Bit embarrassing for OPTA

[rudebhoy]

They ran 50,000 simulations and came up with the following -

“How Many Points Will Be Needed to Finish in the Top 24?


The top eight positions are likely to be reserved for the cream of the crop. It would make sense for the best eight teams in the competition to take those positions.

The play-off therefore represents the best chance of making it through to the knockouts for most teams in the competition.

Our analysis found that 10 points will as good as guarantee a place in the top 24 places, with 99% of simulations seeing teams on 10 points make it into the play-off round.

Nine points is enough for a play-off spot in 69% of simulations, while eight points only saw 16% of teams make it into the top 24.

Put more simply, nine points – or just three wins from eight games – is very likely to be enough to make it into the play-off round.

Meanwhile, eight points – or two wins and two draws – is very unlikely to be enough.

So, there’s your minimum target: nine points.”


Zagreb were eliminated on 11 points.

 

[hambugerpete]

They weren't far off. Zagreb lost out on goal diff and I doubt those Sims had city being so crap.

 

[Tashyboy]

They weren't far off. Zagreb lost out on goal diff and I doubt those Sims had city being so crap.

We’re through

 

[Swango1980]

They weren't far off. Zagreb lost out on goal diff and I doubt those Sims had city being so crap.

That's the problem with Sims, based on completely unrealistic assumptions.

Maybe nobody predicted City being rubbish. But, amongst all the teams, always a reasonable chance at least one top side will be garbage, and at least one poor side will play brilliantly

 

[Neilds]

Just goes to show how rubbish stats are when trying to predict real life. Maybe those who are continually quoting stats in the WHS thread should take note

 

[rudebhoy]

They weren't far off. Zagreb lost out on goal diff and I doubt those Sims had city being so crap.

“Our analysis found that 10 points will as good as guarantee a place in the top 24 places, with 99% of simulations seeing teams on 10 points make it into the play-off round”

Stuttgart eliminated on 10, Zagreb out with 11. Pretty poor forecasting.

 

[Arthur Wedge]

Is it all not a bit of guess work based on previous performances and a lot of simulations

 

[rudebhoy]

Is it all not a bit of guess work based on previous performances and a lot of simulations

50,000 of them

 

[Arthur Wedge]

50,000 of them

And they end up a point out ?

 

[Swango1980]

And they end up a point out ?

Yeah, but you are missing out on the 99% probability bit.

Had they said 10 points gives you a reasonable or a 50/50 chance of getting through to the next stage, fair enough. But, if they start saying 99% gets you through with 10 points, then it does look a bit daft now. I wonder what their probability was for 11 points? 99.9999%?

 

[Arthur Wedge]

Yeah, but you are missing out on the 99% probability bit.

Had they said 10 points gives you a reasonable or a 50/50 chance of getting through to the next stage, fair enough. But, if they start saying 99% gets you through with 10 points, then it does look a bit daft now. I wonder what their probability was for 11 points? 99.9999%?

It’s just a simulation

And on a format that’s not been done before

It’s all pretty meaningless at the end of the day

 

[Swango1980]

It’s just a simulation

And on a format that’s not been done before

It’s all pretty meaningless at the end of the day

Well then I think you pretty much agree with the OP then. It is all meaningless.

Hence, it is a bit embarrassing for Opta, unless they are happy to be known as the company whose does work that is all a bit meaningless? I bet if that accusation was thrown at them at the start of the tournament, they'd have robustly defended their analysis. Why? Because it is based on 50,000 simulations. Clearly the 50,000 simulations are meaningless if the assumptions made are poor.

A nice overview of software use is:

Rubbish In = Rubbish Out

 

[Grizzly]

Any predictive element as regards sports is tainted by the human factor - and in this format, being played for the first time, there are 32 sets of those human factors. As others have said, the performance of teams like Manchester City, PSG and Juve was probably not what anyone expected, nor at the other end of the table would many have expected Liverpool to win seven straight games.

 

[Swango1980]

I assume the simulation results would have been the same for Europa League.

Looks like 2 teams have been eliminated with 10 points.

 

[sunshine]

50,000 of them

I don’t think anyone understands the concept of confidence intervals or Monte Carlo simulations.