Supercomputer Predictions

[harryCLSsafc]

No surprises.

As it stands we will go down, but the world moves on and who knows how our new team will do. I don't expect top 10 but I do expect us to be competitive with at least the bottom 6. But we will have an early indication of how things may well pan out after our first few fixtures, which includes teams we would certainly have to beat at home and away eg Burnley) to give ourselves a chance of staying up.

first rain check might be say the end of October.

 

[Balboa]

Last season it even had us more likely to finish 24th than 23rd.

Logon or register to see this image

Hope it’s had a new hard drive fitted…

 

[chriswallace85]

Can anyone find their prediction from this time last year? I can only find one with about 5 games to go.

 

[Comeback_Kenwyne]

I care about this as much as I care about that octopus who used to predict the outcomes of World Cup games.

Harsh on Paul imo.

 

[PhilY]

Harsh on Paul imo.

I make no apologies. I didn’t like the sqiggley sucker.

 

[Hardyman's Yugo]

Programmed by Scouse gobshite Mickey Quinn

 

[Madcap11]

Last season it even had us more likely to finish 24th than 23rd.

Logon or register to see this image

and look at middlesbrough, make me laff

 

[GipseT]

Can anyone find their prediction from this time last year? I can only find one with about 5 games to go.

You must be logged on to see media items

 

[MoneyballII]

Footballs not played on computers, it's played on grass

Well 90% grass and 10% plastic these days. But aye.

 

[Lucky1985]

The Opta super computer doesn’t fancy us much this year.
Apparently in 10,000 simulations of next season, we are the only team not to win the league and in 40% of them, we finish bottom.

You must be logged on to see media items

I don’t believe any computer that thinks Leeds, Burnley, West Ham, Brentford, or the Mags could win the premier league this season.

 

[Frijj]

You must be logged on to see media items

So last year it was out on average (mean) by 2.8 places, getting 8/20 spots correct, including the bottom 3.

It’s not great at the middle of the table but nailed 3 of the top 5 and 4 of the bottom 5.

I don’t believe any computer that thinks Leeds, Burnley, West Ham, Brentford, or the Mags could win the premier league this season.

No supercomputer would have “predicted” Leicester would either, but how these models work is they do thousands upon thousands of simulated seasons.

Leicester were 5,000 to 1, meaning if a supercomputer ran a simulated PL 10,000 times, Leicester would have won it twice.

It just so happens that we experienced a 5,000 to 1 event.

 

[Zweiguys]

I care about this as much as I care about that octopus who used to predict the outcomes of World Cup games.

We have a groundhog that can predict the weather.

 

[Lucky1985]

So last year it was out on average (mean) by 2.8 places, getting 8/20 spots correct, including the bottom 3.

It’s not great at the middle of the table but nailed 3 of the top 5 and 4 of the bottom 5.


No supercomputer would have “predicted” Leicester would either, but how these models work is they do thousands upon thousands of simulated seasons.

Leicester were 5,000 to 1, meaning if a supercomputer ran a simulated PL 10,000 times, Leicester would have won it twice.

It just so happens that we experienced a 5,000 to 1 event.

You’re conflating stats and odds.

 

[fyl2u]

Last season it even had us more likely to finish 24th than 23rd.

Logon or register to see this image

Gerrin, we're gonna win the league!

 

[Frijj]

You’re conflating stats and odds.

Those odds are derived from probabilities though.

It’s not impossible for any team to win the league next season and so if you simulate the season enough times, you’ll get weird results. If they’d stopped at 100,000 simulations we might have had a single first placed finish.

As it stands it gives us a 33% chance of staying up. That’s decent.

 

[Lucky1985]

Those odds are derived from probabilities though.

They are derived from what the bookmaker thinks will goad punters into giving them their money.

 

[PhilY]

They are derived from what the bookmaker thinks will goad punters into giving them their money.

And where - and how much - money has already been placed.

 

[Frijj]

They are derived from what the bookmaker thinks will goad punters into giving them their money.

In part, yes.

The point still remains that if you run enough simulations you’ll get weird results, so not trusting a model because it predicted Brentford winning the league doesn’t make sense.

 

[PJH]

You must be logged on to see media items

Big difference in spread between last year an this year. Puts only Liverpool and Arsenal at 100% safe from relegation, and only us as 100% not getting top 4. 17 teams managed to win a simulated league, compared to 4 the year before.

Lowest highest team position certainty last season (as in looking at teams finishing position percentages, which teams individual highest percentage is the lowest of all 20 teams) was West Ham with a 9.5% of finishing 10th. This season it's both Bournmouth and Man Utd with 7.4%. For Bournmouth it's finishing 9th, for Man Utd it's for finishing 13th. From what I see 9 teams in total have a lower highest percentage for this season than West Ham's 9.5%.

Despite out dire supercomputer performance, we still survive in 33.6% of the simulations!

 

[Frijj]

And where - and how much - money has already been placed.

See above regarding why the “supercomputer” has teams like Brentford winning the league in some simulations.