Football Musings

[andy28]

Everton, Villa match I am having issues with my model, It is telling me Everton should be 2.26 with recent home matches. Also I am thinking they will be more confident that Villa. How ever my racing side says Villa will win based on recent form. The last 5 away losses were to Newcastle, Forest, Chelsea, Liverpool, Spurs, which is a tough run, so I think they will be up for it as they will see this as a must win. So when I frank that form Villa come out on top

Another match today Arsenal, Spurs one model shows a lack of goals however another a goal fest. A North London Derby maybe action packed, however it all indicators I use say Arsenal are too short. I think these are sit and watch games for me

Bet delay is on 8 seconds, can sup a full can of lager before you get in and out the market

Coffee for me is 10.00am in morning here, I love these late kick off matches

[wearthefoxhat]

Everton, Villa match I am having issues with my model, It is telling me Everton should be 2.26 with recent home matches. Also I am thinking they will be more confident that Villa. How ever my racing side says Villa will win based on recent form. The last 5 away losses were to Newcastle, Forest, Chelsea, Liverpool, Spurs, which is a tough run, so I think they will be up for it as they will see this as a must win. So when I frank that form Villa come out on top

Another match today Arsenal, Spurs one model shows a lack of goals however another a goal fest. A North London Derby maybe action packed, however it all indicators I use say Arsenal are too short. I think these are sit and watch games for me

Some models do conflict that's for sure. It also depends on how many games, home/away you take into account.

Soccer Stats give the Last 4 and Last 8, Fishy give the Last 6. Then of course there's the whole season you can cover too.

The Everton v Ast Villa game on the last graphic posted (whole season) showed how bad a trend Everton has at home, however, it's not clear what impact the manager might have. The Ast Villa away trend was not bad. The Poisson had it leaning towards Everton, so I gave it a swerve.

The other Poisson predictions all stood up well, including the Leicester L2B trade.

It's all looking promising looking ahead.

[andy28]

I've been comparing actual goal data and expected goals (xG) from the last six games against the model I've been developing. So far, the results are promising.

For example, during the recent Arsenal vs. Spurs match, my main model predicted an xG of 3.84, while the six-game xG average was 1.56, and the actual goals from the last six games averaged 2.97. It's important for me to remember that xG predicts the expected goals in a game, not the actual score. For this match, the xG was 2.25, which isn't far off from the model's prediction.

I still need to work out the conversion rate from Xg to goals, but the last six games' average of 2.97 proved to be accurate.

For the most part they went well still plenty to work on Everton threw all my models out but I was confident Villa would win in a tight game

[wearthefoxhat]

I've been comparing actual goal data and expected goals (xG) from the last six games against the model I've been developing. So far, the results are promising.

For example, during the recent Arsenal vs. Spurs match, my main model predicted an xG of 3.84, while the six-game xG average was 1.56, and the actual goals from the last six games averaged 2.97. It's important for me to remember that xG predicts the expected goals in a game, not the actual score. For this match, the xG was 2.25, which isn't far off from the model's prediction.

I still need to work out the conversion rate from Xg to goals, but the last six games' average of 2.97 proved to be accurate.

For the most part they went well still plenty to work on Everton threw all my models out but I was confident Villa would win in a tight game

Well done.

You might find the sweet spot between actual goals and xG with your research. I'm averaging the actual score with the xG that I produce using the sofa-score metrics. (Normal xG probably okay too) That way it will normalise the score-line and offer a truer picture of the game. This should then lead to a more accurate model. (that's the cunning plan going forward)

With the Arsenal v Spurs game, there seemed more action than what the normal xG calculated, happy with the adjusted xG version. The Poisson prediction model had the game on the lower side (1.80 goals), so something for me think about.

Arsenal v Spurs.png

I'm in the process of reconstructing the Championship league at the moment. Nothing remarkable about it (average total goals 2.48), it's one I know fairly well, along with some of players in the teams, so reckon I could nick a few sovs before the end of the season.

Stats.png

[wearthefoxhat]

Tonights Burnley v Sunderland game is intriguing for different reasons.

Seems to be some value to Lay2Back Sunderland on Betfair, based on combining the data in a game "predicted" to be low scoring and therefore, more than likely, a draw.

Data.png

However, when comparing the each teams' home and away trends for the last 10 games, Burnley are less likely to score and not concede, and Sunderland should score but also concede.

Trend.png

Score-lines suggest 0-0, 1-1, 1-0, but if Burnley score early doors, then it might turn out to be one of those games they go on to score freely and win easily....clear as mud.

----------

Update.

FT 0-0 The normal xG was 0.69/2.86 - so looks though Burnley got away with one.

---------

[wearthefoxhat]

Gone thorugh the games in the Premier & Championship, one or two to get involved with today.

One game in the championship that caught my eye where there appears to be a conflict with the live market.

Portsmouth v Middlesbrough

Indicates a Home win, Total Goals 4.56, 15% Home Edge.
Trade - Lay2Back Middlesbrough @ 1.89
Score-line could be anything. (depends which Middlesbrough team turns up)
Long arse journey to the south coast for the away team.

Ports v Middls.png

----

Portsmouth have been beating mid-tables sides convincingly at Home recently, but their away form home is why they're languishing near the bottom.

Portsmouth H.png

----

Middlesbrough aren't really a mid-table team, but need more consistency to be called a top 6 side. Their away record at one point was strong, scoring for fun, this is taken into account by the Poisson Distribution model, but still leans towards Portsmouth.

Middlesbro A.png

------

Added

Result Portsmouth win 2-1 - Easy this game innit!

[wearthefoxhat]

Here's how it looks from a "Value" perspective, if I just back all the ones indicated as Value bets to level stakes on Betfair.

Championship.png

My own Poisson model, if taken at face value, was average at best. Football can be unpredictable at times, so it makes sense to use a value odds approach. I quite like how the sheet breaks down the other markets, but they could be just a sideshow distraction.

I ran the Premier one too, equally good on value. Arsenal stopped a clean sweep even though they were coasting 2-0 at one stage.

Premiership.png

[andy28]

It appears I have been working out predicted goals wrong

[wearthefoxhat]

It appears I have been working out predicted goals wrong

In what way have you been doing it wrong?

[andy28]

I have been averaging the home for and away concede and then average away for and home concede. Then add the two together. Then I would use Poisson Dist to get match odds but they weren't terribly accurate. The total goals was bang on in fact I am within 4 goals of the total for the EPL but when I looked at home & away separate their is a discrepancy, which in turn has thrown the match odds out.

[wearthefoxhat]

I have been averaging the home for and away concede and then average away for and home concede. Then add the two together. Then I would use Poisson Dist to get match odds but they weren't terribly accurate. The total goals was bang on in fact I am within 4 goals of the total for the EPL but when I looked at home & away separate their is a discrepancy, which in turn has thrown the match odds out.

At 7.12 in the video, Kie Millet goes through how to use the goal averages info (home & away) to set up the poisson predictive element.Play it at 0.75 speed cos he flies through it and pause it at each stage as you create the excel sheet.

https://www.youtube.com/watch?v=24e_Z4WHR48&t=545s

[andy28]

I have been averaging the home for and away concede and then average away for and home concede. Then add the two together. Then I would use Poisson Dist to get match odds but they weren't terribly accurate. The total goals was bang on in fact I am within 4 goals of the total for the EPL but when I looked at home & away separate their is a discrepancy, which in turn has thrown the match odds out.

At 7.12 in the video, Kie Millet goes through how to use the goal averages info (home & away) to set up the poisson predictive element.Play it at 0.75 speed cos he flies through it and pause it at each stage as you create the excel sheet.

https://www.youtube.com/watch?v=24e_Z4WHR48&t=545s

Thanks, I have watched that video and my spreadsheet is very similar to his now, I did watch that part several times but as I wasn't using his exact cells it confused the hell out of me so I just carried on with my own sheet and formula. Then I saw another video explain this better so asked Chat Gpt and it gave me the formula.

I am slowly getting there with my project for football and I must say I have learned a lot from your thread so keep up the good work

[wearthefoxhat]

I have been averaging the home for and away concede and then average away for and home concede. Then add the two together. Then I would use Poisson Dist to get match odds but they weren't terribly accurate. The total goals was bang on in fact I am within 4 goals of the total for the EPL but when I looked at home & away separate their is a discrepancy, which in turn has thrown the match odds out.

At 7.12 in the video, Kie Millet goes through how to use the goal averages info (home & away) to set up the poisson predictive element.Play it at 0.75 speed cos he flies through it and pause it at each stage as you create the excel sheet.

https://www.youtube.com/watch?v=24e_Z4WHR48&t=545s

Thanks, I have watched that video and my spreadsheet is very similar to his now, I did watch that part several times but as I wasn't using his exact cells it confused the hell out of me so I just carried on with my own sheet and formula. Then I saw another video explain this better so asked Chat Gpt and it gave me the formula.

I am slowly getting there with my project for football and I must say I have learned a lot from your thread so keep up the good work

That part needs careful editing, and the issue, if using it exactly the way Kie M describes, although not insurmountable, will be setting up each league when teams get promoted/relegated. Those four cells would need to be edited carefully.

I circumvented that process, with GPT's help, to calculate the home/away averages using seperate tables to the side, and use VLOOKUP, and then get the matrix with the Poisson calculation to look at resulting xG to populate all the other stuff including the odds.

All I need to do now is copy and paste (or input) the data I want it to review, ie: Last 4, 6, 8 or the whole seasons home/away, and then combine with Club Elo (if league is covered) or Pinnacle, to normalise the odds.

I'm finding the value approach works well, when comparing the normalised odds, with the Betfair market. Of course, the sheet gives plenty of other value opportunities, ie: BTTS, Correct Scores...etc but I tend to like the match odds. This approach takes a contrarian perspective as you have to sometimes back a team against the grain, but it's "value".

Last night it indicated Preston/QPR Draw/Away odds as value, with a total profit across all the games so far. I tested it out in league 1 and it indicated Away/Draw for Burton v Wigan at odds of 6.20 so that encouraging going forward.

WIth the "value" approach, you'd likely have to take all the value spots and not cherry pick the games.

----

I'm also looking at a way to take into account the strengths of the team to be played. This would again adjust the model further and give some more realism. Will be leaning on "Me Ol' Mucker" - GPT.

Swells.png

[andy28]

I set all my spreadsheets up the same, I use VBA and then just apply to a new worksheet, then I do all my workings to the side so all I have to do if I want to set up a new league is copy all my workings into that sheet and everything fits, I just need to change the name of the teams.

Trying to automate as much as I can to save time

[wearthefoxhat]

I set all my spreadsheets up the same, I use VBA and then just apply to a new worksheet, then I do all my workings to the side so all I have to do if I want to set up a new league is copy all my workings into that sheet and everything fits, I just need to change the name of the teams.

Trying to automate as much as I can to save time

Nice one.

VBA has so far eluded me, but, will approach it at some time. My own maths and excel knowledge has always been basic, and I'm no spring chicken, but, I'm closing the gap fast, and, who knows, StarFox might be born soon...(FFS Fox, what are you on?)