Holebas - 4th best full back in the Premier League

In my opinion if an algorithm or set of data tells us that Guediora is more effective than Doucoure over last year's games then at that point you make a decision that the methodology for producing these "facts" is useless.

 

Could our favorite mad Greek be off to Turkey?

http://sportwitness.co.uk/watford-player-offered-european-giants-alternative-signing/

 

Good for you. Nothing wrong with you making your own conclusion based on the information you have in front of you. And people can then consider that opinion based on the thought processes you invested in it.

But to me, suggesting that Doucoure was more effective than Gueds in a game he didn't even play in tells me all I need to know about your opinion and understanding of the subject you have chosen to post on here.

 

Care to explain your thinking? Surely these stats are even more useless if they don't create an average comparable so you can rank players side by side? Number of games should be irrelevant (perhaps with a measure that removes players who have only played a handful).

 

Championship team Brentford also use a complicated algorithm scouting tool to sign players. Using this, they have got promoted from league 1, got to the play=offs and finished in the top 10 in the other 2 season. They have got progressively worse each season since promotion though, finishing 8 and 10th since their 5th place finish. Not sure if it has worked or not!

 

Prodl aside maybe, we need fresh defenders to take us to the next level. We weren't exactly water tight last season and Holebas lets crosses in too easy as I've said before.

Interestingly Kyle Walker is 11th yet Spurs prepared to let him go.

 

This thread is quite incredible. There's a nonsense article that claims Holebas is the 4th best full back in the league, which some people jump on in an "I told you so" way, as though this vindicated their rare opinion that he was good this season (even though he won the forum poll for POTS). Then it's pointed out to them that the same performance ratings come up with some truly bizarre rankings amongst the rest of the squad and people argue that even though those they don't agree with those findings themselves, somehow WhoScored rankings are better than human opinion because they're more consistent (or as one particularly confused individual calls them "facts").

Of course stats are hugely important in the modern game and people like Opta are experts in what they do. There are all manner of aspects of the game, from crosses to tackling and shots on target that can usefully be compared when players are being scouted. But it's all about the raw numbers, or the percentages, not the quality of the actions. Pereyra's Goal of the Season increases his score in that game by the same as Capoue being gifted a goal against Chelsea. An algorithm can't tell the difference in quality between those two actions, to them they're both just goals, but a human can.

Incidentally, N'Golo Kante, who cleaned up all the individual POTS awards this year, is only the 34th best player according to WhoScored, behind such top performers as Otamendi, Kolarov and Mustafi. Says it all really.

 

Why do you want me to defend your misconception of what the WhoScored stats are produced for? You are the one that thinks they are there to tell you if one player is better than another, when I have said over and again that they are not. Read my words, they are just a statement of statistics, the conclusions are for you to make, and clearly, though you despise your own deduction, the stats tell you that Gueds performed better than Doucoure. I have never said that. That is what you have concluded, not me, and because you don't like that conclusion, you then decide that the stats must me 5hite.

My thinking is that you think Gueds is 5hite, and so much so that he can't possibly have been more effective than Doucoure, even though the stats you refer to are from different games, and Doucoure, however well he played, was involved in the the worst run of games this team has had in decades.

 

Steve. Thanks for this opinion piece.

I prefer facts though mate. Can we get it back on thread...

How about that then? Holebas fourth best in the Premier League. Brilliant. Completely justifies my opinion that he was our best player this season.

Fantastic.

 

Just read it an weep Stevey boy. Fourth best in the Prem. FOURTH best eh. Brillo. Let me know when this rubbish gets as annoying as your rubbish.

 

Who'd a thunk it. Kante only the 34th best player on WhoScored. You clearly, for all your indignation at my posts, haven't understood a word I have posted. Baahhh.

 

You agree Doucoure did well at least. You mention him playing in an out of form team so presumably these stats are not valid in comparing player output with player output over a 10/15 match sample.

Even if their measurement is valid, they need to be subjectively adjusted for

A) Form of team as a collective
B) strength of opposition
C) tactics deployed by manager

In summary you might as well have a starting point how much asparagus each player consumed during the season and go from there.

The shortcomings of these stats is so obvious when you look at some of the perverse rankings Steve showed above. I do appreciate you have boxed yourself in now though so will need to defend them irrespective.

 

424 are you aware these ratings get adjusted by an algorithm for the team scoring and conceding goals? And whether they keep a clean sheet? It isn't just pure stats but an unspecified adjustment is applied according to a "who scored" formula.

 

You keep airing your theory about people prejudging Guedioura and calling him useless before he kicked a ball.
This is just nonsense - he had tremendous credit in the bank and the clear majority loved him to bits based on Championship form and his isolation at the hands of QSF, in spite of which he always remained upbeat and committed. They desperately wanted him to succeed but were forced to accept he wasn't good enough. No doubt a few wrote him off sooner, but you always get those.

 

I can understand if you don't want to read my posts Jumbo.

But please, in future, if you are going to reply to them, please can you make sure you read them, because everything you bring up here has been responded to a few times already.

You are now back to saying that statistics have to be subjective to mean anything. Which again, IMHO, indicates that your grasp of what is meant by 'fact' is e tremely shakey. For the sake of tbe forum, I'll refer you to my previous answers rather than repeat myself.

 

How can it be a fact if the whoscored algorithm adjusts the players performance by collective team factors? The adjustment doesn't even make sense as you get punished for playing well against Man City but your team gets tonked but you get boosted if you put in a mediocre performance against Sunderland but your striker has a great game. It's utter nonsense and you repeating it is a "fact" makes no difference to that. It's a rules based number based on underlying stats.

 

Because it is an algorithmn and not an opinion. Betting sites would not use the data if it was subject to an opinion. Clubs would not look at the data if it was subject to an opinion. That such people trust it, when their own investments rely on it, and you don't, when only your opinion is being judged by it.

Why are you asking ME, again, to explain why YOU think it should be a valid measurement of how good one player is compared to another, even though I have repeatedly said you are mistaken in thinking that is what it is for?

 

I'm not asking anything of you. I'm merely pointing out that if it uses an utterly basic algo to adjust the score (which can be argued to be a non logical adjustment anyway) then the rating is NOT a "fact" that you can take joy in bashing people with in an attempt to show yourself as the intellectual football fan.

I can create an algorithm that adjusts the scores by a different set of rules and that isn't a fact either

 

so clubs trust it and make investment decisions based on it? So we are bracing ourselves for an incoming £60m bid for Holebas but might be able to pick up Kante with a % of the funds received ? Good stuff.

 

No. You see, the clubs understand the value of the data and consider it along side other information they have. They then make an investment decision based on all factors.

Where as Jumbo and Stevie look at it and start crying because they think some nasty man wants to take their dummy away.

 

Can you? Give me an example that only uses facts that will do that.

 

We're only questioning the methodology as it seems to result in some bizarre outcomes. To be honest it just looks to me like you weren't even aware that whoscored applied an adjustment and are now waving your arms around trying to retain credibility.

Easy. Whoscored effectively gives you a boost when you play a poor team at home. Instead I would adjust by average starting price of your team across the main 10 bookies (fact and observable). I would give you a boost if your performance is in a game where you are underdogs on the basis that you get less time on the ball and are able to make key passes. The extent of the boost would be determined by how big a statistical outsider you are.

Or you can even adjust by wage bill of opposition starting eleven over your team. Fact.

Adjust by home verses away. Fact.

Any of those algos will have the opposite impact of whoscored's primative adjustments.

 

You mean they look at it and then discount it because it doesn't help assess how good a player truly is ?

You are clearly intelligent and very articulate (if a little condescending) and support the right team, but I think we are all a little confused why you don't just put down your spade, climb out of the hole, and join everyone else on this thread in laughing at the bizarre conclusions that can be drawn from this data.

 

Err. You haven't got a clue have you.

All of those prove my point, though some of the facts have little to do with a players performance and I doubt you'd get very far plugging it to clubs or the bookies.

My invitation was for you to prove your claim that you could create an algorithmn that generated something that was not a fact, but you take great delight in demonstrating alternatives that equally, according to your own claim, produce facts, thus proving what I am saying to you.

You also decry WhoScored for boosting the home team, and then state that one of your revolutionary boosts that is far more effective than theirs is to boost the home team. A claim you then joyfully declare to be a fact.

Please keep going. Its like watching Nathan Ellington trying to score a goal in a Watford shirt.

 

Blimey Undies! That's a bit harsh.

More likely to join the others who got outa here days ago when dunb and dumberer decided to flex their logic circuits. Don't know which one I am, but you guys ensure I'm not on my own.

 

Apparently Fenerbahce sniffing round after they couldn't Clichy. I wouldn't mind Holebas going and we got Clichy instead, but both on the wrong side of 30 so would prefer we either give Mason a go or a younger new transfer.

http://www.hertfordshiremercury.co....jose-holebas/story-30387452-detail/story.html

 

You're really struggling here so I'll leave it. My methods would penalise a player who plays for the strongest team. Get it now?

I'm going to leave you waving your arms in the air in indignation because you've got to the point where you don't even think about what people have posted.

You are so desperate to defend a flawed model that you just look like a ranting maniac. Enjoy analysing your useless stats over the summer.

And guess what. All these algos still are not facts. Have fun.

 

That is sweet and it sums up the validity of your argument perfectly. You don't even have a clue what an algorithm is.

Sorry for the maniacal ranting everyone.

 

I don't really get this fixation on the 'algorithm' when it's just going to be a glorified vector of weights. One could easily pick a set of weights that output complete nonsense.

 

My fixation is with the data that was used. I think the algorithm talk was an attempt to blag the matter with cod science to justify some cheap comments made regarding some decent Watford players.

 

Unfortunately not. You got yourself into the mess because you didn't understand that the raw data was adjusted to arrive at the whoscored ranking. From then on you just kept digging.

Of course none of this lack of understanding of the figures you were talking about changes the fact that we both agree Holebas was one of our best players last year.

 

I had suspected for a while that this was the source was the source of your confusion but thanks for confirming it. You don't understand the difference between the primary data (the stats) and the WhoScored ratings. I'll try to explain it as simply as possible for you:

Opta collect all different types of data during the game, this is primary data and takes the form of statistics about all manner of things (crosses, blocks, tackles, shots etc etc).

WhoScored then run all this data through an algorithm to come up with a performance rating for each player. The algorithm is written by people who have used their subjective judgment to decide how much weight to give each action (e.g. something like +0.2 for each block, +0.4 for a shot, -0.3 for losing possession). So the ratings can be described as objective, but only within WhoScored's own parameters. They are not "facts" about how effective a player is.

WhoScored comes up with perverse rankings due to the inability of an algorithm to distinguish the quality of certain actions. So, a tackle by the corner flag that gives away a corner counts the same as a fantastically timed last ditch tackle that prevents a goal. Anyone watching would be able to tell which is the more important action, but plugging it into an algorithm takes that element of judgment away. It also can't pick up on other, less tangible things such as positioning. Prodl was rated 5/6 of our centre backs this year, and one of the reasons is that he makes fewer tackles per game so he accrues a lower score on this action. But what if he doesn't need to make as many tackles because his positioning is better than our other centre backs? That's also not reflected in ratings such as these.

Anyway, the original disagreement was about whether you could use these ratings as evidence that Holebas was the 4th best full back in the league. You've accepted that WhoScored ratings don't tell you how well a player has performed, although you do seem to keep referring to them as measures of effectiveness (somewhat confusing, as most people would think that there is a direct correlation between a player's performance and their effectiveness). Either way, the logical conclusion is that you agree that they don't help to prove that Holebas was the 4th best player in his position in the league, so at least we've made some progress.

I also refer you back to Holebas winning this forum's POTS vote. You seem to think that it's only you, as a particularly enlightened fan, that values him and can see how well he played, yet that's manifestly not true.

 

The

Their algorithm co-relates factual data, not a weighting subject to WhoScored's opinion. Their weightings are based on factual statistics, not random, subjective opinions. Their data would not be taken seriously by the professionals who pay them for it if there was any whiff of data that was not subject to fact.

Have you never seen a player play well but not achieve anything? If you have, then I wonder why you are confused by the difference between performance and effectiveness. If you haven't, then I can only assume you don't watch much football.

Why do you think I am denegrating everyone on the forum? I am very aware that Holebas is highly rated on here. That is why I have included Gueds and Amrabat to illustrate that you are misunderstanding the WhoScored data. I am trying to explain that misconception to you and a couple of others.

And nice that you only now pick up on the fact that I have been saying from the start that the WhoScored ratings don't prove anything. That is exactly what I have been trying to impress on you from the start (read the thread and you will see that).

My argument, as I have explained far too many times, is not that the stats create a measure of how good a player is, but that they are a valid, however flawed, source of data with which to inform your opinion or support an argument.

You have rubbished WhoSored as a measure of how good one player is compared to another. I have been telling you that you are being unfair because that is not what the data they put out is for. You and a couple of others are the only ones that seem to think it is.

 

Maldini made an average of 1 tackle every two games.

WhoScored would think he's turd.

 

I suspect Clichy's wages are significantly higher than Holebas' though, and at his age, will have no resale value. If some sort of "pay as you go" deal can be agreed to, where his earnings would be dependant on appearances, with bonuses for clean sheets, assists etc, then maybe he'd be worth it.