Storeytime 2015/16

I just had a look at last years Lge Table and imo there were 16 fixtures that you could convincingly-ish say were harder to win than the other 24. Altho, for everybody 2 of the 40 in total were impossible to win, although they do kinda exist. Chelsea got to play 2 games that weren't against themselves whereas everyone else had the disadvantage of facing 2 of THE most difficult games being chelsea home & chelsea away.

So, you could say, you don't have to balance any sections/groups you create because it kinda does it itself. The fixtures for the weaker sides are bound to be more tricky, innit. So any gains they make do have very slightly more value in the wonderful world of theoretical maths & stats.
 
BazzaBear said:
People are getting carried away here with a quest for accuracy. It doesn't matter.

It doesn't matter if 83 points is really what the winners will get, it doesn't matter if the teams stated as bottom 6 are really in the bottom 6. It simply matters that every team is judged against the same criteria, so their relative performance can be gauged.

Putting tons of effort into improving the formula is pointless.
Click to expand...

Not quite. In fact, if that is all that mattered we wouldn't have a point for a storeytime table at all. We could just say "three points for a win, one point for a draw. It simply matters that every team is judged against the same criteria". The idea behind these things is that while every team plays every other team home and away by the end of the season, the sequence that each team plays them is different. Thus the desire to measure how each team has actually fared, where the simple table doesn't give enough information.

So we need to evaluate the difficulty of each fixture played to judge a team by its merits. If we simply use the original peterstorey model of assigning teams into groups at the beginning of the season we run into a couple of problems. The first is that it is inflexible and doesn't account for more recent and much more relevant data. The second is that it creates arbitrary cutoff points. The expected value of a match against Swansea or Southampton is the same as the expected value of a match against Leicester City. A match against Spurs is the same as a match against Chelsea. A match against West Brom is supposedly much harder than a match against Aston Villa. Forcing the tier system onto non linear data erodes the value of the whole thing. Bad data in, bad data out.

The best way to do it, that I can think of, is to evaluate based on actual data home and away fixtures against each opponent during the current season. But even that is useless until about January. And when I post that table, people don't seem to understand what it is measuring. It's not a predictor, it's just a league table with adjustments for strength of schedule built in.
 
@Eboue I hadn't realised this was 2013, I was trying to work out how it fitted into last season, :wenger:. I am an idiot & can only offer my grovelling apologies for this.

It's perfect - the Arse / Spurs is an interesting comparison <--- we 'think' they are closer than the real table because of the fixture pattern, as we 'think' are Chelsea to City.

where 'think' = data suggests, obv :)

http://i.imgur.com/EJzGXrV.png

although it's more obv again at the bottom - Sunderland & Wigan
 
Last edited:
Points collected in 15/16 versus the same fixtures in 14/15:

Leicester +5
Liverpool +3
Swansea +3
Man Utd +2
Man City +2
Palace +2
Everton + 1
Tottenham + 1
Watford** + 1
Bournemouth* 0
West Brom 0
West Ham 0
Newcastle 0
Aston Villa -1
Norwich*** -1
Sunderland -1
Arsenal -3
Chelsea -3
Southampton -5
Stoke -5

*Hull = Bournemouth
**Burnley = Watford
***QPR = Norwich



The league table - as it stands:

1. Chelsea 84
2. Man City 81
3. Arsenal 72
4. Man Utd 72
5. Tottenham 65
6. Liverpool 65
7. Swansea 59
8. Southampton 55
9. Crystal Palace 50
10. Stoke 49
11. Everton 48
12. West Ham 47
13. Leicester 46
14. West Brom 44
15. Newcastle 39
16. Sunderland 37
17. Aston Villa 37
18. Bournemouth 35
19. Watford 34
20. Norwich 29
 
Eboue said:
The best way to do it, that I can think of, is to evaluate based on actual data home and away fixtures against each opponent during the current season. But even that is useless until about January. And when I post that table, people don't seem to understand what it is measuring. It's not a predictor, it's just a league table with adjustments for strength of schedule built in.
Click to expand...

I think this is what you've already tried to do Eboue (as per the quoted paragraph); but I would suggest the most accurate way to do things would be to look at Average Points Yield (APY) from a visit to Villa Park, for example, and use that to determine where teams stand against one another

e.g.

It's January 1st and Villa have played 10 games:

5 at home} 2W 1D 2L average points yield 1.4
5 away} 0W 1D 4L average points yield 2.6

Anybody who plays Villa after Jan 1st gets 1.4pts added to their points total if it's at Villa Park or 2.6 if they play them at home.

Do this for all the teams and BAM we can just call it a day in January.

Does that make any sense? I hope so! It still fails to account properly for the schedule because Villa could have just had 5 tricky aways pre-January but I'm gambling that will be evened out by somebody elses easier schedule
 
Now I understand it, I like Eboue's. Also, now I'm in the correct season it seems to work out kinda 'right' too.
 
Eboue said:
Not quite. In fact, if that is all that mattered we wouldn't have a point for a storeytime table at all. We could just say "three points for a win, one point for a draw. It simply matters that every team is judged against the same criteria".
Click to expand...

No, because that wouldn't be judging them by the same criteria would it? You need weighting, but as long as the weighting is vaguely sensible and consistent, it does the job. In theory your system of judging on the current seasons merits is more accurate. Except, as you admit yourself, it renders itself almost completely useless for the first half of the season. And after that point, the difference between such an alternative table and the actual table gets smaller and smaller anyway. So tell me again how adding extra complexity to try to get a higher degree of accuracy makes an alternative table better?

In reality, using last years standings as a measure of likely performance is a reasonable approximation. There will be teams that perform very differently, outliers, but as a blunt tool it's enough to give the vague indication which is all this idea should be aiming to achieve. An attempt to get 'accuracy' is doomed to failure anyway. It just adds complexity, confuses 99% of the people viewing it, and doesn't actually achieve any better a guide.
 
BazzaBear said:
An attempt to get 'accuracy' is doomed to failure anyway. It just adds complexity, confuses 99% of the people viewing it, and doesn't actually achieve any better a guide.
Click to expand...

Of course it is better. It's silly for you to keep saying this. It might not be as much better as you would think justifies extra time but it is clearly superior and more accurate. If people are incapable or unwilling to exercise critical thinking when entering this thread, that's their problem. I'm not interested in convincing the masses.
 
Blatzo said:
I think this is what you've already tried to do Eboue (as per the quoted paragraph); but I would suggest the most accurate way to do things would be to look at Average Points Yield (APY) from a visit to Villa Park, for example, and use that to determine where teams stand against one another

e.g.

It's January 1st and Villa have played 10 games:

5 at home} 2W 1D 2L average points yield 1.4
5 away} 0W 1D 4L average points yield 2.6

Anybody who plays Villa after Jan 1st gets 1.4pts added to their points total if it's at Villa Park or 2.6 if they play them at home.

Do this for all the teams and BAM we can just call it a day in January.

Does that make any sense? I hope so! It still fails to account properly for the schedule because Villa could have just had 5 tricky aways pre-January but I'm gambling that will be evened out by somebody elses easier schedule
Click to expand...

Yes that's the gist, difference being that what I did before would be updated as more data came in.
 
Eboue said:
Of course it is better. It's silly for you to keep saying this. It might not be as much better as you would think justifies extra time but it is clearly superior and more accurate. If people are incapable or unwilling to exercise critical thinking when entering this thread, that's their problem. I'm not interested in convincing the masses.
Click to expand...

It's better towards the end of the season, it's worse towards the beginning. Does that balance out as better over all? When it's less needed at the end of the season anyway?

Nothing that can be done is anything better than a blunt instrument, since we're literally trying to predict the future. Your system pretends at 'accuracy' when in reality it is at least nearly as limited as the simpler system overall. Here's an analogy: It's like measuring a length in mm to 4 decimal places, when you're using a tape measure. Yes. 5.155 sounds 'more accurate' than 5. But what are you really achieving with that accuracy?

In that latter part of the season, I already conceded that it is going to be more accurate, so there's no point you telling me that like you're disabusing me of a misconception. But if we're going to use one system throughout the season, I think it fails to confer a definite advantage - while implying of itself some kind of 'rightness' that it can't promise to achieve.


Now ... if you really wanted to put effort in to get a better system (which I'd still argue is unnecessary, and even counter-productive in terms of the system conveying an understandable message) then you'd start off with results weighted against purely last years standings, you'd end with results weighted purely against this years standings, and throughout the year you'd have a mechanic where the weighting gradually changes between the two from game week to game week! Now that's a complex system! :lol:
 
You're a terrible analogy.

The point is, you're trying to use a scalpel to do the job of a hammer. However much accuracy you pretend to, you can't predict the future.
 
Currently looking at...

Actual Points + (Games Left x 2) - Number of ''Worst 16 Games'' Left = Silly Figure much likely to change by next May but used as a comparison.

and am pleasantly surprised tbh, for the current season it would suggest that Arsenal are 4pts behind City/Utd rather than 3pts because of the West Ham home loss. The 1 pt theoretical difference from actuality is because of City v Chelsea, Utd v Spurs. Arsenal haven't played any of the more 'orrible 16 games yet. (palace away not terribly difficult according to last season).

I'm super confident of this one, :wenger: think it works well in the scenarios I can think of.
 
Matchday 2
After game two:

+2: Leicester City, Liverpool, Manchester City
+1: Swansea City
0: Everton, Manchester United, West Ham United
-1: Arsenal
-2: Chelsea, Stoke City, Tottenham Hotspur, Watford
-3: Aston Villa, Crystal Palace, Newcastle United, Norwich City
-4: AFC Bournemouth
-5: Southampton, West Bromwich Albion
-6: Sunderland
 
If I post this estimated rough draft of how I think mine stands in mythical model points

63pts City, Utd, Liverpool
62pts Leicester
61pts Swansea, Everton
60pts West Ham and others
59pts Arsenal and others
58pts Chelsea and others
others but no one has fewer than 56/57 pts

I've got to ask how (rhetorical) we are a point behind Swansea when we have 2 more than them in the 'real world' That's a 3point swing where they only got 1 point from their game at Chelsea which is what is switching things in their favour. It can't be true can it? Or rather it's too big an error in the model.

(Utd are par, Swans are + 1 to par)
 
Last edited:
adexkola said:
Posters used it as a "facebook like" on here.
Click to expand...

fecking idiots, you know how useful + 1 can be in football discussions? :mad:

Re: the thread... every fecking year, same discussion, same shit as before like with the FIFA Rankings. It's an alternative for early season, it is useful, it's not a fecking crystal ball, get on with it.

PS: Nothing to do with you specifically
 
Points collected in 15/16 versus the same fixtures in 14/15:

Leicester +6
Liverpool +6
Palace +5
Man City +4
Swansea +3
Bournemouth* +2
Newcastle + 1
Everton 0
Man Utd 0
Chelsea 0
Norwich*** -1
Sunderland -1
Tottenham -1
Watford** -1
West Brom -3
West Ham -3
Aston Villa -4
Southampton -4
Stoke -5
Arsenal -5


*Bournemouth = Hull
**Watford = Burnley
***Norwich = QPR



The league table - as it stands:

1. Chelsea 87
2. Man City 83
3. Arsenal 70
4. Man Utd 70
---------
5. Liverpool 68
6. Tottenham 63
7. Swansea 59
8. Southampton 56
9. Crystal Palace 53
10. Stoke 49
11. Everton 47
12. Leicester 47
13. West Ham 44
14. West Brom 41
15. Newcastle 40
16. Sunderland 37
17. Bournemouth 37
---------
18. Aston Villa 34
19. Watford 32
20. Norwich 29
 
Samid said:
Points collected in 15/16 versus the same fixtures in 14/15:

Leicester +6
Liverpool +6
Palace +5
Man City +4
Swansea +3
Bournemouth* +2
Newcastle + 1
Everton 0
Man Utd 0
Chelsea 0
Norwich*** -1
Sunderland -1
Tottenham -1
Watford** -1
West Brom -3
West Ham -3
Aston Villa -4
Southampton -4
Stoke -5
Arsenal -5


*Bournemouth = Hull
**Watford = Burnley
***Norwich = QPR



The league table - as it stands:

1. Chelsea 87
2. Man City 83
3. Arsenal 70
4. Man Utd 70
---------
5. Liverpool 68
6. Tottenham 63
7. Swansea 59
8. Southampton 56
9. Crystal Palace 53
10. Stoke 49
11. Everton 47
12. Leicester 47
13. West Ham 44
14. West Brom 41
15. Newcastle 40
16. Sunderland 37
17. Bournemouth 37
---------
18. Aston Villa 34
19. Watford 32
20. Norwich 29
Click to expand...

I have no idea what this means.
 
Sassy Colin said:
I have no idea what this means.
Click to expand...

Points collected this season compared to the exact same fixtures last season. For example United beat Newcastle at home last season (3 points) but only managed a draw (1 point) this season. That means United are -2 points for that particular fixture.

The table is how it stands if we count the last home/away fixtures (results) played between all clubs.
 
Samid said:
Points collected this season compared to the exact same fixtures last season. For example United beat Newcastle at home last season (3 points) but only managed a draw (1 point) this season. That means United are -2 points for that particular fixture.

The table is how it stands if we count the last home/away fixtures played between all clubs.
Click to expand...

Alright, but why? Just tell me why you are doing this. :(
 
<wades in>

This one is actually a lot more interesting than I thought it would be - although having dismissed it as ''utter b*ll*cks'' that wasn't all that difficult. In this instance however (quiet at the back) I am a buffoon.

This model weights the fixtures (by comparison to last year) and then projects forward via a combination of actual results added to expected results. And it's not a bad idea to take into account that the perceived strength of a team (last years results) will outway current form - theoretically. Chelsea will be hoping that's the case anyway.

And the anomalies of results give this a bit of accidental flexibility too, not sure 'flexibility' is the right word but I know what I mean.

Quite interesting innit, overall.

(West Ham must have b*ggered up the other one something shocking)
 
Samid said:
Points collected this season compared to the exact same fixtures last season. For example United beat Newcastle at home last season (3 points) but only managed a draw (1 point) this season. That means United are -2 points for that particular fixture.

The table is how it stands if we count the last home/away fixtures (results) played between all clubs.
Click to expand...

Are you going to post the latest updated version?
 
Samid said:
To prove that my method is more effective and gives a more realistic view on the league table than "Storeytime" and Eboue's.
Click to expand...
I don't see how it would? Obviously they'll look very similar in the final third of the season, but early on I don't see what benefit it provides? No-one could look at that table and understand that City have had a great start to the season, with Chelsea the complete opposite.
 
GlastonSpur said:
Are you going to post the latest updated version?
Click to expand...

Points collected in 15/16 versus the same fixtures in 14/15:

Man City +6
Palace +4
Everton +4
West Ham +3
Norwich*** +3
Watford** +3
Bournemouth* +2
Leicester +2
Tottenham + 1
Liverpool + 1
West Brom + 1
Swansea 0
Newcastle 0
Man Utd 0
Sunderland -3
Southampton -3
Arsenal -5
Chelsea -6
Stoke -7
Aston Villa -7


*Bournemouth = Hull
**Watford = Burnley
***Norwich = QPR


The league table - last home/away results against every opposition:

1. Man City 85
2. Chelsea 81
3. Arsenal 70
4. Man Utd 70
---------
5. Tottenham 65
6. Liverpool 63
7. Southampton 57
8. Swansea 56
9. Crystal Palace 52
10. Everton 51
11. West Ham 50
12. Stoke 47
13. West Brom 45
14. Leicester 43
15. Newcastle 39
16. Bournemouth 37
17. Watford 36
---------
18. Sunderland 35
19. Norwich 33
20. Aston Villa 31
 
index.php
 
ThierryHenry said:
I don't see how it would? Obviously they'll look very similar in the final third of the season, but early on I don't see what benefit it provides? No-one could look at that table and understand that City have had a great start to the season, with Chelsea the complete opposite.
Click to expand...

What? City are 12 points better off than Chelsea already after 6 games compared to the same fixtures last year. Chelsea have been losing fixtures they were winning last year while City are winning those they were losing. It's fairly evident in the table.

This gives it a more personal touch for every club. Rather than having a general formula for the whole league you look at individual games. Each fixture has its own history. Example: Up until last season Sunderland were beating City 1-0 at home four seasons running. "Storeytime" would've given -3 to City as it is a fixture they're expected to win. My method would give both teams 0 as history from the previous season was repeating itself.

Also, "Storeytime" has a par score of 83 points. That formula focuses mostly on the top 4-5 teams. My formula is for all 20 clubs.
 
Samid said:
What? City are 12 points better off than Chelsea already after 6 games compared to the same fixtures last year. Chelsea have been losing fixtures they were winning last year while City are winning those they were losing. It's fairly evident in the table.

This gives it a more personal touch for every club. Rather than having a general formula for the whole league you look at individual games. Each fixture has its own history. Example: Up until last season Sunderland were beating City 1-0 at home four seasons running. "Storeytime" would've given -3 to City as it is a fixture they're expected to win. My method would give both teams 0 as history from the previous season was repeating itself.

Also, "Storeytime" has a par score of 83 points. That formula focuses mostly on the top 4-5 teams. My formula is for all 20 clubs.
Click to expand...
You're over complicating things for something simple.
 
How does the RAWK alternate premier league table work?
 
Not sure this belongs in here or a thread of its own, but I've made my own version of the alternative league table.

Groupings are Top 4, Next 6, Bottom Half, with the expectations being that you win all of your home games, lose away against the Top 4, draw away against the Next 6, and win away against the Bottom Half. Although distinct groups, to ensure the weighting is equal for all teams, 4th position is considered part of the Next 6 for all non-Top-4 teams, and 10th position is considered part of the Bottom Half for all Bottom Half teams. Meeting the expectations should see a team finish with 93 points, with 29 wins, 6 draws, and 3 defeats.

As an added level, the groupings are dynamic. Week 1 groupings were allocated according to the final positions in 2014/15, with the newly promoted sides replacing the three that were relegated in the Bottom Half. As teams move above or below the expected par, the groupings change following each week's results. Where teams are level on points in the alternative structure, their standings in the actual table act as the tie breaker.

As we're now approaching Week 7 I've completed the 6 games already played and the table currently looks like this:

West Ham United +3
Manchester City + 1
Manchester United -1
Leicester City -1
Arsenal -4
Tottenham Hotspur -4
Watford -4
Everton -5
Liverpool -5
Bournemouth -5
Swansea City -6
Crystal Palace -7
Newcastle United -7
Norwich City -8
Chelsea -9
West Bromwich Albion -10
Aston Villa -10
Stoke City -11
Southampton -12
Sunderland -16
 
You must log in or register to reply here.