Topic: The Alternative Premier League Table 2013-14

[Hij]

Please keep posting Woolster, it's certainly an interesting method, do you have a link for the thread you are currently running, might be easier to look at both, unless people don't mind an amalgamation of methods in this thread -guess its up to Prof.

[Red number seven]

Or as you said stop chucking the par 3's. We need to just focus on ourselves. If we keep parring(in regards to 71 points) from now till the end of the season we will very likely get top 4

IF we keep parring, we will get 87 points and compete for the title, if I understand the system.

[The Woolster]

Please keep posting Woolster, it's certainly an interesting method, do you have a link for the thread you are currently running, might be easier to look at both, unless people don't mind an amalgamation of methods in this thread -guess its up to Prof.

Thanks. What with the other forums I read/use plus twitter, I will probably only be an occasional poster on here.

The forum I am posting this on is the liverpoolway, but I am posting on the members part of the site, which is paid for I'm afraid. Perhaps I'll post my table on here occasionally/semi regularly rather than weekly?

[The Woolster]

Mine is simply from logic, I've not put close to as much thought as Prof into it or confirming with past results, but that was also not really my plan. One thing I have since checked following a comment is the amount of wins the 4th place team has had over the last 5 or 6 years, which has ranged between 19 and 21, and my model has 20 wins needed, so is about right in that sense.

It half came about following criticism of Rodgers last season not being able to beat the top teams, and my arguement at the time was that if we beat all the teams from 10th down home and away we'd have 66 points and would only need 4 points from the rest of our games to get to 70 points, so its not far from that.

My guess is that the past data will show that the teams fighting for 4th place sometimes lose those games that it shouldn't because of random variation, so it has an off day when a weaker team has a good day, and that somtimes they take points off the top 2 or 3 when the opposite occurs, but there are the 3 or so teams around them who are more or less equal and anthing can happen and the team in 4th has got 'lucky'. My guess is that overall historically, the team that finishes 4th will actually get less than the 48 points I assign for the middle set of teams, but have beaten the teams around them (3rd to 7th) to make up the points.

A team that is fighting for 4th will of course be more inconsistant than a team fighting for the title, and there is no way of gussing when they will have their good or bad days, which is hard enough as it is for those top teams as I think Prof has mentioned having read back, so I think the only way to assign the expected scores is by logic, as there won't be a trend in performance of the 4th placed team in terms of who they lose against or whether it is at home or not, other than home scores will be somewhat better than away.

Have also remebered that I am comparing this seasons results to the corresponding fixtures last season (which I think I've seen mentioned on this thread that this is also done elsewhere on here), so I do have last seasons data.

Of those 10 mid table teams, where the pars in my method would see a team realisticall fighting for 4th looking for 48 points (from 60), but my guess being that they actually get a little less, Utd scored 51 points, City scored 45 points, Chelsea scored 42 points, Arsenal scored 44 points, Spurs scored 46 points, Everton scored 37 and we scored 35. The dropped points are distributed pretty randomly though.

Arsenal only scored 11 against the top 7 teams, Chelsea got lucky and scored 22 (more than Utd's 20).

So I think it goes to show that for the teams who are in the fight for 4th, you can't really predict where they will drop/gain points, so can only work on logic, whereas for the teams winning the title, there is much less variation and history shows they win the vast majority of their home games.

[I-was-watching-it-on-Sky]

This is very interesting...I tried to replicate this last night with all the current top 6 teams [I left Everton out....wonder why].

A slight adjustment though...I squeezed the two bars together...top line is home form...bottom line is away form.

With all six bars one on top of another I think it will give a good pictorial representation of the relative strength of a team's home/ away form and also a direct comparison with rival top-6 team's form. 

I was also thinking of a way to automatically build-in the 'par' element into it.

This is my version of this 'Form Bar' which Prof suggested up to date @ game 7.

[Prof]

This is my version of this 'Form Bar' which Prof suggested up to date @ game 7.

Nicely done 

I'll leave it to you keep that up-to-date for the season if you're happy to 

Do the dots above the lines mean anything?  Am I correct saying they indicate how many matches at home/away have been played?

[Prof]

So I think it goes to show that for the teams who are in the fight for 4th, you can't really predict where they will drop/gain points, so can only work on logic, whereas for the teams winning the title, there is much less variation and history shows they win the vast majority of their home games.

This is why I argued against having a different set of pars for top four results.  My theory was that teams who come top four play like  champions for some of the season, but can't sustain it for 38 games.  Predicting where the champions can afford to drop points is hard enough, and they don't drop many.  I doubt a suitable method of doing it for a 70/71 point season could be worked out, but feel free to try.

My approach was to average out the  additional dropped points over the season (0.5 points per game) to create the 71 point line.  Looking at previous top four teams, it seems to illustrate quite early whether teams are on track or not.

[Blinis]

So I think it goes to show that for the teams who are in the fight for 4th, you can't really predict where they will drop/gain points, so can only work on logic, whereas for the teams winning the title, there is much less variation and history shows they win the vast majority of their home games.

I totally agree with you, I did not wanted to say that logic was not fine; it is perfectly fine to me! It's just that, as you said, it is way more difficult to find a simple pattern for top 4 results than for title winning form. But feel free to try, we could even make several models in competition, let them run, and then decide which one is better 

This thread gets more and more interesting, keep going guys!

[I-was-watching-it-on-Sky]

Nicely done 

I'll leave it to you keep that up-to-date for the season if you're happy to 

Do the dots above the lines mean anything?  Am I correct saying they indicate how many matches at home/away have been played?

Yeah...you got it...the dots indicate games home/ away [helps me count].

I'll keep it up to date n post it regularly...possibly learn how to do it and make it look a reasonable size.

The green lines are at the 51 & 38pt thresholds for top spot & the blue lines are for top-4 performance [average over last 10 seasons].

It would be nice if people comment on how your idea indicates differences in performance between these major sides.

[The Woolster]

This is why I argued against having a different set of pars for top four results.  My theory was that teams who come top four play like  champions for some of the season, but can't sustain it for 38 games.  Predicting where the champions can afford to drop points is hard enough, and they don't drop many.  I doubt a suitable method of doing it for a 70/71 point season could be worked out, but feel free to try.

My approach was to average out the  additional dropped points over the season (0.5 points per game) to create the 71 point line.  Looking at previous top four teams, it seems to illustrate quite early whether teams are on track or not.

I think though, if you are looking at a model to predict how a team will score the 70/71/72 points or whatever for top 4, it doesn't need to be as realistic in a prediction sense, as its simply not possible due to the massive role that luck plays in single results and seasons as a whole, and the further down the table a team is, the bigger role luck plays. So for me, as long as its logical, its fine.

Keeping with the golf analogy, and you may have said something similar in these threads but I've only skimmed back and from what I've seen you've covered all bases, but the par score of a hole is not actually the average score of the hole. Some par 4s will have an average of say 3.8, and others will have an average of 4.3, but a par 4  for both still makes sense. So in terms of teams finishing in 4th, My guess is that history will show that from the bottom 3, although par is 18 points, the average might be 15ish, from mid 10, whilst par is 48 points, the average will be around  42ish, and from the top 7, whilst the par score is 5, the average is 14ish. But to attempt to predict where points will be dropped/gained would not make any logical sense and would be futile attempting to do so, so something that simply makes sense is fine.

I've had a bit to drink, so not sure if I've made sense there 

[I-was-watching-it-on-Sky]

I think though, if you are looking at a model to predict how a team will score the 70/71/72 points or whatever for top 4, it doesn't need to be as realistic in a prediction sense, as its simply not possible due to the massive role that luck plays in single results and seasons as a whole, and the further down the table a team is, the bigger role luck plays. So for me, as long as its logical, its fine.

Keeping with the golf analogy, and you may have said something similar in these threads but I've only skimmed back and from what I've seen you've covered all bases, but the par score of a hole is not actually the average score of the hole. Some par 4s will have an average of say 3.8, and others will have an average of 4.3, but a par 4  for both still makes sense. So in terms of teams finishing in 4th, My guess is that history will show that from the bottom 3, although par is 18 points, the average might be 15ish, from mid 10, whilst par is 48 points, the average will be around  42ish, and from the top 7, whilst the par score is 5, the average is 14ish. But to attempt to predict where points will be dropped/gained would not make any logical sense and would be futile attempting to do so, so something that simply makes sense is fine.

I've had a bit to drink, so not sure if I've made sense there 

As a rule of thumb...the more simple the 'model'...the more useful it is for general application.

The beauty of Prof's model is that it doesn't attempt to pick when/ where/ against whom a side will drop points...the model would need to be so complicated...so sophisticated...that ya average bear couldn't follow it.

Besides...each factor you built into a model would be subject to error at every point.

I'm an engineer...we're the kinds of guys who build engines...get men to the moon...

If you're designing an engine...it's 'possible' to build a mathematical model...which could determine the Vector of every single atom inside the chamber at every point in time...but we don't do that.

We make assumptions that a single point remains constant over a given time...then just work out what goes in...what comes out.


Prof's model is a bit like this way of working...all the various/ individual 'factors' which have a bearing on performance are ignored...and we just take as our 'point in time' the status of current form at this moment [after each round of games].

In the early part of the season [say week one] you only have a small amount of data to build the trend line with.

The data which goes in week by week is 100% correct and becomes more reliable to build the trend line on...

And it is only about 'trends'...not hard n fast 'predictions'.

I estimate that by week 25...it's about +/- 3% accurate.

[The Woolster]

My model doesn't aim to predict when where or whom points are dropped against either, its very similar to Prof's just with a different target.

[Prof]

As a rule of thumb...the more simple the 'model'...the more useful it is for general application.

The beauty of Prof's model is that it doesn't attempt to pick when/ where/ against whom a side will drop points...the model would need to be so complicated...so sophisticated...that ya average bear couldn't follow it.

Besides...each factor you built into a model would be subject to error at every point.

I'm an engineer...we're the kinds of guys who build engines...get men to the moon...

If you're designing an engine...it's 'possible' to build a mathematical model...which could determine the Vector of every single atom inside the chamber at every point in time...but we don't do that.

We make assumptions that a single point remains constant over a given time...then just work out what goes in...what comes out.


Prof's model is a bit like this way of working...all the various/ individual 'factors' which have a bearing on performance are ignored...and we just take as our 'point in time' the status of current form at this moment [after each round of games].

In the early part of the season [say week one] you only have a small amount of data to build the trend line with.

The data which goes in week by week is 100% correct and becomes more reliable to build the trend line on...

And it is only about 'trends'...not hard n fast 'predictions'.

I estimate that by week 25...it's about +/- 3% accurate.

Very well put.  Thanks.

[I-was-watching-it-on-Sky]

My model doesn't aim to predict when where or whom points are dropped against either, its very similar to Prof's just with a different target.

Oops...

I didn't mean to offend...rather to reiterate the point of the usefulness of a 'simple' model.

[The Woolster]

No worries, I wasn't offended, just confused. It happens a lot...

[I-was-watching-it-on-Sky]

It depends on the chances though. We've conceded almost 50% of our goals through set-pieces this season, which is a relatively easy fix (relative to open play defending).

Most of the shots (54%) we concede are outside the box. As long as Mignolet is comfortable with that, then there is not as much of an issue as if we were, say, City, who concede 11% of their shots in the 6 yard box, or Spurs, who concede 48% of their shots in the 18 yard box. That means attackers are getting in behind their defences more regularly, while Liverpool are keeping the attackers in front of them more. Unsurprisingly, the top two teams for forcing the opposition to shoot outside the box are the same top two teams who occupy joint first - us and Arsenal. We've only conceded 5 goals in the league in 7 games, and 2 of those were in the same game. So we've only conceded in 4 games. At the rate we're going, we'll be heading for a 25-30 goals against season. That's definitely Top 4 defending, if not title-challenging defending.

I was thinking about your "...25-30 goals against this season..." comment over a pint at lunchtime...

Humm...

Naturally this depends on how you calculate it...but...do you agree with this...?

5 GA in 7 games...gives (5/7)x 31) games left in the rest of the season = 22.14...plus the 5 GA already conceded = 27.14 [we're on track for 27-28 GA].

However, we've conceded in only 4 games...

So...((4/7) x 31) + 4 games in which we've already conceded = 21.7 (22).

That suggests we'd have 38 - 22 = 16 clean sheets...

So...going back ((5/4) x 22) + 5 = 32.5 (33 GA).

In 6 out of the last ten seasons...that's champions form...


I just know that's wrong...but we are human...we dare to dream.

[Prof]

I was thinking about your "...25-30 goals against this season..." comment over a pint at lunchtime...

Humm...

Naturally this depends on how you calculate it...but...do you agree with this...?

5 GA in 7 games...gives (5/7)x 31) games left in the rest of the season = 22.14...plus the 5 GA already conceded = 27.14 [we're on track for 27-28 GA].

However, we've conceded in only 4 games...

So...((4/7) x 31) + 4 games in which we've already conceded = 21.7 (22).

That suggests we'd have 38 - 22 = 16 clean sheets...

So...going back ((5/4) x 22) + 5 = 32.5 (33 GA).

In 6 out of the last ten seasons...that's champions form...


I just know that's wrong...but we are human...we dare to dream.

In the last sum, you don't need to add 5 as you're saying we'll concede in 22 games at a rate of 5/4 in each of these (including the 4 we've already conceded in).  So 27.5 is correct either way of working it out.

What it doesn't take into account is that we've played 4 home and 3 away.  Also, we've only played 1 'big' team.  So it's highly skewed, exactly what the APLT is designed to offset 

[I-was-watching-it-on-Sky]

In the last sum, you don't need to add 5 as you're saying we'll concede in 22 games at a rate of 5/4 in each of these (including the 4 we've already conceded in).  So 27.5 is correct either way of working it out.

What it doesn't take into account is that we've played 4 home and 3 away.  Also, we've only played 1 'big' team.  So it's highly skewed, exactly what the APLT is designed to offset 

Yeah...I see it now...

My excuses are...

A) Attila the wife broke my concentration...most likely talking about important stuff like soft furnishings.

B) I'm not very bright...but I can't half lift heavy things.

Take your pick...

[zanwalk]

Brilliant thread, thanks lads. I too am fascinated by statistics, unfortunately I am also too lazy to learn how to put these charts together...

[Prof]

Yeah...I see it now...

My excuses are...

A) Attila the wife broke my concentration...most likely talking about important stuff like soft furnishings.

B) I'm not very bright...but I can't half lift heavy things.

Take your pick...

I pick A.  I can empathise with that.  Seriously, do we need another 'throw' for the end of the bed?  They serve no purpose and we have 4 in what appears to be a squad rotation system already   

[Frizzo]

I pick A.  I can empathise with that.  Seriously, do we need another 'throw' for the end of the bed?  They serve no purpose and we have 4 in what appears to be a squad rotation system already   

Prof, if I didn't love you enough already, that post seals the deal.

Quality. Absolute balls to the wall quality.

[McrRed]

looks good to me. Top 4 is the goal and that's where we are.

Big result for Chelsea.  Spurs the big losers this weekend.

It puts our real table position into perspective 

But I'm guessing doesn't count interlopers like the current Southampton... not a problem as long as they drop off at some point?

[I-was-watching-it-on-Sky]

In the last sum, you don't need to add 5 as you're saying we'll concede in 22 games at a rate of 5/4 in each of these (including the 4 we've already conceded in).  So 27.5 is correct either way of working it out.

What it doesn't take into account is that we've played 4 home and 3 away.  Also, we've only played 1 'big' team.  So it's highly skewed, exactly what the APLT is designed to offset 

OK Prof...I've had a wee think about this...

Any comments to help keep me on the right track here...see what you think of this as a method.

I'd appreciate anyone chiming-in...

The average GF over the last 10 seasons is 79 [let's call it 78 to make the numbers a tad easier].

So...it takes 78 goals to earn 90pts = 0.866 G/ pt.

Of the 90pts target...78 are earned in your 26 par 3 games... therefore 0.866 x 78pts = 67.6 [68 goals] in your par 3 games.

67.6/ 26 = 2.6 goals in each par 3 game...apply this to par 1 games and we get 10 goals in 12 par 1 games...works out to be 0.866 goals per game.

So...for us this season 6 x par 3 games & 1 x par 1 games

Therefore... ((6 x 2.6) + (1 x 0.866)) = 16.466 [16] goals... we're in fact -5 against a 'par' for goals.


Before I carry on and work this out a bit more...any comments and/ or corrections guys...?

[Prof]

That' an interesting idea.  You are essentially saying that for every point, you need 0.866 goals.  We are -3 on the APLT, so in theory we should be 0.866 x 3 = 2.6 goals behind where we  we should be in the GF column, but we are -5.  Does this mean we are over achieving based on our goal scoring?  The first few results being 1-0 wins probably accounts for this.

It'll be interesting to monitor this, and maybe look at the goals per points correlation across the league from previous seasons.

[I-was-watching-it-on-Sky]

That' an interesting idea.  You are essentially saying that for every point, you need 0.866 goals.  We are -3 on the APLT, so in theory we should be 0.866 x 3 = 2.6 goals behind where we  we should be in the GF column, but we are -5.  Does this mean we are over achieving based on our goal scoring?  The first few results being 1-0 wins probably accounts for this.

It'll be interesting to monitor this, and maybe look at the goals per points correlation across the league from previous seasons.

Yeah...I'll have time on my hands from the end of this month...I'll look into that.

Though I think it's a but more complex to work out than 'X' goals per point...

It seems to me that it would be a reasonable expectation that the 'par' for goals in any particular game would mirror the degree of difficulty of that game [home to Sunderland is different than away to the Scum in Mordor]. 

I was also thinking...if 'par' is 2.6 goals per par 3 game...there's no point in humping a side by 6-0...3-0 is good enough and be good enough to spread the other 3 round a bit.

I was looking at goals GF/ GA/ GD for both home & away games over the last ten years for the top-4 sides.

There's such a wide spread in all the numbers...it's hard to figure out which is the decisive thing to measure/ use as thee factor to build a 'par' structure around.

I'm going to turn this over in my mind a bit more...massage some numbers n see where it goes.

[Prof]

Good luck 

I'd start by looking at correlations between points and goals scored.

[Anthony]

I'm an engineer...we're the kinds of guys who build engines...get men to the moon...

If you're designing an engine...it's 'possible' to build a mathematical model...which could determine the Vector of every single atom inside the chamber at every point in time...but we don't do that.

We make assumptions that a single point remains constant over a given time...then just work out what goes in...what comes out.


Prof's model is a bit like this way of working...all the various/ individual 'factors' which have a bearing on performance are ignored...and we just take as our 'point in time' the status of current form at this moment [after each round of games].

Or, in other words, this is Rocket Science!

Trying to pick this up being a long time fan of the Corresponding Results measure in JP's thread - still boggled but keep up the good work!

[mikey_LFC]

Would there not have to be some adjustments to the 90 point line based on the number of teams capable of competing for the title. Most would agree that this year it will take significantly less than that to win the league due to the amount of teams who could make the top 4, thereby making it our each clubs start better than compared to previous years.

[Jizzinho]

I pick A.  I can empathise with that.  Seriously, do we need another 'throw' for the end of the bed?  They serve no purpose and we have 4 in what appears to be a squad rotation system already   

Ha ha ha ha I'm feelin' this.

Having just recently been married and moved in together I had a sit-down talk with my wife about something she was concerned about. Thought she was going to tell me she had found a lump or the cat had died. Turns out I was getting a bollocking for not pulling back the curtains and making the bed without the "throw" as I leave for work later than her in the morning.

I was tempted to ask her again what the bloody thing was actually for but - just as with the APLT - some people will never get it.

[Prof]

Would there not have to be some adjustments to the 90 point line based on the number of teams capable of competing for the title. Most would agree that this year it will take significantly less than that to win the league due to the amount of teams who could make the top 4, thereby making it our each clubs start better than compared to previous years.

Time will tell, but the aplt has never been about winning against the big teams.

[I-was-watching-it-on-Sky]

Would there not have to be some adjustments to the 90 point line based on the number of teams capable of competing for the title. Most would agree that this year it will take significantly less than that to win the league due to the amount of teams who could make the top 4, thereby making it our each clubs start better than compared to previous years.

Humm...

The thing is that the results aren't really convincing either way on this point.

The two closest title races [in terms of the points spread from champions to 4th place] are 11 in 2008 [87 to 76] and 12 in 2011 [80 to 68].

I've calculated that the 'actual' average points to win the PL over the last 10 years is 88pts [87 over the last 5 years].

So...I don't think that a 'close race' really affects the points total of the champions by very much [depending on how you calculate it - 8%].


In the year with the biggest points spread [34pts in 2005], 4th place achieved 61pts. Not massively lower than 68pts in 2011.

In reality...the 'target' is 2nd place + 1pt...over the last ten years that's an average of 78pts +1 [72pts in 2011 & 90pts in 2012].


I've no idea what conclusions to draw from any of it...

[I-was-watching-it-on-Sky]

Ha ha ha ha I'm feelin' this.

Having just recently been married and moved in together I had a sit-down talk with my wife about something she was concerned about. Thought she was going to tell me she had found a lump or the cat had died. Turns out I was getting a bollocking for not pulling back the curtains and making the bed without the "throw" as I leave for work later than her in the morning.

I was tempted to ask her again what the bloody thing was actually for but - just as with the APLT - some people will never get it.

Hell's teeth...man...!

I've been married 30yrs...it never goes away.

As my dad used to say...you can't live with them...[imagine a l_o_n_g sigh here] and you can't live with them...

[I-was-watching-it-on-Sky]

Time will tell, but the aplt has never been about winning against the big teams.

Yeah...understood Prof...

But I can't help thinking that which games you pick up the points makes a difference.

There are years where a side's total GF or GD is better than the side placed above them...so where they're getting the points from must be relevant.

I'm still not sure how to rationalise it...

[Prof]

Yeah...understood Prof...

But I can't help thinking that which games you pick up the points makes a difference.

There are years where a side's total GF or GD is better than the side placed above them...so where they're getting the points from must be relevant.

I'm still not sure how to rationalise it...

You can't.  It's called variance, as is taken into account in all mathematical models.  Football is a game of chance, so the idea is you maximise your chance of winning, but don't guarantee it.  Over a season, variance of individual results is generally levelled, but never removed.

I could go on, but I'm on my phone.  Let me know if you want further explanation and I'll try later 

[Frizzo]

Ha ha ha ha I'm feelin' this.

Having just recently been married and moved in together I had a sit-down talk with my wife about something she was concerned about. Thought she was going to tell me she had found a lump or the cat had died. Turns out I was getting a bollocking for not pulling back the curtains and making the bed without the "throw" as I leave for work later than her in the morning.

I was tempted to ask her again what the bloody thing was actually for but - just as with the APLT - some people will never get it.

No, don't compare throw pillows with the aplt!!!

The aplt at least has a purpose!! 

[The Woolster]

In reality...the 'target' is 2nd place + 1pt...over the last ten years that's an average of 78pts +1 [72pts in 2011 & 90pts in 2012].

Not sure about that. If you are the team in 1st place all the time, you could say "we would have won it if we only got 1 more than 2nd anyway", otherwise, if you had finished 1 point more than 2nd, you'd have finished 2nd.

Or am I confused again?

[I-was-watching-it-on-Sky]

Not sure about that. If you are the team in 1st place all the time, you could say "we would have won it if we only got 1 more than 2nd anyway", otherwise, if you had finished 1 point more than 2nd, you'd have finished 2nd.

Or am I confused again?

Maybe not my best explanation...I'll try n clarify...

There's an advert on TV now featuring Dick Fosbury...his revolutionary new style of high jump [at least is was when I was wearing penny-round shirt collars].

So I'll try n keep it topical...

If the guy destined to finish in 2nd place has jumped 2m...and you're the last to jump...all you have to do to be first is...2m + 1cm.

There's no point in jumping 2m 34cm...you're still only first...with 33cm of ego.

That's why I say that the 'real' target is only 1pt better than 2nd.

I was wrong bye the way...the average for 2nd place over the last ten seasons is 82pts...so the target is [on average] in the region of 83pts+ [n I've kinda got a wee feeling in my water that about 83pts will win it this year].

[Prof]

All that assumes that the team that comes second hasn't taken their foot off the gas.

Also, if we wanted to win the league last year, we needed the total of the winners plus one.  That's the key point.

[Frizzo]

All that assumes that the team that comes second hasn't taken their foot off the gas.

Also, if we wanted to win the league last year, we needed the total of the winners plus one.  That's the key point.

and then consequently we'd have the total of the second place team +1

[Prof]

If the egg came before the chicken...?