BST points prediction thread…

[ForsolongaRover]

The only advantage I can think of from having played more games is that having an extra midweek game might fatigue our rivals more (when you include the travel, where applicable) and also have a knock-on effect on their following Saturday game due to them having fewer rest/preparation days. Could this take away a fraction of the advantage of having games in hand? Might be clutching at straws though as it's all a bit intangible.

FWIW we'll not have to wait long until the games played (mostly) level out....

Tuesday 25th Feb
Bromley v Bradford
Cheltenham v Walsall

Tuesday 11th March
Wimbledon v Cheltenham
Colchester v Port Vale
Grimsby v Notts County

Tuesday 25th March
Port Vale v Barrow

Mostly tough-looking games-in-hand for our promotion rivals on paper, so you never know. Cheltenham are in good form and have a big say, Colchester away is a tricky game-in-hand for Vale, as is Notts County's visit to Grimsby. Bradford have to go to Bromley who are no mugs, and Bradford's away record as previously highlighted is not great (W3, D6, L6).

Maybe someone has already expressed agreement with this and perhaps it is not at all in dispute, but having to play more than a once a week successively is a very salient factor. It weighs significantly on fitness, stamina and the risk of injury and must reduce the effectiveness of a team to have to play twice a week. If the club only have a small squad it is worse.

[BillyStubbsTears]

You wanted it. Here it is. Although you probably won't like it....

Remember, to save a lot of very tiresome arguing again, this is NOT predicting who will finish where. It's predicting how many points each final position will end up with. Nothing magic about it, but with a couple of exceptions, it's generally getting the finishing points tallies right to within a couple or 3 points for 15 years now.

So, here's this evening's prediction.

1st place: 90
2nd 89
3rd 88
4th 85
5th 85
6th 82
7th 75
8th 70
9th 69
10th 65

Bit of a bugger if that's anywhere near. It means we'd need to win 10 of our final 13 to make the top 3.

88 points feels high for 3rd place, but there are a lot of teams with already slightly better ppg than us, and similar recent form. Personally, I think we are going to struggle to make the top 3, and this confirms it. We still don't and never have consistently looked like clearly one of the top three sides this season.

If the third place team gets 88 points I’ll be genuinely amazed.

I agree it does seem unfeasibly high.  I'm just putting out what the model says. It's not a very sophisticated one, but it has been very close much more often than it's been wildly off the mark.

[TonySoprano]

It'll be at least 3 points less than 88 like it always is.

[GazLaz]

You wanted it. Here it is. Although you probably won't like it....

Remember, to save a lot of very tiresome arguing again, this is NOT predicting who will finish where. It's predicting how many points each final position will end up with. Nothing magic about it, but with a couple of exceptions, it's generally getting the finishing points tallies right to within a couple or 3 points for 15 years now.

So, here's this evening's prediction.

1st place: 90
2nd 89
3rd 88
4th 85
5th 85
6th 82
7th 75
8th 70
9th 69
10th 65

Bit of a bugger if that's anywhere near. It means we'd need to win 10 of our final 13 to make the top 3.

88 points feels high for 3rd place, but there are a lot of teams with already slightly better ppg than us, and similar recent form. Personally, I think we are going to struggle to make the top 3, and this confirms it. We still don't and never have consistently looked like clearly one of the top three sides this season.

If the third place team gets 88 points I’ll be genuinely amazed.

I agree it does seem unfeasibly high.  I'm just putting out what the model says. It's not a very sophisticated one, but it has been very close much more often than it's been wildly off the mark.

Without consistency models are worth nothing!

[BillyStubbsTears]

I couldn't disagree more.

No model is ever either 100% accurate or consistent in its inaccuracy. The question is, are they useful?

In this case, this really simple model typically gets the final points total for each position right to within a couple of points about 80% of the time.

Some years it is spookily accurate, even when predicting what appears to be very unlikely scenarios.

Might well be wrong this year of course.

[Pancho Regan]

Has anyone checked other League 2 clubs to see if they’re discussing this topic

There's a lot of positivity amongst the Bradford fans, which is understandable, although they've had several false dawns over recent seasons of course. They reckon Alexander has finally got them playing as a team.
A couple of them on their forum are suggesting 82 points will be enough for auto promotion and they are confident they can achieve that.

[Dutch Uncle]

I am enjoying this excellent thread. As a person who has spent my life in Operations Research, and including much use of modelling and simulation, I would like to make a few comments – no negativity intended to anyone in any sense.

First with regards to simple data, points-per-game (ppg) is probably the best way of comparing clubs’ performances that have already happened in a season – i.e. I am not talking about any predictions or guides to future performance. That is why ppg was used to adjudicate final positions in the aborted COVID season. It is simple, understandable, and consistent across all clubs. To go any deeper into e.g. current form, ranking of opposition left to play brings more complications and potentially loss of fairness/consistency/transparency.

Then there is a wise old saying in the modelling and simulation community ‘all models are wrong, some are useful.’ Models, by definition, look at a subset of reality. Does that subset contain all the major driving factors?  Sometimes it can be good to look in more detail, but also sometimes the possible benefit of adding more factors is not worth the effort required and can confuse more than illuminate.

In this case, IMHO current form can add usefully to simple ppg for predictions. The ranking of opposition left is also possibly useful, but there is the question of whether to be ranked by table position or current form or some combination of both. Deeper analysts might wish to think about player availability (squad size, injuries and possible suspensions – how many players on a high number of yellows, quality of reserve players), past head-to-head data, do teams play well or badly against certain type teams/tactics, but IMHO that falls in the category of more work than benefit. Then there are always the things that can’t be modelled well – sheer luck with referee decisions, run of the ball, the effect of an early goal on teams, weather and playing conditions on the day. In this case, for us supporters, IMHO it is better to have a simple model and update and track it. BST’s model seems to do that, and IMHO is useful.     

[normal rules]

I find the ppg very telling .
Despite rovers lge position , all 4 clubs directly below have a better ppg. With the points being so close and with em all having games in hand it’s no surprise I guess.
Wombles, county and Bradford are all on 1.781ppg and PV are on 1.774. Whereas rovers are on 1.757. A win changes the score by around 0.05. So a win v Newport still only puts rovers on very similar ppg as the three teams directly below. A defeat for Bradford tonight will be a big boost. But they will
Then be on exact same ppg as rovers.
The gd has me concerned. It could make a big difference. It could be the diff between auto and playoffs .

[Colemans Left Hook]

Walsall, Wimbledon, Notts County and us have been the best four teams in L2 since day one. No reason to think that won’t be the top 4 come the end of the season.

i watched the first half of Morecambe v Walsall and Walsall were not the better side - I was keeping an eye on Hull at Sunderland and watched that second half as I lad backed Hull to stay up after those January "proper" signings .

It seems Morecambe had a player sent off after 52 minutes

https://www.bbc.co.uk/sport/football/live/c6262egzr4qt

"Two second-half goals gave League Two leaders Walsall victory over a struggling Morecambe side who ended the game with 10 men.

The contest was evenly poised until the 50th minute when Morecambe defender Rhys Williams was sent off for pulling back Jamie Jellis as he looked to run onto a long through ball.

To make matters worse for the home side, the Saddlers took the lead from the resulting free-kick as Taylor Allen curled the ball over the wall and into the goal off the foot of the right-hand post.

Despite being a man down, Morecambe looked to hit back with Marcus Dackers missing a golden chance to equalise when he headed a pinpoint Adam Lewis cross over from close range.

He was made to pay for the miss when Jellis scored a second three minutes from the end when he fired past goalkeeper Harry Burgoyne from the edge of the box.

Morecambe had the better chances of the first half but Lee Angol twice failed to find the target from close range."

now Notts County
-- yes they have Jones - "the old man" etc    OUT INJURED     
Amazingly Tranmere were much the better side for well over 70 minutes (they recently got a couple of Liverpool loanees) 

here i  the bbc version  which doesn't do Tranmere justice  Notts County for some reason were hopeless

https://www.bbc.co.uk/sport/football/live/cdryr7jgeg0t

"Late goals from Charlie Whittaker and Alassana Jatta fired promotion-chasing Notts County to a 2-1 win over Tranmere.

Nigel Adkins' side, who saw substitute Kristian Dennis grab a consolation effort, were the better side for long periods.

But they were made to pay for failing to exploit their superiority and remain in deep trouble at the other end of the table.

Despite entering the contest without a win in eight, Rovers carried a threat.

Omari Patrick tested Alex Bass after combining with Jake Garrett before Josh Davison fired over from the edge of the box.

Sam Finley and Chris Merrie were also presented with opportunities to hand Adkins' men the advantage.

Cameron Norman appeared to have done exactly that after meeting a cross by the influential Garrett only for Bass to again intervene.

If Norman could attribute that miss to the County goalkeeper's agility, he had only himself to blame on the stroke of half-time after heading wide from close range.

Whittaker and Jatta punished those misses with clinical 80th and 83rd-minute finishes from Mai Traore assists before Dennis' 88th-minute consolation."





as I keep saying in my world the dangers to us are behind us Bradford and PV

and don't underestimate any of these teams at the bottom Tranmere v Accrington should be "gud un" tonight bare knuckle fight ?

[DonnyBazR0ver]

I am enjoying this excellent thread. As a person who has spent my life in Operations Research, and including much use of modelling and simulation, I would like to make a few comments – no negativity intended to anyone in any sense.

First with regards to simple data, points-per-game (ppg) is probably the best way of comparing clubs’ performances that have already happened in a season – i.e. I am not talking about any predictions or guides to future performance. That is why ppg was used to adjudicate final positions in the aborted COVID season. It is simple, understandable, and consistent across all clubs. To go any deeper into e.g. current form, ranking of opposition left to play brings more complications and potentially loss of fairness/consistency/transparency.

Then there is a wise old saying in the modelling and simulation community ‘all models are wrong, some are useful.’ Models, by definition, look at a subset of reality. Does that subset contain all the major driving factors?  Sometimes it can be good to look in more detail, but also sometimes the possible benefit of adding more factors is not worth the effort required and can confuse more than illuminate.

In this case, IMHO current form can add usefully to simple ppg for predictions. The ranking of opposition left is also possibly useful, but there is the question of whether to be ranked by table position or current form or some combination of both. Deeper analysts might wish to think about player availability (squad size, injuries and possible suspensions – how many players on a high number of yellows, quality of reserve players), past head-to-head data, do teams play well or badly against certain type teams/tactics, but IMHO that falls in the category of more work than benefit. Then there are always the things that can’t be modelled well – sheer luck with referee decisions, run of the ball, the effect of an early goal on teams, weather and playing conditions on the day. In this case, for us supporters, IMHO it is better to have a simple model and update and track it. BST’s model seems to do that, and IMHO is useful.     

Yes, of course, these exercises are useful and in this case, a bit of fun too.

Whatever model BST is using, then I think the fun comes from the accuracy of the original forecast. If you keep adjusting the forecast, then obviously it's going to be more accurate the nearer we get to completion. But the true accuracy of the model should be measured against that original forecast to determine whether another factor needs to be introduced. For example the percentage of matches that go against form, which as you point out could be for any number of reasons, like injuries, suspensions etc.etc.

As said previously, in the last quarter of the season, there's the fight factor, with clubs somehow finding an extra 10% performance to fight off relegation, fight for the play offs and fight for Automatic  which means PPG becomes less of an indicator of where teams are going.

I think this is the reason we all think the 88 pts is too high for 3rd but that said, it's not a million miles away.

Maybe we could add a factor of 5% for 'Squeaky bum time'. Plus 5% on the points projections for those in the relagation zone. And minus 5% for those in the play off/automatic zones? Ha ha.

[Chris Black come back]

We’ve not been great for most of this season. Grimly stacking up points on a consistent basis. Are we in poor form or is this is playing to at or near our ability? Either way we need to start playing more consistently better than we have done for most of this season, to make a really strong case for the automatics. It’s been an odd season.

[drfchound]

I suppose it would be possible to get a reasonably accurate prediction of how many points team positions would finish on if you took a look at the final league table over the last ten years and took an average from that.
So how ever many points the top side ended with each year, added together and then divided by ten.
Then the same again with second, third and fourth places etc.
It might even give a more accurate guess at how the end of season points stacked up.

[drfchound]

I suppose it would be possible to get a reasonably accurate prediction of how many points team positions would finish on if you took a look at the final league table over the last ten years and took an average from that.
So how ever many points the top side ended with each year, added together and then divided by ten.
Then the same again with second, third and fourth places etc.
It might even give a more accurate guess at how the end of season points stacked up.

So, with my above thoughts in mind I did a quick look back over the last eleven seasons, then ignored the covid year and took an average of ten seasons and came up with the following for the top ten placings:
1st.  89 points
2nd. 85 points
3rd. 83 points
4th. 79 points
5th. 76 points
6th. 74 points
7th. 73 points
8th. 70 points
9th. 68 points
10th. 66 points

[Chris Black come back]

To get to 83 the currently third placed side would need 1.85 ppg.

[drfchound]

To get to 83 the currently third placed side would need 1.85 ppg.

….and we need 1.923 ppg.

[Dutch Uncle]

I suppose it would be possible to get a reasonably accurate prediction of how many points team positions would finish on if you took a look at the final league table over the last ten years and took an average from that.
So how ever many points the top side ended with each year, added together and then divided by ten.
Then the same again with second, third and fourth places etc.
It might even give a more accurate guess at how the end of season points stacked up.

Have put up the last 43 years stats (since 3 points per win introduced) earlier at reply #136 in this admittedly long thread Hound 

[drfchound]

I suppose it would be possible to get a reasonably accurate prediction of how many points team positions would finish on if you took a look at the final league table over the last ten years and took an average from that.
So how ever many points the top side ended with each year, added together and then divided by ten.
Then the same again with second, third and fourth places etc.
It might even give a more accurate guess at how the end of season points stacked up.

Have put up the last 43 years stats (since 3 points per win introduced) earlier at reply #136 in this admittedly long thread Hound 

Ah, sorry Dutch, I hadn’t noticed that.
Having seen it now it isn’t too far away from what I came up with is it.
I rounded mine up or down on the decimal points whereas you kept yours (with the decimal points included).
No more than a point difference in all places.

[Bessie Red]

To get to 83 the currently third placed side would need 1.85 ppg.

….and we need 1.923 ppg.

And if Wimbledon draw their game in hand they also would require 1.923ppg. If they lose their game in hand they would need 2ppg. If my Auntie had b*llocks.....

[Bessie Red]

I find the ppg very telling .
Despite rovers lge position , all 4 clubs directly below have a better ppg. With the points being so close and with em all having games in hand it’s no surprise I guess.
Wombles, county and Bradford are all on 1.781ppg and PV are on 1.774. Whereas rovers are on 1.757. A win changes the score by around 0.05. So a win v Newport still only puts rovers on very similar ppg as the three teams directly below. A defeat for Bradford tonight will be a big boost. But they will
Then be on exact same ppg as rovers.
The gd has me concerned. It could make a big difference. It could be the diff between auto and playoffs .

If Bradford lose tonight they will be on a lower ppg than us at 1.727ppg. If they draw they will be on the same as us.

[drfchound]

To get to 83 the currently third placed side would need 1.85 ppg.

….and we need 1.923 ppg.

And if Wimbledon draw their game in hand they also would require 1.923ppg. If they lose their game in hand they would need 2ppg. If my Auntie had b*llocks.....

Agreed Bessie, and that is why I spoke about watching how things evolve as matches go by.

[BillyStubbsTears]

No stopping Bradford. That 25 points from the past 10 games since Cook got injured.

[dickos1]

Whereas Walsall have only got 15 points from last 10 games, and we have got 21.

[BillyStubbsTears]

Whereas Walsall have only got 15 points from last 10 games, and we have got 21.

True. But context. They'd got 52 points from the first 23 games and we'd only got 37.

[GazLaz]

No stopping Bradford. That 25 points from the past 10 games since Cook got injured.

Said for years he potentially limits them as a team. Really good striker for the level but possible he’s so much of a focal point that he limits them in other areas. Good case study for individual players being great but a net negative for the team. Ronaldo in his last season at Juve for example.

[adamtherover]

Either way we need to start playing more consistently better than we have done for most of this season,

Is winning 6 from 8 not a sign that we have seen an upturn?

[adamtherover]

No stopping Bradford. That 25 points from the past 10 games since Cook got injured.

spurs always had better results when kane was injured as well..

[BillyStubbsTears]

Either way we need to start playing more consistently better than we have done for most of this season,

Is winning 6 from 8 not a sign that we have seen an upturn?

It's an upturn in results certainly. But that was generally a set of matches against really weak sides. We've got a far harder run in that that. If we can win 3/4s of the games we have left, I'll ll be chuffed to bits.

[ncRover]

How realistic is that we can achieve a higher goal difference than 2 of Walsall, Notts, Bradford and Wimbledon?
We need an extra point really.

[dickos1]

Whereas Walsall have only got 15 points from last 10 games, and we have got 21.

True. But context. They'd got 52 points from the first 23 games and we'd only got 37.

Of course which is why they’re so far clear, but since their striker went back to stoke their form has been midtable. Let’s hope it stays like that

[BillyStubbsTears]

Whereas Walsall have only got 15 points from last 10 games, and we have got 21.

True. But context. They'd got 52 points from the first 23 games and we'd only got 37.

Of course which is why they’re so far clear, but since their striker went back to stoke their form has been midtable. Let’s hope it stays like that

If they keep on picking up 1.5 ppg, Walsall will end up with 87 points.

Given goal difference, to top that we'd need 2.3 ppg from now to the end of the season. That'd mean we'd have picked up 51 points in the 2nd half of the season
Might change, but I've seen nothing this season to suggest we are remotely that good.