Topic: Darwin Núñez (Darwin Gabriel Núñez Ribeiro)

[Cpt_Reina]

So let’s throw it out …. I mean we’ve got 7000 league minutes on Darwin over the last 4 seasons where his numbers are consistently elite but it might be a bad measure and/or have not matured fully as a data set so let’s bin it

Now we’ve binned xa … how would you like to evaluate and judge Darwin’s performance level?

I never said throw it out. There's no need to lash out at me or put words in my mouth.

But I (personally) wouldn't put as much faith in his benfica numbers, different league, different opposition, different tactical set up, different footballing culture you know environmental factors that apparently don't exist, if I was looking to assess his value to a PL team. I'd like to focus on more translatable data, ideally.

I don't know how that affects any analysis if you just go on his Liverpool numbers. Maybe we need 2 more seasons of wait and see until we're comfortable with that.

[Chris~]

Agreed.

But again idk maybe I place too much importance on such things, and maybe the sample again would to be too small, but playing for months as an attacking player without Trent, Robbo and Salah available alongside you would have a significant effect on production.

Someone like Gakpo or Diaz having to play through that period, where someone like Jota sits it out and gets his minutes in before and after they return that's enough to place some doubt regarding even a seasons worth of data.

So what's the ideal sample size to remove that variance? 3 seasons? 6000mins? Football is an inherently small sample sport for a lot of if not most of its highest value actions so the time needed for sample size to get to where it needs to be isn't short. Seems like a fairly slow to market way to adequately assess someone if the data set needs so long to mature.

They still get to play with good-great players though. It's not like we replace them with below league average quality. Without Robertson, Trent and Salah we scored 4 against Bournemouth, Luton and Chelsea (small sample size says get rid of them) we have lots of very good players and Nunez is one of them. I think a seasons worth gives you some confidence, you can assign parameters around that (say in your example Diaz is really low one season you can look at why, or are they a kid, or the team rubbish in general) but you can compare to seasons worth to see his overall quality. With Nunez I think there's enough historical (past seasons) to go with this season that you don't really need to break it all down. If it was the first time it jumped up then yeah, be a bit quizzical about it

[JackWard33]

I never said throw it out. There's no need to lash out at me or put words in my mouth.

But I (personally) wouldn't put as much faith in his benfica numbers, different league, different opposition, different tactical set up, different footballing culture you know environmental factors that apparently don't exist, if I was looking to assess his value to a PL team. I'd like to focus on more translatable data, ideally.

I don't know how that affects any analysis if you just go on his Liverpool numbers. Maybe we need 2 more seasons of wait and see until we're comfortable with that.

We’ve got 4k + minutes for lfc all comps.. 3500 league minutes  .. it’s more than enough to evaluate a player with a high confidence interval especially when it corroborates rather than deviates from his previous club performance

So we’ve got enough minutes how would you like to use them to evaluate him
Not xa … so what ?

[Cpt_Reina]

We’ve got 4k + minutes for lfc all comps.. 3500 league minutes  .. it’s more than enough to evaluate a player with a high confidence interval especially when it corroborates rather than deviates from his previous club performance

So we’ve got enough minutes how would you like to use them to evaluate him
Not xa … so what ?

I know you don't like hearing my opinion that not all minutes are created equal. So I don't expect we'll find common ground here when I say that the Europa league, league cup and fa cup minutes are of less value to me for much the same reasons as the benfica minutes.

Not worthless data, but less valuable data. As always I'm just here for the strongest most robust representation we can get. League minutes, champions league minutes. These are what I value in terms of analysis. The others are noise.

Do we still have enough minutes?

[LallanaInPyjamas]

We’ve got 4k + minutes for lfc all comps.. 3500 league minutes  .. it’s more than enough to evaluate a player with a high confidence interval especially when it corroborates rather than deviates from his previous club performance

So we’ve got enough minutes how would you like to use them to evaluate him
Not xa … so what ?

My question would be why you and others are using one side of the data (xG / xA) to assert that there is enough of a sample size to be certain about something, while wilfully ignoring or disregarding the other side of the data (xG conversion) to be certain that it will turn around through better luck and variance.

That's the part that doesn't make sense to me.

[JackWard33]

I know you don't like hearing my opinion that not all minutes are created equal. So I don't expect we'll find common ground here when I say that the Europa league, league cup and fa cup minutes are of less value to me for much the same reasons as the benfica minutes.

Not worthless data, but less valuable data. As always I'm just here for the strongest most robust representation we can get. League minutes, champions league minutes. These are what I value in terms of analysis. The others are noise.

Do we still have enough minutes?

Yes .. 4000 of those minutes
How do you want to use them to evaluate him? Goals scored?

[mullyred94]

Do assists matter yet?

[Mr Dilkington]

My question would be why you and others are using one side of the data (xG / xA) to assert that there is enough of a sample size to be certain about something, while wilfully ignoring or disregarding the other side of the data (xG conversion) to be certain that it will turn around through better luck and variance.

That's the part that doesn't make sense to me.

In 17k minutes across his entire career he is on 0.60 goals from 0.57xG. Basically an average finisher so far.

[LallanaInPyjamas]

In 17k minutes across his entire career he is on 0.60 goals from 0.57xG. Basically an average finisher so far.

For Liverpool, he's 12.66 goals under what he should be in the Premier League. We've just heard that 3500 minutes is enough of a sample size to be sure about something, so I think that tells us he's considerably below average at finishing.

[Mr Dilkington]

For Liverpool, he's 12.66 goals under what he should be in the Premier League. We've just heard that 3500 minutes is enough of a sample size to be sure about something, so I think that tells us he's considerably below average at finishing.

No it doesn't. That's completely ridiculous.

Finishing is by nature noisy and subject to variance whereas underlying performance level is much more stable.

I.e. Lewandowski can put up 0.85xG one season and score 1.10 goals, then the next season put up 0.85xG and score 0.60 goals.

Whereas we wouldn't see Lewandowski putting up 0.85xG one season and then 0.5xG the season after.

[LallanaInPyjamas]

No it doesn't. That's completely ridiculous.

Finishing is by nature noisy and subject to variance whereas underlying performance level is much more stable.

I.e. Lewandowski can put up 0.85xG one season and score 1.10 goals, then the next season put up 0.85xG and score 0.60 goals.

Whereas we wouldn't see Lewandowski putting up 0.85xG one season and then 0.5xG the season after.

Do you seriously expect Nunez to outperform his xG to those levels?

If using data that tells us he should have 12 more goals than he has scored to conclude that he is below average at finishing is completely ridiculous, then I don't see the point of using this data whatsoever.

[mullyred94]

For Liverpool, he's 12.66 goals under what he should be in the Premier League. We've just heard that 3500 minutes is enough of a sample size to be sure about something, so I think that tells us he's considerably below average at finishing.

Yet two seasons prior in UCL he out basically perfomed his XG didnt he?

I mean last season he scored 4 goals out of a xg of 5 - I would say thats fine would you?

Prior season at Benfica he scored 6 goals out of an xg of 3. - Doubled his XG expectation ?

This season in Europa 5 goals from a XG of 6.

Or am I reading the stats wrong?

[JackWard33]

As an aside I don’t know where 12.66 goals is from but it’s way off… fbref has reliable numbers

[mullyred94]

As an aside I don’t know where 12.66 goals is from but it’s way off… fbref has reliable numbers

Which is were I got his european stats from.
This season in the league

11 from 15 XG

11 from 14.2 NPXG
 

[Mr Dilkington]

Do you seriously expect Nunez to outperform his xG to those levels?

He literally did it at Benfica two years ago!

Honestly you're boxing yourself into a really dumb corner with this.

How many minutes do you think you need to be sure someone is a good or bad finisher?

[mullyred94]

He literally did it at Benfica two years ago!

Honestly you're boxing yourself into a really dumb corner with this.

How many minutes do you think you need to be sure someone is a good or bad finisher?

I literally just posted his european stats for them

[Redley]

For Liverpool, he's 12.66 goals under what he should be in the Premier League. We've just heard that 3500 minutes is enough of a sample size to be sure about something, so I think that tells us he's considerably below average at finishing.

Where on earth are you getting those stats?

[LallanaInPyjamas]

He literally did it at Benfica two years ago!

Honestly you're boxing yourself into a really dumb corner with this.

How many minutes do you think you need to be sure someone is a good or bad finisher?

A noisy, variant outlier?

And I would say more than 3500 minutes personally, but hey we just heard from a numbers guy that 3500 minutes is enough to be certain about something. So if we use that logic, we can be certain that Darwin is a bad finisher.

[mullyred94]

Just incase anyone didn't know his stats are free ....


https://fbref.com/en/players/4d77b365/Darwin-Nunez

[mullyred94]

A noisy, variant outlier?

And I would say more than 3500 minutes personally, but hey we just heard from a numbers guy that 3500 minutes is enough to be certain about something. So if we use that logic, we can be certain that Darwin is a bad finisher.

Not by his UCL and Europa league stats his basically bang on isnt he?

[LallanaInPyjamas]

Where on earth are you getting those stats?

Understat. So fbref is putting him 7-8 below from what I can see? Quite different but still not even average is it.

[mullyred94]

Understat. So fbref is putting him 7-8 below from what I can see? Quite different but still not even average is it.

https://fbref.com/en/players/4d77b365/Darwin-Nunez

[LallanaInPyjamas]

https://fbref.com/en/players/4d77b365/Darwin-Nunez

If these numbers can be trusted so heavily, how can they differ so greatly? Models shouldn't be that far apart if it's something we can use with such confidence to judge a player?

[Mr Dilkington]

A noisy, variant outlier?

And I would say more than 3500 minutes personally, but hey we just heard from a numbers guy that 3500 minutes is enough to be certain about something. So if we use that logic, we can be certain that Darwin is a bad finisher.

You're better than this.

Basically what you are saying is this: Darwin Nunez is a bad finisher because in 3.5k minutes at Liverpool he is under his xG.

Response: but across his whole career he is basically on par for xG to goals, in fact slightly above and just two years ago at Benfica he massively outperformed his xG.

Yes but this is just a noisy, variant outlier. Which is why to determine the quality of his finishing I am using a 20% sample of his entire career and ignoring the other 80%.

Surely when you step back you can see the absurdity of what you're saying?

[mullyred94]

If these numbers can be trusted so heavily, how can they differ so greatly? Models shouldn't be that far apart if it's something we can use with such confidence to judge a player?

Well I'm fairly sure if you look at the video the EPL posted with the XG from Jota, Salah etc misses against Palace you'll see that they match up?

Happy to be told I'm wrong if I am remembering wrong.

EDIT: Fairly sure Jotas chance on the EPL IG account was .45 or .47 ? ( Open goal basically )

FBref has it as a .5 XG from that game for Jota.

[LallanaInPyjamas]

You're better than this.

Basically what you are saying is this: Darwin Nunez is a bad finisher because in 3.5k minutes at Liverpool he is under his xG.

Response: but across his whole career he is basically on par for xG to goals, in fact slightly above and just two years ago at Benfica he massively outperformed his xG.

Yes but this is just a noisy, variant outlier. Which is why to determine the quality of his finishing I am using a 20% sample of his entire career and ignoring the other 80%.

Surely when you step back you can see the absurdity of what you're saying?

I'm being deliberately obtuse by using another poster's logic to point out the one-sidedness of this debate, and how you and others are using numbers. You are using the positive side of predictive / theoretical stats to be certain about something, while disregarding the actual stats to reinforce that certainty.

[JackWard33]

Understat. So fbref is putting him 7-8 below from what I can see? Quite different but still not even average is it.

26.3 npxg ... 20 goals

Yes he's underperformed his expected goals in the league for us.. but we already knew that - that's the point of the whole debate

The reason conversion and xg are different measures is that chance conversion is highly volatile and not predictive in a useful time period (ie a player can have a high conversion % one year and a low one the next ...as illustrated by our long haired striking god currently being discussed) where as xg is much more stable
They're just two different measures. When you say why is one data point reliable and one isn't it's like asking why is a radar forecast for the next 2 hours more reliable compared to a 30 day forecast - they're differenent measures with different levels of predictive ability

[Mr Dilkington]

I'm being deliberately obtuse by using another poster's logic to point out the one-sidedness of this debate, and how you and others are using numbers. You are using the positive side of predictive / theoretical stats to be certain about something, while disregarding the actual stats to reinforce that certainty.

So you don't accept what I said about finishing skill relative to underlying performance? I.e. finishing is noisy and by nature prone to high variance, whereas underlying performance is more stable.

[Sheer Magnetism]

So let’s throw it out …. I mean we’ve got 7000 league minutes on Darwin over the last 4 seasons where his numbers are consistent but it might be a bad measure and/or have not matured fully as a data set so let’s bin it

Now we’ve binned xa … how would you like to evaluate and judge Darwin’s performance level?

But they AREN'T consistent - they fluctuate wildly. That's the problem! He's had one season where he massively outperformed his xG but he's been significantly below in the other three of the past four seasons. That info is on the FBef page! Neither you or Dilkington seem willing to acknowledge this.

[Mr Dilkington]

But they AREN'T consistent - they fluctuate wildly. That's the problem! He's had one season where he massively outperformed his xG but he's been significantly below in the other three of the past four seasons. That info is on the FBef page! Neither you or Dilkington seem willing to acknowledge this.

And yet still basically exactly xG to goals across his career.

[mullyred94]

But they AREN'T consistent - they fluctuate wildly. That's the problem! He's had one season where he massively outperformed his xG but he's been significantly below in the other three of the past four seasons. That info is on the FBef page! Neither you or Dilkington seem willing to acknowledge this.

significantly below is a stretch his 4 below in the EPL by fbref?

Stats stay the same but I guess we interpret them differently

[LallanaInPyjamas]

So you don't accept what I said about finishing skill relative to underlying performance? I.e. finishing is noisy and by nature prone to high variance, whereas underlying performance is more stable.

Sure, I can agree that finishing is more variant than underlying performance. But I don't think it's as variant as you are making out for high-class finishers. Of course you will find one bad season of finishing for players like Lewandowski and Messi, especially towards the latter stages of their career in lesser functioning teams.

If you take Darwin as an example, he has underperformed in every season except one. Therefore that overperformance becomes the outlier until proven otherwise. Him taking lots of shots = he will score lots of goals. We agree there. Where we don't agree is this "reversion to the mean" bollocks. Footballers aren't robots, and they're not controlled like casino games. Their skills matter.

Like I said, the absurdity of all of this is that very overperformance led the stat guys on here to question his signing. Yet now he is consistently underperforming, they are in his corner while slating Jota for being on a hot streak. There is some very skewed logic going on here - it comes across as smartarsery.

[JackWard33]

But they AREN'T consistent - they fluctuate wildly. That's the problem! He's had one season where he massively outperformed his xG but he's been significantly below in the other three of the past four seasons. That info is on the FBef page! Neither you or Dilkington seem willing to acknowledge this.

Yes.. finishing fluctuates wildly... now we're getting somewhere

Xg accumulation does not

[mullyred94]

Sure, I can agree that finishing is more variant than underlying performance. But I don't think it's as variant as you are making out for high-class finishers. Of course you will find one bad season of finishing for players like Lewandowski and Messi, especially towards the latter stages of their career in lesser functioning teams.

If you take Darwin as an example, he has underperformed in every season except one. Therefore that overperformance becomes the outlier until proven otherwise.

Like I said, the absurdity of all of this is that very overperformance led the stat guys on here to question his signing. Yet now he is consistently underperforming, they are in his corner while slating Jota for being on a hot streak. There is some very skewed logic going on here - it comes across as smartarsery.

Whos slating Jota?

Only point re Jota I've seen is that he went 32 games without scoring yet most people would say his a good finisher right?

EDIT - Infact Jota has been the best finisher for Liverpool since Klopp took over IMO.

[JackWard33]

Like I said, the absurdity of all of this is that very overperformance led the stat guys on here to question his signing. Yet now he is consistently underperforming, they are in his corner while slating Jota for being on a hot streak. There is some very skewed logic going on here - it comes across as smartarsery.

You're so so close to getting it.. it's like its on the edge of your consciousness

Also no ones slating anyone for being on a hot or cold streak... that bit was weird.. dont know a single 'data' driven guy that doesn't love Jota

[Redley]

But they AREN'T consistent - they fluctuate wildly. That's the problem! He's had one season where he massively outperformed his xG but he's been significantly below in the other three of the past four seasons. That info is on the FBef page! Neither you or Dilkington seem willing to acknowledge this.

You've said this before and then done a runner when this is actually pointed out...

significantly below is a stretch his 4 below in the EPL by fbref?

Stats stay the same but I guess we interpret them differently

[Sheer Magnetism]

And yet still basically exactly xG to goals across his career.

Okay sure, citation please. On that page, it gives info for five seasons where his xG is 73.4 and his actual goals are 68. That is not near exactly the same. If he'd hit average in that abnormal last season it would be at something like 60 against 73. His shots-to-goal total has also been similarly bad for three of those last four seasons.

[mullyred94]

Okay sure, citation please. On that page, it gives info for five seasons where his xG is 73.4 and his actual goals are 68. That is not near exactly the same. If he'd hit average in that abnormal last season it would be at something like 60 against 73. His shots-to-goal total has also been similarly bad for three of those last four seasons.

So 5 below his XG over 5 seasons is poor finishing?

[LallanaInPyjamas]

Whos slating Jota?

Only point re Jota I've seen is that he went 32 games without scoring yet most people would say his a good finisher right?

EDIT - Infact Jota has been the best finisher for Liverpool since Klopp took over IMO.

Slating may be an exaggeration, but go and look in his thread. Dilks and Jack will gleefully point out every time he misses a good chance, just because they can't accept that finishing is a skill, and Jota is the best player in our squad at that.

[Sheer Magnetism]

You've said this before and then done a runner when this is actually pointed out...

I'm sorry but I have no idea what 'significantly below is a stretch his 4 below in the EPL by fbref?' means , so it's difficult to really mount an argument against it.