.37 3.38 0           Each Person: £34.04

[Matthew]

Re: Predicting The English Premier League-using stats

Date Home Away Prediction Odds Bookie Stake Correct Return 15/09/2007 Sunderland Reading D 3.50 Betfair £31.32 15/09/2007 West Ham Middlesbrough AW 4.30 Betfair £31.32 15/09/2007 Wigan Fulham D 3.30 vcbet £31.32 15/09/2007 Everton Man Utd AW 1.83 Betdirect £31.32 15/09/2007 Birmingham Bolton AW 3.00 Bet365 £31.32 15/09/2007 Chelsea Blackburn D 4.50 Skybet £31.32 15/09/2007 Portsmouth Liverpool D 3.80 Betfair £31.32

[Matthew]

Re: Predicting The English Premier League-using stats

Date Home Away Prediction Odds Bookie Stake Correct Return 15/09/2007 Sunderland Reading D 3.50 Betfair £31.32 N 15/09/2007 West Ham Middlesbrough AW 4.30 Betfair £31.32 N 15/09/2007 Wigan Fulham D 3.30 vcbet £31.32 Y £103.36 15/09/2007 Everton Man Utd AW 1.83 Betdirect £31.32 Y £57.42 15/09/2007 Birmingham Bolton AW 3.00 Bet365 £31.32 N 15/09/2007 Chelsea Blackburn D 4.50 Skybet £31.32 Y £140.94 15/09/2007 Portsmouth Liverpool D 3.80 Betfair £31.32 Y £114.50

The dancing bananas are back............:nana:nana:nana:nana Nearly £200 profit on the day............a big dent in those losses!!!!!

ALL HW D AW No of Bets: 37 5 22 10 Staked: £1,267.52 £162.20 £753.09 £352.23 Returned: £1,097.38 £177.49 £862.47 £57.42 P+L: £-170.14 £15.29 £109.38 £-294.81 Strike Rate: 35.1% 80% 36.4% 10% Yield: -13.4% 9.4% 14.5% -83.7% Avg Back Odds: 3.29 1.62 3.63 3.37 Avg Win Odds: 2.76 1.37 3.56 1.83 Each Person: £42.60

[Matthew]

Re: Predicting The English Premier League-using stats

Date Home Away Prediction Odds Bookie Stake Correct Return 16/09/2007 Man City Aston Villa AW 3.4 WillHill £39.19

[happygooner]

Re: Predicting The English Premier League-using stats

The dancing bananas are back............:nana:nana:nana:nana Nearly £200 profit on the day............a big dent in those losses!!!!!

ALL HW D AW No of Bets: 37 5 22 10 Staked: £1,267.52 £162.20 £753.09 £352.23 Returned: £1,097.38 £177.49 £862.47 £57.42 P+L: £-170.14 £15.29 £109.38 £-294.81 Strike Rate: 35.1% 80% 36.4% 10% Yield: -13.4% 9.4% 14.5% -83.7% Avg Back Odds: 3.29 1.62 3.63 3.37 Avg Win Odds: 2.76 1.37 3.56 1.83 Each Person: £42.60

Fairplay to you mate, you've certainly made my day! I've got Schmeichel and Richards in my fantasy team though, why did you're system have to come up with this game haha??!

[Wumpking]

Re: Predicting The English Premier League-using stats Well done Matthew - nice results today :clap

[Matthew]

Re: Predicting The English Premier League-using stats

Date Home Away Prediction Odds Bookie Stake Correct Return 16/09/2007 Man City Aston Villa AW 3.4 WillHill £39.19 N

ALL HW D AW No of Bets: 38 5 22 11 Staked: £1306.71 £162.2 £753.09 £391.42 Returned: £1097.38 £177.49 £862.47 £57.42 P+L: £-209.33 £15.29 £109.38 £-334 Strike Rate: 34.2% 80% 36.4% 9.1% Yield: -16% 9.4% 14.5% -85.3% Avg Back Odds: 3.29 1.62 3.63 3.37 Avg Win Odds: 2.76 1.37 3.56 1.83 Each Person: £40.90