Premier League Table 2019/20 (as predicted using big data)

[adexkola]

The lack of transparency in their method is disappointing. Still, the inner workings of models like this tend to include IP valued by companies as being worth million of pounds. Being deliberately obtuse and protective of the actual formulas and metrics that produce outcomes like this mirrors all of my experiences with private sector 'big data's companies thus far.

It's a shame it didn't come from a university or some PHD project as I quite enjoy drilling down into the nuts and bolts of how they arrived at their outcome.

Nevertheless, their final table is pretty close to what I would predict if I was basing it on the squads as of today - city and Liverpool as the top two, Spuds third with United, Chelsea and Arsenal in some order rounding out the top 6.

Agree with you on all points.

For sure, proprietary methods need to be protected. However there is a middle ground where you can explain general techniques used, and demonstrate valuable insight, without giving away the secret sauce. The table shown is a general jerk off, complete with the buzz words "big data", "predictive analytics" thrown in. The only way this has value is if it is accurate with regards to the values and positions predicted, and I'll bet that it will be off considerably. Then what is the point?

Probably should have put some sort of stats abuse trigger warning in there when I shared the link. Apologies for the oversight.

no worries.

 

[Raven96__]

This is of zero importance.

Football is played out there on the pitch, 90 minutes, I can see united finishing higher than that if they sign maguire and a midfielder.

 

[Adcuth]

With the added 66 goal difference, just to rub it in.

That's points mate

 

[RedDevilRoshi]

Load of rubbish.

It’s virtually an identical table to last season

 

[sebsheep]

That's points mate

Indeed. I should've known I'd made a mistake, that number was far too big for our GD.

 

[Bullhitter]

Ignoring the points totals and goal columns, most positions look plausible to me apart from I think it's underrating Burnley who I would say will be comfortably mid table.

 

[ColvaleGoa]

Moral victory for us!

 

[Foxbatt]

This is of zero importance.

Football is played out there on the pitch, 90 minutes, I can see united finishing higher than that if they sign maguire and a midfielder.

True and no one ever thought Leicester was going to win the PL. Football is a round object and anything can happen on a football pitch. No one ever thought Greece was going to win the Euros too.

 

[bosnian_red]

Just read some of their actual script... it has Sheffield United beating Man United 5-0 in gameweek 13 What the feck kind of predictive model is this. I'm all for using stats and trends and models to run simulations like this, but come on! This looks like people just picked random numbers out of a hat. No model would ever have that result happen.

 

[D4X73r]

As long as 'pool don't win the league..

 

[Suedesi]

Guess what? United are finishing 6th again. Yay!

Link

Sounds about right.

 

[GhastlyHun]

It seems this is the first time they have done a prediction like this, at least I can't find one even for last season.
Well then. I hope to see this again at the end of the season and compare their 'big data' thing with reality.

 

[Tom Cato]

Cant complain really. I fully expect us to have similar season to last. We havent improved with the exception of the RB

Yes we have. Have you not seen Mason Greenwood play in the offseason? The kid is ready.

 

[Pagh Wraith]

True and no one ever thought Leicester was going to win the PL. Football is a round object and anything can happen on a football pitch. No one ever thought Greece was going to win the Euros too.

This is not this works. Of course people "predicted" Leicester and Greece to win, i.e. assigned them a non-zero probability value.

 

[Adisa]

Not much wrong.

 

[Chipper]

Just read some of their actual script... it has Sheffield United beating Man United 5-0 in gameweek 13 What the feck kind of predictive model is this. I'm all for using stats and trends and models to run simulations like this, but come on! This looks like people just picked random numbers out of a hat. No model would ever have that result happen.

Long post alert, just read the first 4 or 5 lines if you don't want to be bored.

That's a bit of strange one!

To be a little fair to them the model could not predict what it thought was the most likely outcome for every match and run with that if the aim was to predict a reasonable looking final league table. It would have to throw in some surprise results along the way.

Now 5-0 seems like a bit too much of a surprise result mind you.

If it didn't put in surprises you'd end up with tables that look like the one some have been compiling from Mark Lawrenson's predictions down the years, and then taking the piss out of him somewhat wrongly in my opinion. His job is to put what he thinks the most likely score will be for each game and the big criticism with him is that the table produced from those predictions would often have the top clubs lose less often than they would in the real world, and get more points than they really would.

Thing is, on a match by match basis, you're not going to predict the top club, or top 2-3 clubs to lose very often at all if your aim is to put down the most likely scoreline for each game. When Fergie was here, in any individual game, when would we be second-favourites to win a game? Maybe marginally away at Arsenal or Chelsea, but still you might well predict a draw and end up with us undefeated if you were compiling a table based off predicting our scores back then. Lawro tables always looked like that, with many people failing to grasp that kind of concept. That said, City have probably actually come close to racking up the unrealistic points tally of tables of his predictions would produce in the last couple of season.

One thing Lawro always has had/has is Liverpool bias.That's fair critique of him.

That was a bit of a divergence. Anyway, the overriding point I'm trying to make is that to try to model a table properly you'd have to throw in shock results rather than just the most likely outcome for each match because shock result happen in the real world so I suppose that's what they're trying to do here. 5-0 to Sheff Utd does seem a bit of a huge shock though, a shock of shock results even when trying to allow for shocks! Bet if you asked the same people to produce the most likely scoreline just for that match, outside of the context of trying a model a full season they'd have us winning or a draw.

 

[duffer]

Tweet

— Twitter API (@user) date

 

[BusbyMalone]

Was just reading through our "Script" for this season, made by the boffins over at BT. Suffice to say, it doesn't make for great reading.

(GAMEWEEK 1-3)
Big questions loom over Old Trafford. Ole Gunnar Solskjaer appears to have all the answers again. Man Utd win their first three games by an aggregate score of 12-3. Luke Shaw is involved in a third of their goals. The form of his career. Everyone gets carried away as they end a gameweek top of the league for the first time in two years.


(GAMEWEEK 4-9)
The wheels are off. Man Utd suffer deflating draws domestically and in the Europa League. Arsenal beat them away in the Premier League for the first time since 2006. Liverpool pick up three points at Old Trafford for the first time under Jurgen Klopp.

Man Utd drop to eighth.

(GAMEWEEK 10-20)

Solskjaer remains optimistic. Even as Premier League new boys Sheffield Utd and Aston Villa take points off his side in consecutive games. In Europe Man Utd win their last two group games to finish top. These results give them confidence to keep the top four within sight throughout the festive period.

(GAMEWEEK 21-23)
Man Utd start the new year with another defeat to Arsenal. It is about to get so much worse. Liverpool demolish Man Utd at Anfield. The 20-time English Champions watch their biggest rivals go top of the league.


(GAMEWEEK 24-37)
The buoyant mood at Man Utd survives an historic fourth straight derby defeat to City. Their European progress has Old Trafford buzzing as only a good cup run can. Man Utd dominate North London. They get revenge on Arsenal in Europe and beat Tottenham in the league. Solskjaer uses this momentum to guide his side all the way to the Europa League final.

(GAMEWEEK 38 + UFL FINAL)
Man Utd’s top-four hopes die with a wonder strike from Leicester’s Demarai Gray. They miss fourth by one point.

1. Man Utd concede 50+ goals for the second consecutive season and the second time in the PL era.

1. Man Utd are the first team to score 2000+ Premier League goals, finishing the season with a total of 2061

1. Paul Pogba is the first player to top both Man Utd’s goals and assists charts in back-to-back Premier League seasons.