Beat the Mathematical Model

[Darren]

In my alleged obsessive desire to denigrate all alleged mathematical models, I want to see if such a model does, in fact, have a better predictive record than models with no such claims or top of the head predictions. So....

I have taken this coming weekend's fixtures and predicted them on the basis of my own 2 minute completely non-thought out non-mathematical model which is as follows:

Where the home team is placed higher in the table than the away team = home win
Where the home team is 7 places or less lower in the table than the away team = draw
Where the home team is 8 places or more lower in the table than the away team = away win

Adjusting for where Palace would have been but for the 10 point deduction gives me the following prediction:

Blackpool v Ipswich HW
Bristol C v Doncaster HW
Cardiff v Middlesbrough HW
Palace v SUFC D
Derby v Watford D
Newcastle v Barnsley HW
Forest v Swansea HW
Peterborough v Coventry AW
Plymouth v Preston AW
QPR v WBA AW
Sheff Wed v Leicester AW

If Ollesandro would kindly ask his mate to give his predictions and if anyone else who is interested would like to post theirs we can see if the model does have greater predictive power than one would expect.

 

[AlienWoodward]

In my alleged obsessive desire to denigrate all alleged mathematical models, I want to see if such a model does, in fact, have a better predictive record than models with no such claims or top of the head predictions. So....

I have taken this coming weekend's fixtures and predicted them on the basis of my own 2 minute completely non-thought out non-mathematical model which is as follows:

Where the home team is placed higher in the table than the away team = home win
Where the home team is 7 places or less lower in the table than the away team = draw
Where the home team is 8 places or more lower in the table than the away team = away win

Adjusting for where Palace would have been but for the 10 point deduction gives me the following prediction:

Blackpool v Ipswich HW
Bristol C v Doncaster HW
Cardiff v Middlesbrough HW
Palace v SUFC D
Derby v Watford D
Newcastle v Barnsley HW
Forest v Swansea HW
Peterborough v Coventry AW
Plymouth v Preston AW
QPR v WBA AW
Sheff Wed v Leicester AW

If Ollesandro would kindly ask his mate to give his predictions and if anyone else who is interested would like to post theirs we can see if the model does have greater predictive power than one would expect.

To put that into laymans terms,Darren has predicted some results for this weekend.
If anyone would like to do the same please feel free to join in

 

[derrywan]

as a randomly generated control sample, here's predictions generated by tossing a coin (draw = landed on the pile of files to my left rather than the carpet)

Blackpool v Ipswich HW
Bristol C v Doncaster HW
Cardiff v Middlesbrough AW
Palace v SUFC D
Derby v Watford AW
Newcastle v Barnsley D
Forest v Swansea AW
Peterborough v Coventry AW
Plymouth v Preston AW
QPR v WBA AW
Sheff Wed v Leicester HW

(Are you coming to Palace on Sat Olly?)

 

[LouTheBlade]

In my alleged obsessive desire to denigrate all alleged mathematical models, I want to see if such a model does, in fact, have a better predictive record than models with no such claims or top of the head predictions. So....

I have taken this coming weekend's fixtures and predicted them on the basis of my own 2 minute completely non-thought out non-mathematical model which is as follows:

Where the home team is placed higher in the table than the away team = home win
Where the home team is 7 places or less lower in the table than the away team = draw
Where the home team is 8 places or more lower in the table than the away team = away win

Adjusting for where Palace would have been but for the 10 point deduction gives me the following prediction:

Blackpool v Ipswich HW
Bristol C v Doncaster HW
Cardiff v Middlesbrough HW
Palace v SUFC D
Derby v Watford D
Newcastle v Barnsley HW
Forest v Swansea HW
Peterborough v Coventry AW
Plymouth v Preston AW
QPR v WBA AW
Sheff Wed v Leicester AW

If Ollesandro would kindly ask his mate to give his predictions and if anyone else who is interested would like to post theirs we can see if the model does have greater predictive power than one would expect.

Interesting model, although if these rules applied in all situations the league table would look ridiculous. It also doesnt take into account current form. I would be happy with a few of those results though

 

[Linz]

Go on then, this is my non-mathematical look at last Saturday's scores and guess model:

Blackpool v Ipswich D
Bristol C v Doncaster D
Cardiff v Middlesbrough AW
Palace v SUFC AW
Derby v Watford HW
Newcastle v Barnsley HW
Forest v Swansea HW
Peterborough v Coventry AW
Plymouth v Preston D
QPR v WBA AW
Sheff Wed v Leicester AW

 

[Highbury_Blade]

As another control group, I will predict the outcomes of this weeks matches using the Bristol stool scale. I'm doing a variety of cuisines this week, so the results should differ markedly.

To add an element of randomness I'll draw the fixture out of the hat, as by the end of the week, my insides are likely to be mush, thus affecting the results.

Healthy stools 3-5 on the scale HW.
Post beer dump 6-7 AW
lacking in fibre 1-2 D

first result forest v swansea D (2 on the scale if you're interested)

I'll keep you all posted as and when I take my twice-daily constitutional.

 

[Darren]

Interesting model, although if these rules applied in all situations the league table would look ridiculous. It also doesnt take into account current form. I would be happy with a few of those results though

Ah but what you have to understand is that my model is, in fact, 100% accurate in estimating which teams are better than other teams and that where the results differ from my predictions it is because of randomness and exogenous factors.

(sorry Olly, couldn't resist )

 

[Ollessendro]

In my alleged obsessive desire to denigrate all alleged mathematical models, I want to see if such a model does, in fact, have a better predictive record than models with no such claims or top of the head predictions. So....

I have taken this coming weekend's fixtures and predicted them on the basis of my own 2 minute completely non-thought out non-mathematical model which is as follows:

Where the home team is placed higher in the table than the away team = home win
Where the home team is 7 places or less lower in the table than the away team = draw
Where the home team is 8 places or more lower in the table than the away team = away win

Adjusting for where Palace would have been but for the 10 point deduction gives me the following prediction:

Blackpool v Ipswich HW
Bristol C v Doncaster HW
Cardiff v Middlesbrough HW
Palace v SUFC D
Derby v Watford D
Newcastle v Barnsley HW
Forest v Swansea HW
Peterborough v Coventry AW
Plymouth v Preston AW
QPR v WBA AW
Sheff Wed v Leicester AW

If Ollesandro would kindly ask his mate to give his predictions and if anyone else who is interested would like to post theirs we can see if the model does have greater predictive power than one would expect.

There you go steam-rollering into something without me actually agreeing!?

I've asked him for predictions, but for reasons out of my hands it may not be possible (see the other thread).

 

[PeterNdlovu081]

Blackpool v Ipswich AW
Bristol City v Doncaster HW
Cardiff v Middlesbrough D
Crystal Palace v Sheffield United AW
Derby v Watford HW
Newcastle v Barnsley HW
Nottingham Forest v Swansea AW
Peterborough v Coventry D
Plymouth v Preston HW
QPR v WBA AW
Sheff Wednesday v Leicester AW

 

[LouTheBlade]

My non-mathematical guess model for saturday;

Blackpool v Ipswich - AW
Bristol C v Doncaster - HW
Cardiff v Middlesbrough - AW
Palace v SUFC - D
Derby v Watford - HW
Newcastle v Barnsley - HW
Forest v Swansea - D
Peterborough v Coventry - HW
Plymouth v Preston - D
QPR v WBA - AW
Sheff Wed v Leicester - AW

 

[Paulus]

In my alleged obsessive desire to denigrate all alleged mathematical models, I want to see if such a model does, in fact, have a better predictive record than models with no such claims or top of the head predictions. So....

I have taken this coming weekend's fixtures and predicted them on the basis of my own 2 minute completely non-thought out non-mathematical model which is as follows:

Where the home team is placed higher in the table than the away team = home win
Where the home team is 7 places or less lower in the table than the away team = draw
Where the home team is 8 places or more lower in the table than the away team = away win

Adjusting for where Palace would have been but for the 10 point deduction gives me the following prediction:

Blackpool v Ipswich HW
Bristol C v Doncaster HW
Cardiff v Middlesbrough HW
Palace v SUFC D
Derby v Watford D
Newcastle v Barnsley HW
Forest v Swansea HW
Peterborough v Coventry AW
Plymouth v Preston AW
QPR v WBA AW
Sheff Wed v Leicester AW

If Ollesandro would kindly ask his mate to give his predictions and if anyone else who is interested would like to post theirs we can see if the model does have greater predictive power than one would expect.

OK I'll use your model but I'll base it on the current form table (last 8 games overall) rather than the full table you are using.

Blackpool v Ipswich D
Bristol C v Doncaster D
Cardiff v Middlesbrough D
Palace v SUFC D
Derby v Watford HW
Newcastle v Barnsley HW
Forest v Swansea D
Peterborough v Coventry AW
Plymouth v Preston D
QPR v WBA AW
Sheff Wed v Leicester AW

On first glance too many draws but we will see.

 

[mall]

There are 33 points at stake.
My prediction is based on a quick perusal of every match ever played anywhere, excluding other planets, that the majority of points will end up going to the home teams.

 

[Ollessendro]

Ollessendro's predictions:

Blackpool v Ipswich D
Bristol C v Doncaster H
Cardiff v Middlesbrough D
Palace v SUFC D
Derby v Watford HW
Newcastle v Barnsley HW
Forest v Swansea D
Peterborough v Coventry HW
Plymouth v Preston AW
QPR v WBA AW
Sheff Wed v Leicester AW

Please note that, although I based this on my friends model, this is not his predictions! I have looked at his ratings table, the league table (and particularly goal difference), current form and home and away form (dependent on circumstances). Ideally I'd like to take into consideration bookmakers odds (however, I can't as I'm at work) and factors on the day (such as team selection, formation, injuries) however I'm still fairly confident of this.

 

[Highbury_Blade]

Slightly earier than expected due to what I can only assume must have been a dodgy lasagne at lunch from the work canteen

Sheff Wed v Leicester AW (6 on the scale)

 

[tomtheblade]

My predictions, based on nothing but my little brain:

Blackpool v Ipswich D
Bristol C v Doncaster HW
Cardiff v Middlesbrough D
Palace v SUFC AW
Derby v Watford HW
Newcastle v Barnsley HW
Forest v Swansea D
Peterborough v CoventryAW
Plymouth v Preston HW
QPR v WBA AW
Sheff Wed v Leicester AW

 

[moDtheGod]

How long are we trialling this for? One week of fixtures proves nothing.

 

[Ollessendro]

How long are we trialling this for? One week of fixtures proves nothing.

Mod has a very good point of course, but I didn't want to spoil Darren's fun. He has a bee in his bonnet and set this up without even getting my opinion on it. He should know full well that if someone predicts better results than my friends model I will merley (and quite rightly) put it down to luck or randomness.

Luck, exogeniety and randomness mean that Tom, Dick or Harry could predict the fixtures better than my friends model this week. Still if we can do it for all the fixtures between now and the end of the season we should be able to see a difference.

For someone as clever as Darren is he should see that no matter what happens we will not agree and even this method will not prove anything. However he seems hell bent on proving me wrong. The fact is that I never argued the model can predict the league table. So even if my friend gave me his ratings and predicted a league table for the end of the season, randomness would assure that it was not fully accurate. The chances are that it would be more accurate than Darren's, Matt B's and HH predictions, but there is no guarantee.

I'm happy to go along with this little chirade if necessary. I've enjoyed the debate and I think that this is quite fun. Plus I am confident that, despite randomness, my friends model will predict results better than the majority, if not all, people on here over time. Then i'll be interested to see what Darren thinks. However I get the impression he is as stubborn as I am and he is unlikely to give any ground, even if it proves the model works well.

 

[raul]

How long are we trialling this for? One week of fixtures proves nothing.

A random guess would be as long as it takes for either Darren or Olly to wear the other down.....

 

[tomtheblade]

Olle, the key thing in your mate's model is the equation isn't it? (i admit i haven't read the whole of the other monster thread on this subject)

Anything else is just a formalised version of what every Tom, Dick and Harry does when gambling on football results. That is, an educated guess based on form, fitness of players, opponent, home or away etc. It's no surprise to me that someone who goes into a much larger amount of detail and research on his teams comes out with better results than the average Joe, but the amount of work and input required means relatively it's a similar payoff to someone who predicts based on a quick glance at the league table.

 

[wardlowwanderer]

Based on which team I like most of the pair. If I don't like either D.

Blackpool v Ipswich HW
Bristol C v Doncaster AW
Cardiff v Middlesbrough D
Palace v SUFC AW
Derby v Watford HW
Newcastle v Barnsley HW
Forest v Swansea D
Peterborough v Coventry D
Plymouth v Preston HW
QPR v WBA HW
Sheff Wed v Leicester D

this will most definitely be 100% correct according to my barometer.

 

[FlyingFjortoft]

This gives me an excellent chance to test my new model to predict football results, which after some recent success I am becoming fairly confident about.
I have delved into my sock draw 11 times, and pulled out a sock for each result. A white sock is a home win, a black sock is an away win and anything else (red sock, packet of crisps, condom) is a draw.
This gives me the following results which i can highly recommend getting on an accumulator:
Blackpool v Ipswich D
Bristol C v Doncaster HW
Cardiff v Middlesbrough D
Palace v SUFC AW
Derby v Watford HW
Newcastle v Barnsley AW
Forest v Swansea D
Peterborough v Coventry HW
Plymouth v Preston HW
QPR v WBA AW
Sheff Wed v Leicester AW

 

[Pinchy]

As another control group, I will predict the outcomes of this weeks matches using the Bristol stool scale. I'm doing a variety of cuisines this week, so the results should differ markedly.

To add an element of randomness I'll draw the fixture out of the hat, as by the end of the week, my insides are likely to be mush, thus affecting the results.

Healthy stools 3-5 on the scale HW.
Post beer dump 6-7 AW
lacking in fibre 1-2 D

first result forest v swansea D (2 on the scale if you're interested)

I'll keep you all posted as and when I take my twice-daily constitutional.

Your system is shit isn't it?

 

[Sirius]

Interesting thread olle. Horace Bachellor's infra draw method 208 Radio Lux.(Yonks ago) Will watch with interest how the predictions do

 

[Ollessendro]

This gives me an excellent chance to test my new model to predict football results, which after some recent success I am becoming fairly confident about.
I have delved into my sock draw 11 times, and pulled out a sock for each result. A white sock is a home win, a black sock is an away win and anything else (red sock, packet of crisps, condom) is a draw.
This gives me the following results which i can highly recommend getting on an accumulator:
Blackpool v Ipswich D
Bristol C v Doncaster HW
Cardiff v Middlesbrough D
Palace v SUFC AW
Derby v Watford HW
Newcastle v Barnsley AW
Forest v Swansea D
Peterborough v Coventry HW
Plymouth v Preston HW
QPR v WBA AW
Sheff Wed v Leicester AW

I'm going to come back with something serious to your sarcastic post. If you really believe that Barnsley will win away at newcastle and Posh will win at hyome against Coventry (this goes for all people) then you should back them, as you'll get good odds. The 3 tips my mate gave me (oin terms of the bookies mis pricing were those 2 and Swansea to win at Forrest). You can get 7/1 for the Dingles to win at the Toon. Peterborugh at 7/4 are a really good bet. Posh are not a bad side and Covenrty are shite. I fully expect them to take 3 points (in fact i've whacked £35 on it).

Olle, the key thing in your mate's model is the equation isn't it? (i admit i haven't read the whole of the other monster thread on this subject)

Anything else is just a formalised version of what every Tom, Dick and Harry does when gambling on football results. That is, an educated guess based on form, fitness of players, opponent, home or away etc. It's no surprise to me that someone who goes into a much larger amount of detail and research on his teams comes out with better results than the average Joe, but the amount of work and input required means relatively it's a similar payoff to someone who predicts based on a quick glance at the league table.

Yes matey the equation, or more to it the whole model section, is the importsant part. Bear in mid that this is an old paper and the models used by my friend and companies that use this have been updated. This is merley for illustrative purposes.

There's more to it than just being well informed and spening time investigating football though Tom. You need to have some statistical expertise, or at least be very good with probabilities. The model predicts the likelehood a team will win a game via the probaility they will score and concede a certain amount of goals aginats another team. However adjustments are then often made for exogenous factors (such as injuries, formation, tactics, weather etc). This is where the combination of footballing and probability expertise come in.

 

[Selly]

Your system is shit isn't it?

This has been crminally ovelooked :thumbup:

Now attempting to explain to the rest of the office what can be quite so funny at 9:40 on a wednesday morning

 

[Darren]

I am quite happy to keep doing this until the end of the season to get a broadly based result but I suspect everyone else might get bored.

My basic scepticism about all alleged mathematical models is that, as a matter of logic, they cannot possibly work in the long run. Firstly because, if they did work all astute gamblers would used them and the bookmakers would all be bankrupted (or bookies would stop taking bets on football). Secondly, I rather suspect that William Hills, Ladbrokes etc have enough dosh to employ people at least as clever as Olly's friend to work out their odds and ensure that they are not out-foxed by these alleged models.

 

[AlienWoodward]

I am quite happy to keep doing this until the end of the season to get a broadly based result but I suspect everyone else might get bored.

My basic scepticism about all alleged mathematical models is that, as a matter of logic, they cannot possibly work in the long run. Firstly because, if they did work all astute gamblers would used them and the bookmakers would all be bankrupted (or bookies would stop taking bets on football). Secondly, I rather suspect that William Hills, Ladbrokes etc have enough dosh to employ people at least as clever as Olly's friend to work out their odds and ensure that they are not out-foxed by these alleged models.

I'm sure a few years ago you would have been able to get a grant from the "loony left" to look into this more deeply.

 

[Darren]

I'm sure a few years ago you would have been able to get a grant from the "loony left" to look into this more deeply.

That's an extemely poor effort at provocation...

 

[raul]

Are you accusing Darren of a Militant Third World Lesbian past Alien?


More interestingly......RADIO LUXEMBOURG!!


Grumpy, Broomhill, Trigger - your Hour is nigh.

 

[AlienWoodward]

That's an extemely poor effort at provocation...

I stopped provoking you after the 50th attempt to win an argument with you