Opta

[sidefootsitter]

Ladies and germs, does anyone here have the Opta formulas or is willing to cooperate to re/create a fascimile of it?

[The Philosopher]

if yr talking about opta stats from the uk, i can tell u to forget about them. i had some friends that used to work for them in the data gathering, they made up almost all the numbers, apart from the obvious major ones, rather than do the wrok of properly counting stuff. they dont get paid much and they generally hire young people, who couldnt give a f''k about the real numbers

[sidefootsitter]

Well, then I am willing to (re)create a reasonable fascimile of what Opta is supposed to be like.

[sidefootsitter]

Woe be me ... no one is interested in creating a credible system for judging a player's performance on the field???

Come on, people ...

[NoSix]

Ladies and germs, does anyone here have the Opta formulas or is willing to cooperate to re/create a fascimile of it?

I think you have your cart before the horse - before you can begin worrying about the formulas, you need the raw data.

[sidefootsitter]

I think you have your cart before the horse - before you can begin worrying about the formulas, you need the raw data. You can create a reasonable formula at least by position and then adjust it if you feel that it's leading you astray for the overall grade.

Opta must be something like (I approximate):

a simple pass = 1 point

a 20+ yd pass = 2 points

a 20yd dribble = 3 points

a cross = 3 points

shot on goal = 5 points

an assist = 10 points

a goal = 25 points

a turnover = -1 pt

corner surrendered = -2

PK surrendered = -10

and so on... 25 or so different categories...

[ChaChaFut]

Why do you need a formula, or why does everything have to add up to a total?

Why don't, for instance, separate things into categories, and get some meaningful data out of what you count. Let's say for example that you want to know how accurate a player is in his crossing. Get his total of correct crosses, divide it by the total of crosses attempted. This should be kept separate from, say, the average of balls recovered per game. A player could be good at the former, but bad at the latter. That way you can compile meaningful rankings in different areas of the game.

[Andy_B]

Categories by position is much better than total points.

It would be great to compare all the right wing backs distribution % or comparing how many crosses found their target from all left wingers or comparing how many balls were lost by the holding midfielders.

Andy

[ChaChaFut]

I agree. For example, to know if an attacker is a good attacker, you measure his scoring, chances created, etc. For defensive players, you focus in ball recovery, clearances, etc. Other data, like passing %, balls lost, etc. can be used for all player positions to rate their overall qualities.

[numerista]

I agree. For example, to know if an attacker is a good attacker, you measure his scoring, chances created, etc. For defensive players, you focus in ball recovery, clearances, etc.

While I see where you're coming from, I'm not a fan of the offensive/defensive dichotomy. It wouldn't give much credit to a team with strikers who force turnovers, outside backs who serve dangerous crosses, and center backs who are excellent set piece targets. In MLS (and probably most leagues), a team like that could do very well.

[ChaChaFut]

Oh no, I'm not saying rate defenders based only on their defense, or forwards only on their attack. That would be a waste. I was just saying, for instance, that if you want to rate/rank players on what they are best at, you can have specific areas to observe. For instance, you will still keep a record of how many shots on goal Alessandro Nesta made in a tournament, but since he is not likely among the leaders in this category, you can still use other categories, such as balls recovered, or blocked shots, to confirm how good of a CB he is.

And of course, as you said well, it is valuable for a team, for example, that his centerforward steals balls, or that their backs create chances, or that his DM scores on 75% of the chances he gets, etc. That's my point. Stats need to be categorized, in order to be accurate.

sidefootsitter's method suggests that you give overall ratings to all player types by putting together a total of points based on all areas of the game, which I see displaying misleading or meaningless results.

[Chad Sexington]

Resurrecting a dead thread, but...

I just saw some Opta data for the first time--one of those maps of all of the passes that happened in an EPL match. Pretty cool. I'm wondering, though, how reliable the information is--like, if the Opta map says that a particular pass was made halfway between the center line and the top of the 18, can you trust that that's actually true to some degree of accuracy (like, +/- 2 yards, or +/- 5 yards)? Or is it not really possible to get precise location data from an Opta map?

Anyone have any knowledge about any of this?

[Ventjeluke]

I was wondering whether someone had discovered the real OPTA-formula or whether someone has created its own formula. I'm very interested in these kinds of statistics.

[RichardHKirkando]

I tend to ignore stuff like the Opta and ACTIM player rankings. If they can't be bothered to share how they come up with it, I can't be bothered to care. Some arbitrary number assigned to a player doesn't tell me anything. I don't even know what they're measuring - points? What is this, basketball?

I do rely on Opta's event data quite a bit for my own stats, though. I have sort of a work-in-progress player rating system for the Premier League (in a "goals above average" type of format), and most of the numbers that go into it ultimately come from Opta. I think the stuff you see on the Telegraph, Guardian, ESPN, etc. websites are fairly accurate, for the most part.