Predicting The English Premier League-using stats-Strike Rate 57.2% - Yield 16.9%

[Matthew]

Originally Posted by plopplop

Saturdays Selections:
04/11/2006 Bolton Wigan - Lay Draw - 3.4
04/11/2006 Charlton Man City - HW - 2.46
04/11/2006 Fulham Everton - Lay Draw - 3.35
04/11/2006 Liverpool Reading - HW - 1.36
04/11/2006 Man United Portsmouth - HW - 1.32
04/11/2006 Newcastle Sheffield United - Lay AW - 5.6
04/11/2006 Watford Middlesboro - HW - 2.68

Being a Boro fan i am already not a big fan of newcastle. Now I have one more reason to dislike them.

[Matthew]

Staked:	17
Returned:	£18.43
P+L:	1.43
Strike Rate:	76.5%
Yield:	8.4%

[Matthew]

Date	Match	Prediction	Bookie	Odds
11/11/2006	Chelsea v Watford	HW	Stan James	1.16
11/11/2006	Man City v Newcastle	HW	Betfair	2.18
11/11/2006	Middlesbrough v West Ham	AW	Paddy Power	3.40
11/11/2006	Wigan v Charlton	HW	Expekt	1.95

[Matthew]

Date	Match	Prediction	Bookie	Odds
11/11/2006	Chelsea v Watford	HW	Stan James	1.16
11/11/2006	Man City v Newcastle	HW	Betfair	2.18
11/11/2006	Middlesbrough v West Ham	AW	Paddy Power	3.40
11/11/2006	Wigan v Charlton	HW	Expekt	1.95

No good today.

I hate betting against my team (BORO) but its bitter sweet with them betting WH - shame we neglected to do this last season in the FA cup semi final.

Staked:	21
Returned:	£21.54
P+L:	0.54
Strike Rate:	71.4%
Yield:	2.6%

[Matthew]

Date	Match	Prediction	Bookie	Odds
12/11/2006	Arsenal v Liverpool	HW	Coral	2.10

[Danny Cash]

From one Boro fan to another, well done so far and Up the Massimo!

[Matthew]

Up the massimo indeed - i was lucky enough to be at the uefa semi last year where he worked his magic yet again. Easily my best football related experience ever.

Originally Posted by plopplop

Arsenal V Liverpool: HOME WIN - 2.10 @ CORAL

Staked:	22
Returned:	£23.64
P+L:	1.64
Strike Rate:	72.7%
Yield:	7.5%

[Matthew]

Can somebody help…………

I am hoping someone can advise me on what they think is the best method to determine the cut-off’s for my prediction.

I have three models which in turn predict the likelihood of a HW, D & AW.

The package I use enables me to conduct some cost/benefit analysis to determine where the cut-off should be – but what are the costs & benefits.

I’ve looked at it in a two ways but can’t decide which is right and which is wrong.

Method 1:
I use the average HW decimal odds for the season 2005-2006 of 2.12
So, when I get the prediction right (predicted 1 x True 1) – I return 2.12pts, when I lose, -1pt.

This suggests the cut-off for my raw score is -0.172 (approx > 28% chance)
Classification rate tells you what %age of games I correctly classify. At this point Strike Rate is expected to be around 58.7% for HWs.

But is this right? Is it correct to put 2.12 in the profit box as that is what I am returned but not exactly profit – so should it be 1.12.

With the profit set to 1.12 my cut-off shifts to 0.003 (48% chance) and lifts my strike rate to 64.7%.


What should my profit be? 2.12 or 1.12?
I have the same problem for D and AW.


Method 2:
Very similar to the above but I derive the odds slightly differently.
The models I have built can either output the raw score which I have used above or a probability of the outcome happening.

When outputting the probability for each of HW, D, AW for a given match I can decide which outcome I think is more likely and back this.

I did this for the whole of the 2005-2006 season. I then took the average odds of all the HW games I predicted correctly to derive my expected profit, this was 1.8.

The same debate still rages about whether it should be 1.8 or 0.8 profit.


Shout up if this hasn’t made any sense. I should really be asleep!!!!!!!!!


My problem with Method 1 is that it includes the long odds of fluke results or results that go against expectation. This artificially raises the expected odds that I plumb in to find my cut-offs. I feel this is reasonable as I am questioning win odds that I would never predict and therefore should not expect to receive.

This is where Method 2 comes in – I remove the odds I am never going to achieve and take the average odds of the bets I get right in the season before.

I’ve no major problems with method 2 apart from I feel it might be self-fulfilling in some way but am yet to put my finger on it.


Anybody got any advice?

[Matthew]

Method 3:
Take the average odds for all the home games where home win has the lowest odds. This allows me to understand what I can expect to get if the game goes to form.

It doesn't include the freak results that method 1 does and it also allows me to understand what I would be gettng on average and not just what I might expect off the back of my predictions (method 2).



I'm probably gonna try all these variations and see what effect it has. But i am not a fan of retro fitting things to achieve the optimal result.

I'd much prefer to come up with my plan/strategy and play it out......

[Matthew]

Date	Match	Prediction	Bookie	Odds
18/11/2006	Arsenal v Newcastle	HW	Sporting Odds	1.3
18/11/2006	Chelsea v West Ham	HW	Expekt	1.23
18/11/2006	Man City v Fulham	HW	totesport	2.20
18/11/2006	Portsmouth v Watford	HW	totesport	1.67
18/11/2006	Sheff U v Man Utd	AW	betfred	1.36

[Matthew]

Originally Posted by plopplop

Staked:	£27.00
Returned:	£30.10
P+L:	£3.10
Strike Rate:	74.1%
Yield:	11.5%

Nothing for today.

[Matthew]

Date	Match	Prediction	Bookie	Odds
24/11/2006	Aston Villa v Boro	HW	Bet365	1.8
24/11/2006	Liverpool v Man City	HW	Totesport	1.5
24/11/2006	West Ham v Sheff Utd	HW	Totesport	1.83

[Matthew]

Originally Posted by plopplop

2 out of 3 aint bad..........

a small profit of .33pts.

Staked:	£30.00
Returned:	£33.43
P+L:	£3.43
Strike Rate:	73.3%
Yield:	11.4%

[Matthew]

Date	Match	Prediction	Bookie	Odds
26/11/2006	Man Utd v Chelsea	AW	Totesport	3.1
26/11/2006	Tottenham v Wigan	HW	Premierbet	1.83

[mcgin]

Good work Plopplop, looks like your doing quite well. Will be keeping an eye on this.

[Matthew]

Originally Posted by mcgin

Good work Plopplop, looks like your doing quite well. Will be keeping an eye on this.

Thanks mcgin, I'm trying my best

I'm nearly at the same point with the Spanish Primeria Liga. I'll term that as "coming soon......." but it is looking promising

Date	Match	Prediction	Bookie	Odds
26/11/2006	Man Utd v Chelsea	AW	Totesport	3.1
26/11/2006	Tottenham v Wigan	HW	Premierbet	1.83

Staked:	£32.00
Returned:	£35.26
P+L:	£3.26
Strike Rate:	71.9%
Yield:	10.2%

[Danny Cash]

Originally Posted by plopplop

2 out of 3 aint bad..........

a small profit of .33pts.

Surely you didn't really want 3/3 though?!

[Matthew]

Originally Posted by Danny Cash

Surely you didn't really want 3/3 though?!

Quite right, I didn't.

It's always very bittersweet when Boro crop up in my selections.
I'd happily lose 1pt everytime to see Boro progress.

I was happy to see Christie back - irrespective of how far offside his stands - it's about time we got one of those.

TONIGHTS GAME:

- no selection.
Although I don't think it is going to be a draw - I keep getting stung on the lays so I will leave it.

[Matthew]

No surprises here...........

Date	Match	Prediction	Bookie	Odds
29/11/2006	Bolton v Chelsea	AW	BlueSQ	1.67
29/11/2006	Man Utd v Everton	HW	Stan James	1.33

For those that might care:
I dont think Man City will win
I dont think Fulham will win
I dont think Portsmouth will win


Maybe I should have layed Watford v Sheff Utd............lessons learned.

[Matthew]

Originally Posted by plopplop

For those that might care:
I dont think Man City will win
I dont think Fulham will win
I dont think Portsmouth will win


Maybe I should have layed Watford v Sheff Utd............lessons learned.

No lessons to learn - I should stay away from laying!!!!