AFL Power Ratings
 

These are my own power ratings for 2004. I plan on using them in 2005.

Port 92.5
Bris 92.8
St K 92.2
Geel 91.7
Melb 91.1
Syd 90.7
WCE 90.4
Ess 90.3
Fre 90.1
Kang 90.1
Carl 87.8
Ade 89.6
Coll 89.1
WBD 87.5
Haw 85.8
Rich 85.6

How do I turn them into percentage probabilities?

Benny

Can you please tell me how you arrived at those figures??? Then I might be able to help.

Port allowed 1823 pnts which you multipled by 10 then divide the pnts scorded. Subtract that from a hundred.

E.G Port 1823*10= 18230/2413=7.5 100-7.5= a power rating of 92.5

Does that help?

Benny

Take Geelong and Melbourne. You've got Geelong 91.7 and Melbourne 91.1. I'm assuming that it basically means Geelong is a six point better team than Melbourne on a neutral ground??? Is that right?

I haven't factored in the home ground advantage.

Benny

If you want to work out ratings based on this years Home & Away, you could possibly think about this: Take a team's winning % and add it to their for & against % and then divide by two to get an average score. This should then give you some indication of the points differences between the teams.

In this case, it would be:
105.0 Brisbane
104.9 Port Adel
100.3 St Kilda
94.1 Geelong
87.8 Melbourne
83.2 Sydney
81.4 West Coast
78.5 Essendon
75.3 Fremantle
72.9 Kangaroos
66.0 Adelaide
63.8 Collingwood
63.6 Carlton
51.2 W.Bulldogs
44.2 Hawthorn
43.7 Richmond

That's not counting the Finals series though. If you did that, of course Port would be ahead of Brisbane.

Thanks. I might try it next year, but after a couple of rounds. What do you think?

Benny

How do I covert the ratings into a percentage figure?

Benny

Okay, let's use your ratings. Take a game between say Carlton and Hawthorn at the MCG. You have Carlton on 87.8 and Hawthorn on 85.8. I've taken the liberty of just assuming that every .1 of a point is equivalent to 1 point on the scoreboard, because in the case of your ratings, I think that's not a bad assessment. Therefore, we have Carlton winning by 20 points.

Mo had a good formula for working out fair prices. It probably sounds a bit complicated, but I think it works well. I think the magic number was 0.955, wasn't it Mo? To find out the true value price for the favoured team Carlton, you multiply 0.955 by itself 20 times (once for every point you have Carlton winning by) and then add 1. Doing that, you end up with $1.40. To get Carlton's % chance, you simply divide 1.40 into 100 and you get 71.42%. Therefore Hawthorn's % chance is obviously 100 - 71.42, so 28.58%. Divide 100 by 28.58 and you get Hawthorn's value price of $3.50.

So therefore in that match, you would get:
Carlton (71.42%) $1.40
Hawthorn (28.58%) $3.50

Personally, I think it's a tad dangerous using ratings based on this year's results. I always like to use my own opinion. I've already done a revised set of ratings for next year. Obviously, I'll probably change it again slightly by the time the season rolls around, but at the moment, my ratings are like this:
126 Port Adel
120 St Kilda
117 Geelong
114 Brisbane
111 Sydney
108 West Coast
105 Essendon
102 Collingwood
99 Fremantle
96 Melbourne
96 Carlton
93 Adelaide
90 Kangaroos
84 West B'dogs
84 Hawthorn
84 Richmond

If you want, I could work out a table based on that formula of Mo's, so that you wouldn't have to work it out for every single game.

It would be handy to have a speadsheet to do all the calculations.

Just a thought.

Benny

In working out your ratings you say take the winning % and add it to the for and against %.

Please explain with an example as I'm havig trouble with it.

Benny

I'm having second thoughts about that idea. While I certainly think it's a good idea if you want to base it on a whole season's results, you wouldn't be able to do it on just a few games as the % figures would be all out of whack. You'd need a fair number of games to base it on.

Anyway, here's what I mean: Using this year's AFL Ladder, Port Adelaide had 17 wins from their 22 games. That's a winning % of 77.3%. Their for and against % was 132.4% Add 77.3 and 132.4 and you get 209.7. Then divide that by two to get their final rating of 104.85 which I rounded up to 104.9.

How can it be adapted for soccer?

Benny

Hmmmm, I'll have a look at that.

I've had a bit of a think about it. This looks like an okay system. You could add a team's Competition Points to it's Goal Difference and then divide that total by the number of games they've played. Again, you must make sure you have a fair few games to go on. Using this year's EPL table so far:

Team Pld GD Pts (RATING)
Chelsea 17 25 40 (3.82)
Everton 17 7 36 (2.53)
Arsenal 17 22 35 (3.35)
Man Utd 17 12 31 (2.53)
Middlesbrough 17 7 29 (2.12)
Liverpool 17 7 27 (2.00)
Aston Villa 17 3 25 (1.65)
Charlton 17 -8 24 (0.95)
Bolton 17 1 23 (1.41)
Tottenham 17 -1 22 (1.24)
Portsmouth 17 -2 22 (1.18)
Newcastle 17 -4 21 (1.00)
Man City 17 3 20 (1.35)
Fulham 17 -9 18 (0.65)
Birmingham 17 -5 17 (0.71)
Norwich 17 -11 15 (0.24)
Crystal Palace 17 -7 14 (0.41)
Blackburn 17 -14 14 (0.00)
Southampton 17 -9 13 (0.24)
West Bromwich 17 -17 10 (-0.41)

Now, I'm not totally sure how much advantage to give for home grounds. Perhaps 1 goal??? So say Arsenal is playing Man City at Arsenal, on these ratings they should win by exactly 3 goals. But most of them won't work out that well. Not quite sure what you should do when you have a team winning by .67 of a goal or something.

I reckon we should round off to the nearest half goal just to give us an idea. I've tried it out on this week's games. Let's see how they go.

Birmingham City vs West Bromwich Albion (Birmingham by 2 goals)
Blackburn Rovers vs Everton (Everton by 1.5 goals)
Bolton Wanderers vs Manchester City (Bolton by 1 goal)
Chelsea vs Norwich City (Chelsea by 4.5 goals)
Middlesbrough vs Aston Villa (Middlesborough by 1.5 goals)
Tottenham Hotspur vs Southampton (Tottenham by 2 goals)
Manchester United vs Crystal Palace (Man. Utd by 3 goals)
Liverpool vs Newcastle United (Liverpool by 2 goals)
Portsmouth vs Arsenal (Arsenal by 1 goal)
Charlton Athletic vs Fulham (Charlton by 1.5 goals)

Hmmm, not sure about those predictions. You'd think there'd be at least 1 draw somewhere.

So why not do that!
I do.
A1 Home team score
B1 Away team score
C1, 0.955^(A1-B1)+1

Of course your Inputs into the data cells must be in terms of actual points.
So with the ratings you originally came up with,Benny,you may have to multiply them by a factor of 10.
Sportznuts seems to have a much firmer grasp of what ratings are trying to represent!
Your ratings Benny,seem to get the teams in correct ability order,but don't really show the actual difference in ability between any 2 teams.
But that is fine for tipping competitions.

And Sportznut,It's nice to see someone remembers something useful that I have contributed,from amongst all the utter bull******** that have escaped from the clouded analysis of too much "Vitamin B".

Sportznut!
Had a look at your modified ratings.
Really have to disagree about what you have done to Brisbane.
Is it because Brisbane tend to be slow starters early and build up form through the year?
In that case I applaud your modification.
I think that shows why I tend to do poorly early in the season.Because I am using form from late last year and don't modify my ratings until I have some match results.
And the pre-season wizard cup doesn't really reflect what most teams will achieve during the early home and away games.
Anyway,114 for Brisbane for early games could be fine,but I wouldn't use such a low rating as that to predict their chances of a further premiership!
Good fortune to the new season,Sportznut and hopefully this forum builds up a large base of successful AFL punters.

Well, I've worked out a table of value prices for each predicted margin in the AFL. They are not totally exact, but I think they're close enough to give you a good rough guide. Thanks again to Mo for the formula which I've based this on. Haven't done it all in a spreadsheet for you Benny, but this should be some help.

Predicted
Margin - Team A - Team B
1....... - $1.95 - $2.05
2....... - $1.91 - $2.10
3....... - $1.87 - $2.15
4....... - $1.83 - $2.20
5....... - $1.80 - $2.25
6....... - $1.77 - $2.30
7....... - $1.73 - $2.37
8....... - $1.70 - $2.44
9....... - $1.67 - $2.50
10..... - $1.63 - $2.57
11..... - $1.60 - $2.65
12..... - $1.57 - $2.75
13..... - $1.55 - $2.82
14..... - $1.52 - $2.90
15..... - $1.50 - $3.00
16..... - $1.48 - $3.10
17..... - $1.46 - $3.20
18..... - $1.44 - $3.30
19..... - $1.42 - $3.40
20..... - $1.40 - $3.50
21..... - $1.38 - $3.65
22..... - $1.36 - $3.80
23..... - $1.34 - $3.95
24..... - $1.32 - $4.10
25..... - $1.30 - $4.30
26..... - $1.29 - $4.45
27..... - $1.28 - $4.57
28..... - $1.27 - $4.70
29..... - $1.26 - $4.85
30..... - $1.25 - $5.00
31..... - $1.24 - $5.17
32..... - $1.23 - $5.35
33..... - $1.22 - $5.55
34..... - $1.21 - $5.75
35..... - $1.20 - $6.00
36..... - $1.19 - $6.25
37..... - $1.18 - $6.55
38..... - $1.17 - $6.90
39..... - $1.16 - $7.25
40..... - $1.15 - $7.65
41- 42 - $1.14 - $8.15
43- 44 - $1.13 - $8.70
45- 46 - $1.12 - $9.30
47- 48 - $1.11 - $10.10
49- 50 - $1.10 - $11.00
51- 53 - $1.09 - $12.10
54- 56 - $1.08 - $13.50
57- 59 - $1.07 - $15.30
60- 64 - $1.06 - $17.60
65- 69 - $1.05 - $21.00
70- 74 - $1.04 - $26.00
75- 79 - $1.03 - $34.00
80- 89 - $1.02 - $51.00
90+.... - $1.01 - $101.00

Yes Mo, I've based it on them being slow starters and missing a few regular players from this year. As I said, I'll probably review it a bit more before I settle on the ratings for R1 though.

However, I'm quite willing to predict a slight decline for Brisbane next year. I still think they'll be a top team, but I reckon St Kilda and Geelong both have a chance to pass them.

Personally, I like Port's chances of repeating and I think we both may have gone off a year early with St Kilda. I reckon it's looking good for a Port/St Kilda GF. If not possibly Port/Geelong.

moeee can you give me an example.

Benny

Speak for yourself regarding the Saints in 2005,Sportznut.
I really thought they had what it takes this year.
I don't know if I'm being a bit harsh,but I just wonder what would have been if someone other than Grant Thomas was running the ship!
It's been brought to my attention that to get the best of a person,you need to compliment them on their good efforts.
Still remember the team photo of the Saints after their win of the Wizard Cup.Looked like they just heard of the Bali incident!
Anyway,Sportznut,I think the Saints have shot their load and will now fall back to earth.

Benny baby.You're a hard man to satisfy.
Try this!
Open up Excel.
For example,the game of Brisbane v St.Kilda.
Insert the score you think Brisbane will get into cell A1.
Insert the score you think St.Kilda will get into cell B1.
Insert the formula I mentioned into cell C1.
How hard could it be?
Any dramas please let me know,but please make an effort to try.

We'll see Mo. I'm not a huge fan of Grant Thomas either, but I still think the Saints are ready to win and I like the look of Port and St Kilda for the GF.

Anyway, what did you think of the table I worked out for the ratings?

Hi Sportznut.
I really didn't get into the nitty gritty fine detail of your chart.
But if you used the appropriate formula then it should be fine.
But I do find a spreadsheet program to be much more useful.
I have all those margins in my spreadsheet and I then look up the bookies prices available and input them on my spreadsheet.
Then I have another formula which compares my price to that available and computes the overlay.
And also the overlay available for any combination of games for my multi-bets.

Wow. Well done. But I guess in the end even with all that computer help, it's really down to us and our opinions.

With the soccer. I came up with this for the ManUtd v Crystal Palace.

Man Utd 2.53
Crystal Palace 0.41

Allowing 0.5 for each goal.

2.53-0.41=2.12

Man Utd by 4 goals

Price Man Utd $1.83 %=54.64
Crystal Palace $2.20. %=45.36

I'm not sure about the draw's % or price. Can someone help?

Benny

Benny,

You've got those teams TOO CLOSE!!! If you wait a while, I'll work out a table for you similar to the AFL one.

Well, a bit more difficult than I thought, so that chart will have to wait a little while. Trust me though, you've got those two teams too close together. Let's just see how those ratings go first.

Well, let's see how those Soccer ratings have gone so far:

Birmingham City vs West Bromwich Albion (Birmingham by 2 goals) Birmingham won 4-0
Blackburn Rovers vs Everton (Everton by 1.5 goals) Draw 0-0
Bolton Wanderers vs Manchester City (Bolton by 1 goal) Man City won 1-0
Chelsea vs Norwich City (Chelsea by 4.5 goals) Chelsea won 4-0
Middlesbrough vs Aston Villa (Middlesbrough by 1.5 goals) Middlesbrough won 3-0
Tottenham Hotspur vs Southampton (Tottenham by 2 goals) Tottenham won 5-1
Manchester United vs Crystal Palace (Man. Utd by 3 goals) Man Utd won 4-2

Not too bad I guess. Perhaps we should just stick to teams rated at least 2 goals ahead.

Got the wrong score there. It was 5-2, so Man Utd DID win by 3 goals. :)

Two more EPL games overnight. Here's how those ratings went:

Liverpool vs Newcastle United (Liverpool by 2 goals) Liverpool won 3-1
Portsmouth vs Arsenal (Arsenal by 1 goal) Arsenal won 1-0

Starting to look a little bit promising.

I have that problem all the time at the TAB Sportznut.
Seem to be getting a fair share of winners,but my wallet tells me different.
Where did all my money go?

You really need to back up your selections with a level staking profit/loss statement!

Benny,

I've worked out that prices table for Soccer. Word of warning though. I don't have anywhere near as much experience with Soccer prices as the AFL, NRL etc and the Draw option does make it a touch tricky. Even so, I've had a bit of a go and come up with something. I've rounded off the margins to the nearest quarter of a point. So let's say you work out Team A is rated 2.23 points ahead of Team B. You would go down to 2.25 and it says Team A should be $1.47, the Draw $4.80 and Team B $9.00. Looks a little harsh on Team B really, but after looking at some Soccer odds over the last week or so, these prices seem to work out okay. I didn't go down any further than 7 goals. Shouldn't need to go down any further than about 5 or 6 really.

Rated
Margin - Team A - Draw - Team B
0 Goals - $2.85 - $3.35 - $2.85
0.25.... - $2.57 - $3.40 - $3.17
0.50.... - $2.35 - $3.45 - $3.55
0.75.... - $2.15 - $3.50 - $4.00
1 Goal.. - $2.00 - $3.60 - $4.50
1.25.... - $1.87 - $3.70 - $5.10
1.50.... - $1.75 - $3.90 - $5.85
1.75.... - $1.65 - $4.10 - $6.70
2 Goals - $1.55 - $4.45 - $7.70
2.25.... - $1.47 - $4.80 - $9.00
2.50.... - $1.40 - $5.20 - $10.50
2.75.... - $1.35 - $5.70 - $12.00
3 Goals - $1.30 - $6.30 - $14.00
3.25.... - $1.26 - $7.00 - $16.30
3.50.... - $1.22 - $7.80 - $19.00
3.75.... - $1.19 - $8.80 - $22.00
4 Goals - $1.16 - $10.00 - $25.00
4.25.... - $1.14 - $11.40 - $30.00
4.50.... - $1.12 - $13.00 - $35.00
4.75.... - $1.10 - $15.00 - $40.00
5 Goals - $1.09 - $17.00 - $46.00
5.50.... - $1.08 - $20.00 - $51.00
6 Goals - $1.07 - $23.00 - $56.00
6.50.... - $1.06 - $27.00 - $66.00
7 Goals - $1.05 - $33.00 - $76.00

Remember, that as well as those Ratings that I worked out (which you can update after each round), I also think it's important to give the home team an extra goal for home ground advantage. The games in this round have worked out pretty close to how they were rated, so it's looking okay so far.

i have an idea for soccer ratings...

1. insert FIFA2005 to your computer
2. click 'play now'
3. choose teams to play each other
4. put 'controller select' in the middle so u arent controlling either team.. i.e CPU vs CPU.
5. see what happens

players and teams have quite accurate ratings so the system isnt THAT crazy.

Well, I don't have that game, or any other computer games for that matter. Why don't you do the honours for us? Play next weeks round of EPL and see what happens.

How much should one allow for the home ground advantage? And what other information should i use in compling the power ratings.

Benny

I came up with these:

Home ground advatage.

Adelaide 2.45
Brisbane 1.97
Carlton 2.04
Collingwood 2.95
Essendon 1.93
Fremantle 1.27
Geelong 1.65
Hawthorn 2.84
Melbourne 1.82
Kangaroos 1.73
Port Adelaide 1.02
Richmond 2.97
St Kilda 1.45
Sydney 1.63
West Coast 1.06
Western Bulldogs 2.87

These are an approximate.

Benny

Not quite sure what you've done there. Again, as with your other figures I guess we can move the decimal point one place to the left to get a real points advantage.

Those figures look wierd though. I assume you must have simply worked out how much better a team performs at home than away. That is not really home ground advantage and I'll tell you why. Port Adelaide has a very good home ground advantage, yet they have the lowest figure you've listed there. I guess that's because they have a very good away record and that must have skewed your figures.

In any case, when the Melbourne teams play each other, on most occasions there's NO real home ground advantage. The real home ground advantages come when teams have to travel interstate.

Do you see what I see?
 

Fremantle has the greatest home advantage in the league when they play at Subiaco.
Benny,my friend,you are not looking at your figures and asking yourself the question,"Do these look about right or have I made an error in judgement somewhere?"