Which Stats? [Split from J vs A thread]

[cesc408]

Lots of essays. Jesus has looked poor recently. Will be shipping him for Lukaku after internationals

[Resonare]

A more concrete example would be this

Goal threat = Goal attempts x Conversion rate

That's just a function of "goals scored" Not goal threat.

It's one of the most simple equations of what goal threat (which is dependant on goal attempts and conversion rate) could look like.

The fact my point is true for that means it's even more true when it comes to other more complex formulas of e.g. f(x, y, z) etc.

That's an example formula that shows how goal threat is dependant on both goal attempts and conversion rate.

It is dependent on both of those stats, as well as many other stats. Each relevant stat can also be used in isolation to measure goal threat, they all just have a different correlation coefficient.

The red part is wrong. Again, you cannot use goal attempts in isolation to get goal threat. Going back to my earlier example,

Goal threat = Goal attempts x Conversion rate, so with a Goal attempts stat of 100 that means Goal threat = 100 x Conversion rate which is an algebraic expression not a value. You can come up with literally any expression involving Goal attempts and Conversion rate and it will be the same i.e. you will always end up with an algebraic expression.

You can use it to measure goal threat. You can say that players with more goal attempts are on average more likely to score than players with fewer goal attempts.

Two things being correlated is not the same as being able to one of those things to measure the other.

This stat works in isolation. Goal attempt is one relevant metric that can be used to measure goal threat, among other relevant measures. Goal attempts and goal threat are correlated with each other, even though they are not 100% correlated.

Ok mate, with a goal attempt value of 100 please find the corresponding value of goal threat.

[Stevieste]

Think might forget the usually bed time story to my kids tonight and read the last page on here to them, they be asleep in minutes

[bloggie]

Think might forget the usually bed time story to my kids tonight and read the last page on here to them, they be asleep in minutes

I had to use an advanced statistical calculation on whether it was worth reading.

[BuckJones]

The title of this thread should be renamed to Resonare or Finisher1?

[Tallerico]

Yikes, I have been tempted out of serial lurking. Finisher is 100% correct and has been very patient in explaining why. By all means, choose to believe something else, but please spare the rest of us.

Note, post edited to remove my uninteresting comments about Aguero given the thread has been split.

Now, I must rest.

[Stemania]

This topic has now been split from the original Jesus vs Aguero thread. Sorry for any inevitably lost context.

[Finisher1]

Yikes, I have been tempted out of serial lurking. Finisher is 100% correct and has been very patient in explaining why.

Thank you. I think every rationally thinking person here agrees with us.

[hancockjr]

If you are going to use a stat don't you want evidence it is predictive of the target variable? Has anyone gathered data to see what stats are actually most predictive of future FPL points?

[Ruth_NZ]

Yeah was done on FFS a while ago but I can't be arsed to look it up.

The best single predictor of goals that will be scored is goals that have been scored.

[hancockjr]

So why do you use more complex stats?

[Patrician]

Yeah was done on FFS a while ago but I can't be arsed to look it up.

The best single predictor of goals that will be scored is goals that have been scored.

As we get deeper into the season, that will be increasingly true, but not in the early part. After the first game, who would you prefer for your team? Player A who has had 2 shots and scored 2 goals, or player B who has had 10 shots and not scored? The problem with goals already scored is that data is easily available to all, and I think it is important to idenitfy the Mahrez from 2 seasons ago, or perhaps the Richarlison this season before the hoards do. Once the goals are flowing there is no edge.

Some data on shot conversion:

Season 16/17

Shot conversion – 16%
Shot in the box conversion – 19%

Season 17/18

Shot conversion – 16%
Shot in the box conversion – 20%

So very consistent. When you look at individual players though, big inconsistencies. Looking at two well known players, Kane converted 26% of shots in 16/17, 14% so far in 17/18. Aguero 14% in 16/17, 26% so far in 17/18.

So for predicting future points, which is what FPL is about, what do I assume the conversion % is for either of these two players?

Analysing number of shots and shot conversion %, is just a complicated way of looking at historical points/goals, which is easy insight that everyone has. The key for me is identifying the future bandwagons that should score at a faster rate than they have done so far, and bringing them in at the right time (likewise bravely ditching high scorers who don't have the underlying stats). In the early part of the season especially, I think looking at the number of shots in and outside the box is the best clue.

[Ruth_NZ]

So why do you use more complex stats?

Because no single predictor is very accurate. But I'm not really stats obsessed anyway and I use the macro picture (a player's track record) as much as recent stats.

[Resonare]

Yikes, I have been tempted out of serial lurking. Finisher is 100% correct and has been very patient in explaining why.

Thank you. I think every rationally thinking person here agrees with us.

At least I've taught you how to properly give your opinion

[BuckJones]

Yikes, I have been tempted out of serial lurking. Finisher is 100% correct and has been very patient in explaining why. By all means, choose to believe something else, but please spare the rest of us.

Note, post edited to remove my uninteresting comments about Aguero given the thread has been split.

Now, I must rest.

He is most certainly the correct one imo. However, you exaggerate his patience a touch imo.

[Resonare]

By all means - if you feel he is right then go ahead and answer the problem I posed to him in his place.

Each relevant stat can also be used in isolation to measure goal threat, they all just have a different correlation coefficient.

Ok mate, with a goal attempt value of 100 please find the corresponding value of goal threat.

[BuckJones]

Maybe Tallerico was right about Finishers patience...you are unrelenting.