Todays Championship Games

Laughing

Well-known member
I ran my algorithm over todays championship games, such as they are. I thought folks might be interested in what it had to say about the Stoke game, it's interesting to use in the race for the top 6.

I use 2 main algorithms to decide on the win, loose, draw and then sanity check against a third deepnet that I didn't write. The first is the most accurate and it predicts a home win for Stoke with a fairly small PHI correlation of 0.43. The algorithms accuracy is about 50%, which is to say guesswork.

The second algorithm has predicted a draw with a correllation of 0.32, again no better than pure chance. The monitor algorithm I use is even less sure of a Stoke win.

Stoke just don't score enough goals and have been getting worse lately. In the last 8 games, all they managed was 0.875 goals perr game. Preston are at 1.25 goals per game over the last 8. There will not be many goals in this and don't be surprised if its a draw or a Preston win.

Derby are predicted to win today with a fairly low correlation, but not so interested in that one.
 

Attachments

  • Screenshot 2022-01-03 10.41.23 AM.png
    Screenshot 2022-01-03 10.41.23 AM.png
    23.1 KB · Views: 13
Lose - Become unable to find ( something or someone ). " I've lost my car keys"

Loose - Not firmly or tightly fixed in place; detached or able to be detached.
 
Lose - Become unable to find ( something or someone ). " I've lost my car keys"

Loose - Not firmly or tightly fixed in place; detached or able to be detached.
Thank you Angel, I would edit it to reflect the excellent advice, but it is also littered with other typos, use instead of us, for example, or the stuck rrrrr key which repeats sometimes.
 
Laughing

Do you account for the strength of the opposition - say if Stoke played Fulham and scored is it treated the same as playing Peterboro.

Many predictive methods look at past data to predict the future but one weakness is it misses key future changes say signing a key player or key player recovering from long term injury.
 
Laughing

Do you account for the strength of the opposition - say if Stoke played Fulham and scored is it treated the same as playing Peterboro.

Many predictive methods look at past data to predict the future but one weakness is it misses key future changes say signing a key player or key player recovering from long term injury.
It uses this seasons statistics plus the last 8 games. The last 8 games is a blunt tool to capture improving teams or teams on a bad run. It does know the difference in league position also.

What a deepnet does, and this will dissapoint people who don't understand machine learning, is take a set of data points, points per game, league position, average goals for and against per game, home, away etc and applies weights to them to get as close to the correct outcome as it can 1,2 our 3. During training it goes over the result set thousands of times, adjusting the weightings of each of the data points until it starts to get the right outcome.

It then saves those weightings and applies them to a data set its never seen before the Stoke, pne game.

The algorithm doesn't understand any of the data and I don't tell it. It doesn't even know if its looking for cancer markers or a football result.

It works because it applies metrics to the data points and averages out how significant that data point is when taken together with the other data points.

It is very good when it's phi is above about 0.55. That is how strongly the data points correlate to the training data. A confidence score of you like.
 
Forest to get back to winning ways
Bournemouth to win 3 in a row and go further ahead in the table
Derby to continue their astonishing survival act and further cement Rooney as possible manager of the season
Swansea V Fulham and Stoke v Preston to fight out draws.
 
Forest to get back to winning ways
Bournemouth to win 3 in a row and go further ahead in the table
Derby to continue their astonishing survival act and further cement Rooney as possible manager of the season
Swansea V Fulham and Stoke v Preston to fight out draws.
Or what Abel says. There are only 2 championship games today mind Abel.

Fulham have dropped away but that's because they aren't playing. Win their 2 games in hand and they are clear top again.

Bournemouth are 9th in the 6 game form table. I can't see them winning 3 on the bounce.
 
The fixture layout over NY has been strange hasn’t it? Why so many games today and not Saturday? Teams getting an extra couple days compared to others isn’t really fair.

Fulham playing today after having played the Wednesday last week (obviously postponed but would have if it went ahead), meanwhile Boro play the same night but expected to play again Saturday lunch time.
 
Or what Abel says. There are only 2 championship games today mind Abel.

Fulham have dropped away but that's because they aren't playing. Win their 2 games in hand and they are clear top again.

Bournemouth are 9th in the 6 game form table. I can't see them winning 3 on the bounce.
Home to Peterborough would have made it 3 in a row but Laughing. Postponement has stopped them.
 
Back
Top