- Another St. Totteringham’s Day; Arsenal up a bit Man C down a bit
- Teen scores seven for Arsenal, the Arsenal team and what happens afterwards
By Tony Attwood (and the calculator button on his mobile phone)
Football data with the name “Opta” attached to it, is omnipresent. Yet there is no Opta.com website. Further, Opta is often criticised by Untold Arsenal. So I thought it might be a good idea to explain what Opta actually is.
Opta is simply a brand owned by Sports Data which itself is the brand name of Stats Perform, so the first thing to understand is that it is not a separate organisation run by a university or trust or charity or government. It’s a brand. Like Levis.
That is not to say that there is anything wrong with their data, but it is just helpful to know who we are talking about. It’s a brand owned by Stats Perform – a company that feeds sports data into a computer and then gets it to make predictions, which can be right or wrong.
My prime beef with Opta is the claim (made largely or exclusively by others, not by Opta itself) that a supercomputer is being used for the predictions. I doubt this because a) no one ever says which supercomputer is being used, b) supercomputers are still very rare, c) supercomputers are phenomenally expensive and d) supercomputers where they do exist are being used for more important things than predicting football results (such as protecting the nation from nuclear attack, and predicting the weather for airlines and shipping companies).
None of this is Opta’s fault as far as I know. Rather it is the people who write about Opta and make things up. What Opta says about itself is “Opta data is generated in real-time though a combination of human annotation” [ie people typing numbers into computers], “computer vision and AI modelling.” [These last two are basically the same thing.]
So Opta can take videos of football matches, turn them into a format that the computers can read, and then make predictions. A bit like Untold Arsenal only a) a million times faster, b) not prone to misreading what the pocket calculator said and c) less interested in red wine.
The next thing to know about the Opta figures is they keep changing. For example around 1 April Opta had Liverpool as favourites to win the league. Now they are third favourites – which makes their prediction about Liverpool being favourites look pretty much like everyone else’s.
In fact Opta figures jump around because they emphasise the last result or two. Before Match Day 31, Liverpool had an 11% greater chance of winning the league (according to Opta) than Manchester City, with Arsenal another 15% behind them.
Now Manchester City have a 45% greater chance of winning the league than Arsenal and a 70% greater chance than Liverpool. All because of recent results which you or I could have worked out just by looking at results.
My argument is that I don’t need an alleged supercomputer (which I suggest is not a supercomputer at all, in the proper sense of the word – and to be fair to Opta I don’t think they who call anything they have got a supercomputer) to tell me that Liverpool, following one win, two draws and two defeats in their last five league games, are very unlikely to win the league. There is nothing computer-like going on here.
The fact is that by using the word supercomputer in association with Opta, various gambling and tipping companies are trying to make out that it is now possible to get closer to the right result than before, whereas in fact all Opta and other firms do is stop focussing on odds, and instead focus on the predictions that make the odds.
And there is an important point to add here, for Opta is predicting the whole Premier League table. And yes I can reveal that according to Opta Aston Villa will be 4th, Tottenham 5th. Just like now. And that’s what it mostly is. “Just like now”.
In other words, Manchester City will win their game in hand, and the only revelation is that further down, Newcastle will continue their better form than Manchester United to make up the one point gap between those two.
Oh yes and they predict that Luton, Burnley and Sheffield United will go down. I bet you didn’t see that coming.
But here’s the twist. At the start of the season the alleged super-computer predicted Bournemouth to be relegated, and Manchester United to finish in the top four. Whatever happens Bournemouth will finish higher than they have ever done in their entire history and Man U are extremely unlikely to rise from their present position of sixth, and could well end up in seventh.
As they say, “this computing lark; it’s a funny ol’ game.”
The notion that football can be predicted accurately by any type of computer, ‘super’ or otherwise, is ridiculous.
Computers simply do not do subjectivity.
What I think you are alluding to Tony is that basically the computers being used are for want of a better word, are ‘gambling’ computers. They are ones used in the ‘gaming’ industry.
They are great at predicting the chances of anything to do with numerical chance. Dice, roulette, cards,
But as soon as you factor in subjectivity, and football is probably the sport most subjective sport there is, then computers have problems.
You hit the nail on the head Tony when you said all they can basically do is look at what’s happened before and predict the same will happen again.
They do it over short term, as you have pointed out above as they simply look at whats happened over the LAST few games and predict the same will happen over the NEXT few games, and they do it long term, which is why when they predict the next seasons finishing positions they will almost always mirror the season just passed.
And that’s because all a computer is programmed with is numbers.
If you input data that shows Man City accumulated 100 points in season ’23 – ’24 why would it predict anything different for ’24 – ’25?
The number of games are the same. The teams are almost exactly the same, save the worst 3 that are effectively being replaced by the next seasons worst 3.
So unless the person programming the computer is imputing all the dozens and dozens of ‘subjective’ factors of course it’s going to predict exactly the same thing will happen.
And even if the programmer is imputing ‘sublective’ numbers, they will be numbers made up by the programmers own subjective numerical valuation of said factor.
Ergo Arsenals young team is a year older, wiser and more experienced.
So I ask you, how does he turn that enormously subjective factor in to a number the computer understands? And in any case who says the programmers subjective assessment is anything like accurate?
Another example of how computers struggle is the weather, that I believe is actually monitored and predicted by a real ‘Super Computer’ held in the Met offices Dorset head quarters.
As anyone who has planned a weekend barbeque knows, weather forecasts are as accurate as a tabloid headline.
In other words it’s a load of bollocks.
Guys,
why do you all get worked up about it. AI is supposedly a form of intelligence, and the only model it can gets its inspiration from is the humna intelligence (or stupidity…) and thus an not be anything else then flawed.
True., because of computers, it can ‘think’ faster, that is make more computations, scenarios in its brain then we could.
So the end result is flawed more often they they would admit. Like the fact that when AI does not ‘find’ a fact, it has a tendency to 1) invent an answer 2) store it forever thus creating an alternate set of facts where it should just say : I don’t know. Ain’t that some very human trait ?!?
Now, the business model of Opta is to get people to bet. As simple as that. And to make believe anyone who uses their data that they have a head’s up on the other bettors. That their hunch is based on some sort of science and will pay off.
All these ‘so-called’ news outlets are getting revenue from the betting industry. So they push that agenda. And we end up with these kinds of predictions.
AFC Bournemouth will not definitely finish higher than they ever have done before as you have stated….they finished 9th in the 16/17 season.
Yes my fault Mark, thank you. What I should have said is that they will finish with more points than in 2016/17 since they already have that, but could finish above ninth as well.
Chris
First off, please don’t think I am ‘worked up’ about this. It’s just something for an old guy to do to waste away the day when the sun refuses to shine. Now that does get me worked up. Grrrrrrrrrrrrrrrrrr
But what you say is right:
“Now, the business model of Opta is to get people to bet. As simple as that”.
As I have said on here a few times, these ‘super computers’ that churn out all this garbage are all owned by Gambling company’s.
I have done research on this and done a couple of big posts, even naming the companies. And they do call them super computers.
I have also done research and posted the results on here, as to what is and what isn’t a super computer, and although they may not be the ones used to predict wars or tsunamis, they are super computers.
And the reason they can be called a super computer is because the definition is so broad.
Apparently the standard iphone of today, has as much power as the super computer of 10 years ago.
So okay, as Tony keeps saying, they may not be using the ‘state of the art ‘super computer’ the ministry of defence uses, but that does not mean they are not using a super computer, all be it a 5, 10 or 15 year old one.
But as you say, it is all about encouraging gambling through the belief that you can ‘beat the bookies’ with the help of technology.
It’s as simple as that.
“…and c) less interested in red wine.”
Thank you Tony, that made me smile
COYG