The Probability Thread 2022

[Grecian-on-Tyne]

So, we have had our first game of 2022 and the season is somewhere around the halfway mark. I thought now would be a good time to start the Probability Thread for 2022.

Ok, it might just be a ploy to share the links to my website and update you on what’s new for this season.

http://www.probs4.club/p4cLge/lge2

http://www.probs4.club/p4cTeam/ExeterCity

The main changes for 2021/22 is under the bonnet. The calculations used have been extensively revised, I’ve ran numerous variations of the ‘model’ against historical data (20 seasons x4 leagues, from 1999-2019) collating the predicted vs actual to find a realistic spread of possible outcomes. The biggest change has been to the ‘random factor’ intended to give a realist spread to the possible outcomes, to achieve this I’ve added what could be called a ‘chaos factor’ to help cater for the dramatic upturn or downturn in form that occurs more often than last season’s model would calculate.

Another addition this year are some charts comparing current season with historical data, hopefully some of these will give an alternative but compliantly view of our (and the other 91) position.

This thread is starting a lot earlier than previous years, two reasons:

1: Because I am confident the calculations and resulting percentage outcomes are about as accurate you can get, without using off-field & other subjective factors (eg. Playing budget, key player injuries, new manager etc.)

2: If our ‘downturn’ continues, even for those who do take a interest in this thread will soon lose interest or if we do start winning before it’s too late you will be able to track our upward progress.

(link to last years thread https://www.exeweb.com/forums/threads/the-probability-thread-2021-…-now-with-website-probs4-club.62808/

 

[Grecian-on-Tyne]

An example showing my calculations as a percentage spread for each position from 0.0% for 24th (0.0% is rounded that will be between 1 to 4 occurrences in 10,000) to 0.6% for 1st, maxing out above 8% for 7,8&9th.
Below that is the historical distribution of the finishing position for all 24 team completed EFL seasons since 3pts for win (1981), 8th most common finish and top 3 only 6 times out of 115.

 

[tonykellowfan]

Great work as always, many thanks.

 

[Grecian-on-Tyne]

A quick update, showing what the charts in #2 now look like.

 

[ORy]

47% probability for automatic promotion, great

 

[Trapdoor]

Approximately only 12% chance of finishing outside of the playoffs. Love to see it.

Average points needed for automatic promotion: 78
Our current median predicted value: 77

 

[grecian-near-hell]

20 points in eight games is great going, 2.5 points per game, if we can keep this up or even drop to 1.75 points average per game would take us to 81 points which would see us in with a good chance of Autos

 

[Grecian-on-Tyne]

A quick update before tonight's game

from here
Chance of Autos just below 50%


this suggests 77 points is close to the 50/50 mark for an auto place & 80 gives greater than 90%.

Come tomorrow these percentages will have changed & again come Sunday.

 

[Grecian-on-Tyne]

I've also created a couple of new pages history & outcomes using a different charting package (plotly).

Still more work to improve these pages & might use plotly for the whole site (for next season).

Perhaps the more interesting chart is here it's showing the percentage position outcomes for each points tally. Using all 24 team EFL completed season since 1981 (start of 3pts for a win) 107 seasons in total, Coloured for league two (3 autos, 4 playoff, 2 relegated).
So top 79 has given to top three finish 50% of the time where 80 & 81 percentage is slightly less 45% & 43%. Where 77 points which might be about 50/50 this season historically has given a top 3 7% (2 from 29)

 

[Grecian-on-Tyne]

Automatic now >50%

https://www.probs4.club/p4cTeam/ExeterCity#TO