How are the supercomputers doing in predicting this season – and what they should be looking at?

 

 

• How simply spending money on players doesn’t take clubs up the league
• Why going to football matches really is good for you and should be encouraged

By Nitram and Tony Attwood

It is repetitious of us to say that it is the silly season again because we are having an unnecessary interruption to the real business of football, which of course is playing league matches.  But an utterly stupidly dopey nonsensical silly season it is.

By way of example, the ‘OPTA Super Computer’ (ho ho) is up to its old tricks again, basically predicting exactly the same top three as last year. This is exactly what it should do, IF it bases its predictions on stats, and stats alone, which is fundamentally what computers, super or otherwise, deal with.

On the other hand, since supercomputers are fundamentally built to help us predict the weather ever more accurately, and thus on occasion save lives of fishermen, and aid the defence of the realm by keeping track of what antagonistic countries are up to, and when rogue meteors are about the strike the planet (and thus again, although more indirectly, save lives), it is rather worrying to think that they can be diverted away from such essential work into predicting who will win the league.

But the media keep on insisting that the security of our nation is of no consequence and thus Super Computers (average cost a quarter of a billion pounds each) can be re-routed into predicting football results.

But worse, the only info these Super Computers seem to have programmed into them are goals scored, goals conceded and points totals.  Anything else considered far too “subjective” to be of interest.  Thus we have discussions without statistics on topics such as how well new players will fit in, who the club should have bought but didn’t, how good is the manager, how many will departed players be missing, who did the club not sign but absolutely should have signed, and what will the boffins do?

Now Boffins is what The Mail calls the people who work with Super Computers (programmers and researchers to you and me).  Yet as with everything else in the Mail this term is somewhat out of date. Nicodemus Boffin also known as “The Golden Dustman”, is a character in Charles Dickens’ “Our Mutual Friend”.  Mr Boffin is rich, but pretends to be miserly.  I’m not quite sure how this relates to supercomputers.

Anyway, as Untold has pointed out before, computers can’t deal with a totally subjective situation like what a player does and doesn’t do. If they could, surely with so many supercomputers allegedly available at a couple of pence an hour, everyone would be using them and each team would buy exactly the players they need.

And yet The Mail says:

“The stats boffins have clearly been impressed by what they have seen from new Liverpool boss Arne Slot in pre-season”.

So, if that is true and these anonymous ‘Boffins’ are inputting THEIR opinions on Liverpool, that is a subjective input. If you input subjective data it becomes a subjective prediction and NOT one based on facts. If, for example, they input that all Liverpool’s players will regain form/ improve this year, and all Arsenal players will get worse then that gives one prediction.   If, however hey predict all of Chelsea’s signings will turn out to be great and all Arsenal signings are rubbish, that too will affect the computer’s prediction.

But these will not be computer predictions based on statistics, they will be computer predictions based on a selection of data and a set of assumptions, masquerading as computer predictions.

Of Liverpool, the Mail says, “The quality of the Reds’ squad provides confidence that they’ll replicate their finish from last term,” which is interesting in that Liverpool’s squad is EXACTLY the same as last season so a five year old would have absolutely no reason to predict anything other than EXACTLY the same finish for them,

And using the same theory of this mythical Supercomputer simply basing this season’s finishing positions on last season’s, explains why it predicts exactly the same top 8 in ALMOST exactly the same order. The only anomaly being Aston Villa and Tottenham swap places which is hardly surprising as both had terrible second halves to last season. Over the second 19 games, Tottenham amassed 30 points and Villa 29. Extrapolate for a season and that’s 60 points for “Spurs” and 58 for Aston Villa.

And surprise surprise Newcastle in 7th and Man Utd in 8th both finished on 60 points.

It’s as if this computer has looked at last season’s stats and simply cut and pasted the results into this season. Which would be a bit like a computer looking at yesterday’s weather and predicting the same for tomorrow.  It could be right, but there’s no science in that, and in fact you don’t actually need a computer.  You can just look out the window.

This season’s prediction:

• 1:Manchester City
• 2:Arsenal
• 3:Liverpool
• 4:Chelsea
• 5:Newcastle United
• 6:Manchester United
• 7:Tottenham Hotspur
• 8:Aston Villa

So the question is, is it actually a supercomputer doing all this advanced mathematics, or is it just a Boffin playing around with an abacus because as we’ve noted on Untold many times, modern state-of-the-art Supercomputers usually tend to drive satellites, predict global markets, follow weather patterns, and make sure Russia isn’t actually sending missiles to drop into the rest of Europe.

But in the end, it’s all pretty irrelevant anyway because if their laptops (which is surely all these journos are using) are programmed with last season’s data, as they certainly must be, they will be incapable of predicting anything other than what happened last year will happen again this year.

For as we have shown recently in the article How simply spending money on players doesn’t take clubs up the league it is not transfers that affect results.

But there is something to be discovered, and we published it on 4 February 2020.   For it was then that we showed that while Arsenal could put in 1.61 tackles per foul, Leicester could put in 2.08 tackles per foul.  And that Arsenal put in 4.32 fouls before getting a yellow card, while Leicester could put in 9.38.

Maybe it was the publication of our article or maybe it was utterly pure chance, but immediately after that publication, the punishment of Leicester by referees for tackles and fouls shot up, and became more in line with the rest of the league.

Why that happened of course is a matter of debate.  It would seem too weird that the PGMO might suddenly read Untold Arsenal and decide to do something. But something immediately changed.   And you never know, maybe when we publish our next set of data, something else will happen. But it would be awfully nice if these supercomputers that all the media have access to, could be used to analyse how different clubs can get away with such astonishingly different numbers of tackles per foul and fouls per yellow card.  But they don’t, so we’ll have to do our analysis all over again.