Supercomputer vs human predicting Arsenal next season. Both lose.



By Tony Attwood

There is a screamingly funny piece in the Sussex Express which really does need to be contemplated.  The headline is…

Predicted Premier League table 2022-23: Supercomputer forecasts Brighton, West Ham, Tottenham, Fulham, Arsenal finish – after latest transfer activity.

Now as I have pointed out before, supercomputers are used to keep tabs on the almost unmeasurable – things like the world’s weather patterns for example.  Or quantum mechanics, oil and gas fields, molecular modelling, nuclear fusion research and that sort of thing.  The latest such computer can make a quintillion (1018)calculations per second.

There are 12 of them in the UK.  So the thought that one of those machines is diverted away from keeping an eye on our nuclear reactors and instead predicts the outcome of next season’s Premier League based on the transfer activity so far, is so screamingly funny its lunacy is only beaten by the notion that the editior’s of Sussex World think someone might fall for the con.  

What’s more, had the guys who wrote this guff actually consulted a super computer, I am sure they would be telling us which of the 12 in the UK they’d managed to buy time on.  And then there would be a public enquiry, because you really can’t get into those machines unless you are supposed to.

But still, let’s see what they got for the several hundred million quid it would have cost them to gain access for a few seconds.

Next season the league table will look like this

  • 1: Manchester City (last season 1)
  • 2: Liverpool (last season 2)
  • 3: Chelsea (last season 3)
  • 4: Tottenham Hotspur (last season 4)
  • 5: Manchester United (last season 6)
  • 6: Arsenal (last season 5)
  • 7: Newcastle United (last season 11)
  • 8: West Ham United (last season 7)
  • 9: Aston Villa (last season 14)
  • 10: Leicester City (last season 8)

I won’t bore you with the rest but Bournemouth, Fulham and Nottingham Forest are doomed to descend.

So mega-mega sums have been spent by Sussex Express on hijacking one of the most powerful machines in the world to be told that… the top four will be the same as last season’s top four.  Oh yes, and the three promoted teams will immediately go back down.  And Arsenal are in decline.

In fact if the top four did remain the same in the 2022/3 season as it finished in the 2021/2 season it would be the first time this century that the top four has remained the same two seasons running.  It all seems very unlikely.

None of this even seems to relate to the level of transfers which is what the article suggests is the basis.  So far we have…

Club In Out Balance
Arsenal £83m £14m £69m
Chelsea 0 0 0
Liverpool £73m £44m £29m
Manchester City £94m £73m £21m
Manchester United 0 0 0
Tottenham Hots £75m £6m £69m

 

These figures are taken from the Guardian’s “Men’s Transfer Window” site at lunchtime on 5 July.

A fractionally more serious approach has been adopted by Give Me Sport who have apparently worked their way through every match of the season.

They have Arsenal finishing in 7th, with 61 goals scored and 44 conceded, ending on 67 points.  The top four is thus Mancheseter City, Liverpool, Tottenham Hotspur and Manchester United.   Leeds, Fulham and Bournemouth go down.

But let’s actually look at this for a moment by checking on Arsenal’s opening four games.  The results are predicted as …

  • Crystal Palace 1 Arsenal 1
  • Arsenal 2 Leicester City 0
  • Bournemouth 0 Arsenal 3
  • Arsenal 4 Fulham 2
  • Arsenal 1 Aston Villa 2

So here we have an unbeaten Arsenal who suddenly lose at home to Aston Villa, Arsenal having scored ten goals in four games.   Quite why this game should be a defeat is of course unclear.  Worse for Arsenal fans, in this scenario Tottenham go through the first 14 games of the season unbeaten, before being walloped at home by Liverpool.

Looking at these results entered manually by a football fan, it would seem that taking either the imaginary supercomputer or fan on his desktop, it really doesn’t matter too much.

So the prediction is Arsenal get 67 points, score 61 and concede 44.   How does that compare with recent years?

P W D L F A Pts
2012–13 38 21 10 7 72 37 73
2013–14 38 24 7 7 68 41 79
2014–15 38 22 9 7 71 36 75
2015–16 38 20 11 7 65 36 71
2016–17 38 23 6 9 77 44 75
2017–18 38 19 6 13 74 51 63
2018–19 38 21 7 10 73 51 70
2019–20 38 14 14 10 56 48 56
2020–21 38 18 7 13 55 39 61
2021–22 38 22 3 13 61 48 69
2022-23 38 20 7 11 61 44 67

 

Clearly what the Give Me Sport results show that we’ll have our fourth worse season by two points.  But notably, we end up scoring the same number of goals, conceding four fewer.

Of course, if you think that everyone Arteta has bought in the transfer window is rubbish, then this might be a valid prediction.  But what it is saying is that having spent £125m last summer and a further £69m (at least) this summer, Arsenal are going backward.  Despite having no recognised centre forward playing as centre forward after the first 13 games of last season, we are not going to score more this season than last.  And that is with an older more experienced Martinelli, Nketiah, Saka and Smith Rowe.  And Jesus.

Plus despite having invested so much in the back line last summer the only benefit we are now going to see four fewer goals conceded.

Well, if that is so, it will be goodbye Mr Arteta because that will have been a huge amount of money spent in order to stand still, or indeed go backwards in 2022/23.  But I rather suspect they are wrong.

Our points total will be in the 70s this coming season.

4 Replies to “Supercomputer vs human predicting Arsenal next season. Both lose.”

  1. This appears to be the same super computer prediction that was in the Yorkshire Post last week.
    Seeing as we have bought Jesus since then and some other clubs have added to their squads, as well as players leaving, it all seems to be a trifle ridiculous.
    I am happy to follow things as they develop through the season and leave these childish guessing games to others.

  2. Silly season indeed. All to keep us clicking on their sites. Can’t wait for Friday!!

  3. @Tony I agree with you that they are probably wrong in their predictions. But like I always tell you, instead of criticizing other people’s predictions especially at the end of the season with the benefit of hindsight, why not also make your own predictions before the season starts, so that while you’re criticizing others, you also will be available for criticism. Put your money where your mouth is for once

  4. Debs I am most sorry I have not made myself clear in such matters but we do have on this site the series
    The opposition in 2022/23 with separate articles on Newcastle, West Ham Manchester United Chelsea Tottenham Hotspur and Leicester City.
    Then we also did publish the article “How easy is it for Arsenal to reach the top four next season?” and indeed I see on the home page there is a listing of a whole series on next season.
    I know there are a few more, as I remember editing them, but surely that must be enough to be getting on with. What more do you want?

Leave a Reply

Your email address will not be published.