Automatic Promotion Target Supporting Data

[ucandomagic]

Cappy asked in my graph thread about my Autos Certain benchmark line, how it was established and whether it is still appropriate in the second half of this season. All of my benchmarks (Autos and Playoffs) are created before the season starts. The Autos Certain line starts with an estimate of the final points total required to finish the season in the Top 2. The teams are grouped into 3 or 4 different quality levels, based on their previous season’s performances, and then I have a model which takes our schedule and comes up with what I believe is the easiest route through our schedule to deliver that final pre-defined total. As a result, comparing our actual results each week against the benchmark as the season progresses is a more realistic comparison than simply comparing against a straight line to the final target.

The end of season target is simply based on historical data. I started doing this a couple of years ago and at that time no team this century, at this level, had ever achieved 90 points and not been in the Top 2. Hence, I took 90 points as a target for the Autos Certain benchmark. Chart 1 below shows the end of season data for the Top 3 teams over the 25 seasons this century. However last year, 2023/24, Leeds did only finish 3rd on 90 points.

Chart 1:



You can see from Chart 1 that, before 2024, even 89 points had only been achieved once by the 3rd place team, in 2016 by Brighton. Graph 1 shows visually this variation over 25 years alongside the 90 point line.

Graph 1:



So, if last year was different from normal, how does this year compare to last year? Chart 2 shows where we are this year compared to last year at the present stage, and at the end of the season. You can see that the 2 seasons are quite similar, except that by this time last year Leicester had established a good gap at the top of the table. In both seasons there are realistically 4 teams with a chance of Top 2, with the 4th team on 51 points.

Chart 2:



So, if we are possibly in another atypical season, what should be the likely target? Blades have the easiest remaining fixtures and Leeds have the hardest. Taking the performances to date of the Top 4, and the strengths of the teams to face in the final 19 games gives the very acceptable end of season Blades forecast in Chart 3.

Chart 3:


So, if Chart 3 is about correct for the other 3 teams, Blades would actually have a target of 92 points to achieve second place.

All of this is highly speculative with 19 games to go, so I will keep the 90 point target for the present but will review it at the 10 and 5 games-to-go points to check that it still appropriate. Significant injuries or transfers in and out can always change the situation.

Hopefully, this hasn’t been too detailed, but in reality I think that the graphs and charts tell the underlying story on their own.

UTB & Slava Ukraini!

 

[WorkingtonBlade]

Really interesting, thank you.

I looked at the table before reading what you wrote, and my instinctive way of looking at it was to look at the 2nd place totals. On this basis, in 9 of the 25 seasons, 90 points would not have been enough to gain automatic promotion (I counted the 2 seasons where it would depend on goal difference as half each). That’s a 36% chance of 90 points not being enough.

It’s also notable that in 3 of the last 4 seasons it wouldn’t have been enough.

The way this season is developing, it feels probable to me that 90 won’t be enough. Your analysis of the remaining fixtures is reassuring though!

 

[Bladezz]

Love the work as always.

 

[GreasyChipBeattie]

Really interesting, thank you.

I looked at the table before reading what you wrote, and my instinctive way of looking at it was to look at the 2nd place totals. On this basis, in 9 of the 25 seasons, 90 points would not have been enough to gain automatic promotion (I counted the 2 seasons where it would depend on goal difference as half each). That’s a 36% chance of 90 points not being enough.

It’s also notable that in 3 of the last 4 seasons it wouldn’t have been enough.

Eh?
Might have been better reading first mate!
I think maybe you're quoting the number of times 90 WAS achieved for second place, so on 16 of the 25 occasions less that 90 points was enough.

 

[Snake]

Fantastic work mate. Usually I go with historical stats but this season I've got a gut feeling that 92 points will be needed for 2nd place. Just hope that 2 point deduction doesn't bite us on the arse

 

[Snowflake]

Cappy asked in my graph thread about my Autos Certain benchmark line, how it was established and whether it is still appropriate in the second half of this season. All of my benchmarks (Autos and Playoffs) are created before the season starts. The Autos Certain line starts with an estimate of the final points total required to finish the season in the Top 2. The teams are grouped into 3 or 4 different quality levels, based on their previous season’s performances, and then I have a model which takes our schedule and comes up with what I believe is the easiest route through our schedule to deliver that final pre-defined total. As a result, comparing our actual results each week against the benchmark as the season progresses is a more realistic comparison than simply comparing against a straight line to the final target.

The end of season target is simply based on historical data. I started doing this a couple of years ago and at that time no team this century, at this level, had ever achieved 90 points and not been in the Top 2. Hence, I took 90 points as a target for the Autos Certain benchmark. Chart 1 below shows the end of season data for the Top 3 teams over the 25 seasons this century. However last year, 2023/24, Leeds did only finish 3rd on 90 points.

Chart 1:
View attachment 202279


You can see from Chart 1 that, before 2024, even 89 points had only been achieved once by the 3rd place team, in 2016 by Brighton. Graph 1 shows visually this variation over 25 years alongside the 90 point line.

Graph 1:
View attachment 202281


So, if last year was different from normal, how does this year compare to last year? Chart 2 shows where we are this year compared to last year at the present stage, and at the end of the season. You can see that the 2 seasons are quite similar, except that by this time last year Leicester had established a good gap at the top of the table. In both seasons there are realistically 4 teams with a chance of Top 2, with the 4th team on 51 points.

Chart 2:
View attachment 202282


So, if we are possibly in another atypical season, what should be the likely target? Blades have the easiest remaining fixtures and Leeds have the hardest. Taking the performances to date of the Top 4, and the strengths of the teams to face in the final 19 games gives the very acceptable end of season Blades forecast in Chart 3.

Chart 3:
View attachment 202283

So, if Chart 3 is about correct for the other 3 teams, Blades would actually have a target of 92 points to achieve second place.

All of this is highly speculative with 19 games to go, so I will keep the 90 point target for the present but will review it at the 10 and 5 games-to-go points to check that it still appropriate. Significant injuries or transfers in and out can always change the situation.

Hopefully, this hasn’t been too detailed, but in reality I think that the graphs and charts tell the underlying story on their own.

UTB & Slava Ukraini!

From a young age I have been fascinated by numbers, graphs etc., so your regular offerings are a pleasure and no more so than this explanation.
I look forward to another PL season, but will blame your figures and not the players if we don’t make it !!!!

 

[Fatlad666]

Brilliant post, alot of time and effort gone into this, thank you for sharing it

 

[WorkingtonBlade]

Eh?
Might have been better reading first mate!
I think maybe you're quoting the number of times 90 WAS achieved for second place, so on 16 of the 25 occasions less that 90 points was enough.

The way I’m thinking of it is, if a team were to finish on 90 points in each of those 25 seasons, would they get automatic promotion? The answer is ‘no’ in 9 out of the 25 seasons - and 3 out of the last 4 - because in those seasons there were at least 2 teams who finished on more than 90 points.

 

[de g.]

Really interesting, thank you.

I looked at the table before reading what you wrote, and my instinctive way of looking at it was to look at the 2nd place totals. On this basis, in 9 of the 25 seasons, 90 points would not have been enough to gain automatic promotion (I counted the 2 seasons where it would depend on goal difference as half each). That’s a 36% chance of 90 points not being enough.

It’s also notable that in 3 of the last 4 seasons it wouldn’t have been enough.

The way this season is developing, it feels probable to me that 90 won’t be enough. Your analysis of the remaining fixtures is reassuring though!

That's the wrong way to look at it - promotion is achieved by beating the third-place team. Only once since 2000 has that required more than 90 points, and never more than 91. Even on the occasions where 2nd got 91+, they still only needed 91 for that position.

 

[WorkingtonBlade]

That's the wrong way to look at it - promotion is achieved by beating the third-place team. Only once since 2000 has that required more than 90 points, and never more than 91. Even on the occasions where 2nd got 91+, they still only needed 91 for that position.

I can see it both ways. You’re working on the basis of “how many points did the team which DID finish 2nd NEED to get to do so”. I’m working on the basis of “if you were another team in that division, how many points would you need to get to finish 2nd?”

In other words, you’re taking the team which actually finished 2nd out of the equation, whereas I’m leaving it in.

 

[Selly]

That's the wrong way to look at it - promotion is achieved by beating the third-place team. Only once since 2000 has that required more than 90 points, and never more than 91. Even on the occasions where 2nd got 91+, they still only needed 91 for that position.

I can see it both ways. You’re working on the basis of “how many points did the team which DID finish 2nd NEED to get to do so”. I’m working on the basis of “if you were another team in that division, how many points would you need to get to finish 2nd?”

In other words, you’re taking the team which actually finished 2nd out of the equation, whereas I’m leaving it in.

The weakness in both methods is it assumes all teams are playing 100% right up to game 46.

Just look at our promotion season under Wilder last time.

Game 46 for us was Stoke - and we were already up.
Leeds were playing someone - no idea who - no idea what the result was

For different reasons neither game was important to the teams in 2nd/3rd playing. Who is to say how many points each team would have got if game 46 was important to both sides?

(I don`t see a way of resolving this issue either)

 

[diode_blade]

Excellent work Ucando
Now can the hospital have my cardiogram chart back please.

 

[WorkingtonBlade]

The weakness in both methods is it assumes all teams are playing 100% right up to game 46.

Just look at our promotion season under Wilder last time.

Game 46 for us was Stoke - and we were already up.
Leeds were playing someone - no idea who - no idea what the result was

For different reasons neither game was important to the teams in 2nd/3rd playing. Who is to say how many points each team would have got if game 46 was important to both sides?

(I don`t see a way of resolving this issue either)

I wonder whether it’s statically true that a team is more likely to win its last game of the season if it needs to do so (to gain promotion or avoid relegation). I remember occasions when we’ve already achieved our aim by the last game when we’ve played with a lot of freedom and looked better than we had prior to that (away at Luton a few years ago?) and occasions when we’ve desperately needed a result and looked paralysed (too many to mention, but Wigan at home will always be a painful memory).

On the other hand, as you say, sometimes a team which doesn’t need a result just looks disinterested (or as if they may have over-indulged somewhat the previous night…) and drops points.

 

[Selly]

I wonder whether it’s statically true that a team is more likely to win its last game of the season if it needs to do so (to gain promotion or avoid relegation).

For me this is a moot point - and fundamentally unknowable.

What we do know is that if the team in 3rd has no chance of catching 2nd then the situation has undeniably changed - and the situation clearly impacts the performance on the pitch (either negatively or positively).

After the Ipswich game in 2019, Leeds suddenly looked more like themselves the day after against Villa - having looked completely shot the prior weekend. pressure was (more or less) off them at that point as even two wins would have seen them needing a 9 goal swing or something.

Whereas against Stoke we looked awful - most likely because of the 5 day bender the entire squad was on.

It would be a huge amount of work - but I wonder what the points threshold (and after how many games) is for 2nd placed teams getting promotion confirmed.

 

[de g.]

I can see it both ways. You’re working on the basis of “how many points did the team which DID finish 2nd NEED to get to do so”. I’m working on the basis of “if you were another team in that division, how many points would you need to get to finish 2nd?”

In other words, you’re taking the team which actually finished 2nd out of the equation, whereas I’m leaving it in.

I don't think your method works, because it relies on reshuffling the pack (one team getting more points) without assuming that this affects the number of points gained by first or second place.

 

[Bennyand the Blades]

GEE - ORE
ME HEAD HURTS......

 

[Iffy Onuora's Shiny Head]

We play a lot of the relegation threatened teams away from home. On paper they look easy, but they will all be fighting for their lives and if they get manage to get ahead, or keep scores at nil nil for a long time, it will give energy to the home fans.

Investing in attacking players (which we seem to be doing) is going to be critical to kill some of these teams off.

 

[WorkingtonBlade]

I don't think your method works, because it relies on reshuffling the pack (one team getting more points) without assuming that this affects the number of points gained by first or second place.

I think the same argument could be used for either method. If you see it from the perspective of the 2nd place team and say (last season for example) that although they got 96 points, they only needed 91 points to get 2nd, then who do the dropped points go to? If any of them go to the third place team (or another team close behind) that changes the answer.

For me the bigger discrepancy is caused by discounting entirely the team that actually finished 2nd. Although it’s only 1 team, it’s 50% of the teams in the division who were good enough to get automatic promotion, so it has a big impact. When considering how confident we can be that 90 points will be enough for promotion it creates a big swing, because the average gap between 2nd and 3rd over the 25 years was 5 points (bigger than I would have guessed).

Either way, this year it really feels as if we have 4 teams capable of getting 90 points. I doubt they’ll all do it, but I suspect 3 will.

 

[de g.]

I think the same argument could be used for either method. If you see it from the perspective of the 2nd place team and say (last season for example) that although they got 96 points, they only needed 91 points to get 2nd, then who do the dropped points go to? If any of them go to the third place team (or another team close behind) that changes the answer.

For me the bigger discrepancy is caused by discounting entirely the team that actually finished 2nd. Although it’s only 1 team, it’s 50% of the teams in the division who were good enough to get automatic promotion, so it has a big impact. When considering how confident we can be that 90 points will be enough for promotion it creates a big swing, because the average gap between 2nd and 3rd over the 25 years was 5 points (bigger than I would have guessed).

Either way, this year it really feels as if we have 4 teams capable of getting 90 points. I doubt they’ll all do it, but I suspect 3 will.

It works because I'm not looking at the details of any specific season, but at the general trends. Over that 24 year period, 140 clubs have finished in the top two of Champ/Lg1/Lg2 (discounting 20-21 for Lgs 1&2 where the seasons were curtailed). Only three times has a club got 90+ and not got promoted: Leeds 23-24, Blades 11-12, and Wednesday 22-23 (96 points!). So there is a roughly 2% chance of 90 not being enough.

 

[WorkingtonBlade]

It works because I'm not looking at the details of any specific season, but at the general trends. Over that 24 year period, 140 clubs have finished in the top two of Champ/Lg1/Lg2 (discounting 20-21 for Lgs 1&2 where the seasons were curtailed). Only three times has a club got 90+ and not got promoted: Leeds 23-24, Blades 11-12, and Wednesday 22-23 (96 points!). So there is a roughly 2% chance of 90 not being enough.

Interesting - thank you. That’s reassuring! I still have a horrible feeling this season is going to be one of the 2% and that - in the Championship at least - that % will grow in the next few years as the impact of the increasing gap between Premier League and Championship takes it toll.

However, all of that aside, it’s an encouraging statistic, as I do still feel we have a good chance of getting 90 points this season, and ultimately I believe more in statistics than I do in gut-feel or my own typical Blades-fan pessimism.

 

[Houston_Blade]

Love this stuff.

Also gives a good opportunity to post this gif:

 

[vas borbokis]

Good work ucando.
Good discussion too!
Let's just win this bastard league eh?

 

[Blade82]

It’s crazy that 95 points would not have been enough last season.

If that’s repeated this season, it could mean we pick up 97 points across the season but still miss out with the -2 deduction.

 

[Houston_Blade]

It’s crazy that 95 points would not have been enough last season.

If that’s repeated this season, it could mean we pick up 97 points across the season but still miss out with the -2 deduction.

You only need to get more than 3rd so 91 points would have been enough.

 

[Blade82]

You only need to get more than 3rd so 91 points would have been enough.

Leeds needed 96 point last season to go up (on goal difference).

Ipswich would have needed another 3 points to win the league.

If two teams set the same benchmark this season 98(1st) & 96(2nd) the other 22 teams need to beat it for automatic promotion.

 

[Houston_Blade]

Leeds needed 96 point last season to go up (on goal difference).

Ipswich would have needed another 3 points to win the league.

If two teams set the same benchmark this season 98(1st) & 96(2nd) the other 22 teams need to beat it for automatic promotion.

3rd placed Leeds got 90 points, therefore to finish 2nd you’d have needed to beat this total or equalled it and had better GD. The fact that Ipswich got 96 doesn’t matter, they only needed 91.

 

[ThirdGenBlade]

There are 2 issues here. Firstly, when you are looking at a statistical sample it is incorrect to argue using a single outlier because you are only working on probabilities.

Looking at this set of data, if the external conditions have been consistent then there is a 96% probability of Top 2 with 90 points.

However, that other 4% was last year, which would seem to imply that external conditions have changed - which they have. Parachute payments mean that there is a group of clubs with significantly greater assets than the majority, and hence points will tend to be much higher in that smaller group and increase the probability of more than 2 clubs getting more than 90 points.

That was the whole reason that ucandomagic looked at the performance and future schedule of the Top 4 and forecast that Blades would be Top 2 from current performance, not based on history.

UTB & FTP!

 

[BrisBlade]

3rd placed Leeds got 90 points, therefore to finish 2nd you’d have needed to beat this total or equalled it and had better GD. The fact that Ipswich got 96 doesn’t matter, they only needed 91.

You can look at this two ways: Ipswich could have got 91 points and still gone up or Leeds needed 96 points to go up. It depends what you treat as variable and what as fixed. Both are perfectly valid.

 

[Houston_Blade]

You can look at this two ways: Ipswich could have got 91 points and still gone up or Leeds needed 96 points to go up. It depends what you treat as variable and what as fixed. Both are perfectly valid.

Fair point, however if our points total is more than 3rd place this season we’ll go up. So that’s the number to worry about.

 

[GymClassHero]

I personally feel 90 pts will be enough. There is clear daylight between the top four and the rest, but nobody is running away with it and teams have been dropping points.