The Probability Thread 2024 … or … Putting a number to doomed!

[Grecian-on-Tyne]

I trust Fujitsu haven't been allowed anywhere near the algorithms being used.

Think that's classified

 

[Grecian-on-Tyne]

@Grecian-on-Tyne Is there somewhere we can see the % likelihood of results on each game in your model?

Sorry, No. There is not a fixed percentage for each match.

Quick overview of how the model works:
Every team has a rating calculated from their home and away strength based of season & current form.
Then for each of the 10,000 simulations run each teams' rating is adjusted up or down, usually by a small adjustment, but sometimes much larger.
Then for each remaining fixture a home/draw/away percentage is applied based on both teams adjusted rating & result assigned based on percentages. So the % used will differ in each of the 10,000 runs.

The model has been refined after extensive testing against historical seasons (since 1998/9), to find the best mix of Season form / Current form & rating adjustment + some other considerations I've not touch on it this quick overview.

 

[Grecian-on-Tyne]

Relegation now at 12.8% ... what a couple of wins does

 

[Nigel E]

Just dragging this up to the top.

After yesterday, despite the disappointing result, our risk of relegation remained hovering around the 10% mark. Though, our chances of finishing in the top half reduced somewhat.

http://www.probs4.club/team/ExeterCity

Just to repeat, all those teams with games in hand are largely playing against each other, they can't all win, poor teams with a point a game after 30-odd matches are likely to carry on averaging a point a game. And so on.

Probably three more wins (or two wins and a clutch of draws) to stay up.

 

[Grecian-on-Tyne]

After last nights results + Reading's additional two points deduction, City's relegation percentage has now dipped to below 10% (9.6%)


https://www.probs4.club/lge/lge1#LOOT-4

With Charlton's there is a whole host of teams that could occupy that 4th place.

 

[SaintJames]

After last nights results + Reading's additional two points deduction, City's relegation percentage has now dipped to below 10% (9.6%)

View attachment 14906
https://www.probs4.club/lge/lge1#LOOT-4

Funny how it doesn't feel like we have a less than 10 per cent chance of relegation

 

[denzel]

So our remaining games are against the teams in the following positions:

3,5,6,7,8,11,17,18,19,21,22

An even spread, but sadly apart from maybe Northampton they are all likely to be playing for something!

 

[Grecian-on-Tyne]

Funny how it doesn't feel like we have a less than 10 per cent chance of relegation

Can think of three reasons ...
1. My model is crap
2. You're a pessimist.
3. Historically our current position 40pts from 35 has seen a bottom four finish around 22% of the time. It just the cluster of teams below us makes it more likely the points needed to survey is a few less than average. https://www.probs4.club/team/ExeterCity#TPaP


the observation in post #4 is still true with the points for relegation/survival has hardly changed

 

[SaintJames]

Can think of three reasons ...
1. My model is crap
2. You're a pessimist.
3. Historically our current position 40pts from 35 has seen a bottom four finish around 22% of the time. It just the cluster of teams below us makes it more likely the points needed to survey is a few less than average. https://www.probs4.club/team/ExeterCity#TPaP
View attachment 14908

the observation in post #4 is still true with the points for relegation/survival has hardly changed

View attachment 14909

I thought we were all pessimists on here ........

 

[Nigel E]

It just the cluster of teams below us makes it more likely the points needed to survey is a few less than average.

Think this is the point. For it to require more than 49-ish points to stay up, you'd be asking about 7 or 8 sides to outperform their season so far. That could happen, but isn't very likely. So I suppose the probability reflects the likelihood of us getting fewer than about 8 points from 11 matches.

The problem of course is that our recent history shows that we have a higher capability than others to go on an absolutely stinking run of results. I'm not sure how easy it is to model that.