Draw the correct number of fingers, for one Or know what I actually listened to on Spotify last year, for another...
I think you definitely need a mix of both to try and find the best possible fits for your club. Both can be used to verify and back up the other's initial suggestion. A scout could come across a player they wasn't aware of, check out the data available to them and process from there. Likewise data could highlight a player previously unknown to the scout/ club who can then go and watch them in action to verify if they'd be a potentially good pick. I think the eye test is probably slightly more important, probably 55-45 in the scouting's favour (providing the scouting of a player is done over weeks and not a one-off) but there will no doubt be a good argument for being the other way round. My reasoning is that data can potentially be misleading, IIRC a few years back @PowerSpurs once brought up some of the data from the CIES website that had Harry Winks as statistically the 5th best midfielder in the Premier League at the time, yet we all know that not at any point in Winks' career was he ever as such, in fact he'd have been closer to the 5th worst. I'd also be curious to see how data represents players such as Modric. He wouldn't be near the top of any lists for assists, chances created or goals, nor would he be high up on tackles, headers or interceptions, he'd likely have good distance covering stats and high passing percentages (so would Winks!) but I'm not sure it'd be able to highlight too much to make him standout from others. Yet all you'd need is 45 minutes of watching Modric in the flesh to know you're witnessing a generational footballer and one of the greatest midfielders to ever play the game.
The question is whether the measures that make Modric great have actually been produced. Or in data analytics terms, the "source data" : 1. was never gathered (irrespective of which measures require it) 2. was available, but the measures were never defined 3. was available, the measures were defined, but were never computed.
You simply give the AI all the data and instruct it to come up with the best algorithm for choosing today's best players on the data known 5 years ago.
He was the midfielder with the best passing percentage during the 18/19 season at one point. Did something similar for Leicester in the Championship when they came up too, to be fair. It's a shame that injuries did him in, as I don't think that we ever got to see him at his best. Had a lot of potential. I'd say the same for Skipp, too. Very unfortunate.
Not from statistics, though. It's regurgitating Wikipedia, which is doing the same to other sources. Modric's stats are very good too, of course. Passing, chance creation and interceptions, mainly. It can't detect things like composure, positioning and dictating the tempo of a game, though. They're all the areas where he's head and shoulders above his competition.
That's true, but he also had very good progressive passing stats, which is surprising. He didn't do that many overall, but they were generally accurate when he did. I think that his main issues were that he became too cautious and lacked physicality. His size and getting repeatedly crocked made him too easy to bully. It'll be interesting to see where he goes next, as he fell out with the Leicester manager.
As machine intelligence currently cannot devise from first principles even the most basic fundamental "best" algorithms in CS, it certainly will not devise a good algorithm to perform the above task. That aside, you did not address the actual question (the video capture system ) .
I think you are out of date on AI capability. But what is certainly true is that if AI can provide an edge in player assessment then it will be short lived because everyone will adopt it. So we need to be at the leading edge not stuck in the past.
It's basically analysing all the commentary from all the people who have watched him. That's part of the data set and is likely to be more correct than what a scout observes by watching him 10 times. But what we really need to know is what he looked like when he was 20 so we can select the best Modric lookalike from the 100s of midfielders we could sign. Or better still identify a pattern from watching De Bruyne, Modric, Xavi, Iniesta etc early in their careers that we can use in recruitment decisions. No scout can do that. AI probably can.
We'd better hope that this isn't accurate, as it'll funnel all the talent to the absolute top clubs. They'll be able to spend the most on AI and offer the best contracts to the prospects. I think that clubs like Brighton currently have a data advantage, but it won't last long.
I think he was always cautious. He had a few stellar performances but there was rarely an extended spell where Winks was looking anything special. I’d be surprised to see him back in the Prem unless it’s a bench player for a bottom five side. I think his best chance of top flight football is looking at another move to Serie A, failing that I wonder if a MLS move might appeal to him.
Genuinely. From myself, and my colleagues. OTOH, I cannot comment on what I think of your knowledge of current machine intelligence capability, because Spurf will probably immediately ban me.
Danny Hall at the Sheffield Star says The Bleeeeeeeeeehds are looking at a double loan deal for Alfie Devine and Ashley Phillips Bit of an issue that Alfie Devine reportedly does not want another loan...
