The joys of ELO

[Eren]

This thread just debunks the ELO applicability for tennis. I mean, to some degree it does work. But it's just not useful to determine the so-called "BOAT/SOAT/XOAT/IOAT" and other invented acronyms by Djokovic fans.

However, without any use of ELO rating, it is still possible to make arguments that put 2014 Federer = 2004 Federer.

Everything is possible, but it wouldn't make sense. It's like arguing that 2015 Nadal = 2010 Nadal. Completely wrong.

@Towny

This is HUGE ownage lol. You completely destroyed the ELO BS propaganda. I still think ELO can be a fun thing to use, but you exposed the **** out of it and showed that anyone who used ELO as an argument don't know much about tennis.

 

[NatF]

Find the post then.

You deleted your old account

 

[TearTheRoofOff]

I didn't insult you as much as you deserved, that's for sure.

Which is also an insult.

 

[Sysyphus]

Personal insight is overrated when you have objective stats.

Tennis has no real objective performance indicators whatsoever, only stats based on a player's performance relative to another at a given point in time, and accumulations of this.

The race time of a 100m is an objective performance indicator. Or the weight on the bar in a clean and jerk. Tennis has nothing even remotely comparable to this. Tennis stats are at best like knowing the distance between the runners in a race, without knowing the actual finishing time.

So what we're left with are a bunch of data that at best only give some opaque glimpse of reality, and we're inescapably forced to extrapolate from these with a degree of personal judgment.

So when someone quite transparently cherry-picks a bunch of usually not-so-rigorous factoids (the word 'statistics' is really not warranted here), without much critical examination of what the data actually entail and their limitations, then we should not be surprised if people do not listen to them very much.

 

[Lew II]

You deleted your old account

If you can't prove I wrote what you say, don't mention it

 

[Sysyphus]

A (not so) little thought experiment to exemplify how misleading the conclusions of a rating system based on relative strength can be in the absence of objective performance indicators. Bear with me:

Suppose that the 100m would stop using the race time(s) as the metric to appraise the quality of a win or the strength of a (race) field. Suppose they instead created an Elo-like system to rate the quality of the performances and racers instead. Suppose they call it Elol. Winning a race against racers with high Elol rating is impressive; winning a race against a racer with a lower Elol rating, less so. Strong races are those were the sum-total Elol rating of the racers is high. And the more races you have been dominating, the higher your Elol rating, and so on.

So let's imagine how Usain Bolt's 100m win in the 2008 Beijing would look under these criteria. Firstly, Bolt himself hadn't really been racing the 100m for a terribly long time at this time. He had racked up a select few very impressive performances to be sure, but not a very deep CV. So we must imagine that his Elol rating wouldn't have been terribly high. And who did he beat? In places 2–4 came Richard Thompson, Walter Dix, and Churandy Martina. Not the most acclaimed sprinters in terms of their results. So judging by the Elol ratings, this race probably wouldn't have gone down in the history books as the greatest.

Now take the 100m final in the 2017 London World Champs. Justin Gatlin won, and in places 2–4 came Christian Coleman, Usain Bolt and Yohan Blake. At this point, Gatlin himself had a long CV of good results, including winning a bunch of races over the last 3 years or so. Those who followed him, like Bolt and Blake, would also presumably have excellent Elol ratings based on their numerous previous successes. We must probably assume that Bolt's Elol rating was slightly better now than it was when he won the Olympics in 2008.

It certainly seems that the Elol system would end up suggesting that the 2017 WC was the greater race of the two. The contestants were more merited. Yet, the 100m has the splendid advantage of allowing to see the actual race times – so what were they? In 2008: 1 = 9.69, 2 = 9.89, 3 = 9.91, 4 = 9.93. In 2017: 1 = 9.92, 2 = 9.94, 3 = 9.95, 4 = 9.99. So, in reality an unmerited Thompson ran a stronger race in 2008 than a legendary Bolt in 2017, and the winning time was also far better in the seemingly weaker 2008 race.

Not hard to imagine that Federer's early dominance over less merited but possibly well-playing opponents might be like the 2008 example, whereas some of the more recent "stronk elo" tennis seasons against past-it greats might be more like the 2017 example. Elo would have no real way of telling us.

 

[Lew II]

Tennis has no real objective performance indicators whatsoever, only stats based on a player's performance relative to another at a given point in time, and accumulations of this.

The race time of a 100m is an objective performance indicator. Or the weight on the bar in a clean and jerk. Tennis has nothing even remotely comparable to this. Tennis stats are at best like knowing the distance between the runners in a race, without knowing the actual finishing time.

So what we're left with are a bunch of data that at best only give some opaque glimpse of reality, and we're inescapably forced to extrapolate from these with a degree of personal judgment.

So when someone quite transparently cherry-picks a bunch of usually not-so-rigorous factoids (the word 'statistics' is really not warranted here), without much critical examination of what the data actually entail and their limitations, then we should not be surprised if people do not listen to them very much.

Of course my criterias are personal too, I was just joking.

But I'm consistent with it, unlike many here. For example you won't see me write 2018 was a great year, despite Djokovic won 2 slams.

 

[NatF]

If you can't prove I wrote what you say, don't mention it

We both remember it You've made it very difficult to search for your old posts as well - I wonder why?

 

[Tennis_Hands]

Of course my criterias are personal too, I was just joking.

But I'm consistent with it, unlike many here. For example you won't see me write 2018 was a great year, despite Djokovic won 2 slams.

We know that you are consistent with your joking Lew.

 

[Sysyphus]

Of course my criterias are personal too, I was just joking.

But I'm consistent with it, unlike many here. For example you won't see me write 2018 was a great year, despite Djokovic won 2 slams.

But cherry-pickers on the opposite side are often equally consistent.

I have seen plenty of Federer diehards who insist that 2017 was terrible even though Federer was successful, because this fits with their overall conviction that tennis was much better in the 2000s than in recent years.

And so we're at scratch, really.

 

[Lew II]

But cherry-pickers on the opposite side are often equally consistent.

I have seen plenty of Federer diehards who insist that 2017 was terrible even though Federer was successful, because this fits with their overall conviction that tennis was much better in the 2000s than in recent years.

And so we're at scratch, really.

So tennis evolved until Federer. Is that consistent to you? They also often write 2007-13 was high level because top players had good results but can't see 2014-16 was just as good too in that department.

 

[Sysyphus]

So tennis evolved until Federer. Is that consistent to you?

Plenty of sports with actual objective measures have world records that are at least a decade old, where the sport evolved with slight peaks and valleys up to that point and with no one having surpassed that level of performance since.

Nothing inherently inconsistent in thinking that this could be the case in tennis as well. Though I personally don't consider the evidence for this to be terribly overwhelming, just like the evidence you present for your views isn't. The different weak/stronk era theorists are ultimately on similar ground in this regard.

 

[ghostofMecir]

It can't be real. If it is, I agree it's simply wrong.

@Lew II ‘s world crashing down around him...

 

[ghostofMecir]

Elo seems to be one of the more popular ratings systems used on this board, especially by certain Novak apologists. It has been used to determine not only peak level but also the level of competition faced by a given player. While I'm not saying the system is total garbage, I thought it'd be interesting to look at a few anecdotal findings of the system to show just how inherently flawed it is. I'll be looking at the Big 3 for this. All data from Ultimatetennisstatistics.com

Federer
Feb 2nd 2004 - 2303
Feb 3rd 2014 - 2321
Immediately following his respective AOs, Fed of 2004 held 2 slams and TMC and was world number 1. Fed of 2014 had just had his awful 2013 season and was now world number 8. 2014 Fed was considered better. For year end Elo ratings, let's compare 2004 to 2014 (3 slams, YEC and 3 masters compared to 0 slams, 0 YEC and 2 masters)

Nov 22nd 2004 - 2420
Nov 16th 2014 -2408
Apparently fairly comparable years.

Nadal
Jan 31st 2011 - 2379
Feb 2nd 2015 - 2383
In the former, Nadal had just won 3 slams and was world number 1. In the latter, Nadal had just one slam to his name (RG) and was world number 3 after being beaten by Berdych at AO. Let's compare the Nadal that Berdych faced in 2015 to the Nadal that Djokovic faced in 2012:

Jan 16th 2012 - 2377
Jan 19th 2015 - 2397
Apparently, Berdych faced the stronger Nadal

Djokovic
September 12th 2011 - 2522
January 30th 2017 - 2516
One of them had just won 3 slams and 5 masters. The other had 1 slam and 4 masters to his name, and had just been bounced by Istomin. In fact, prior to facing Istomin, on January 16th 2017, Novak's rating was 2538, better than it had ever been in 2011

For further comparison, Novak at his lowest Elo in 2012 (August 12th) had 2431
Novak following his garbage 2017 had 2417 (Dec 1st 2017), only 14 points less

There's no real method to my madness but it's simply to illustrate that totting up 'list of players with ???? Elo points or more beaten' is a waste of time, as the system sometimes can't even tell the difference between beating 2013/4 Federer and beating 2004 Federer. But then, what can you expect from a system designed to be used in Chess?

Discuss

Quoted for posterity.

 

[Tennis_Hands]

@Lew II ‘s world crashing down around him...

It cannot crush down based on the realisation that his ELO comparisons are bogus, if he knows about that in advance, hence the cherry-picking, can it?

Otherwise, when applying a consistent approach to the ELO stats, he would have noticed the problem, no?

 

[ghostofMecir]

Quoted for posterity.

Leaked footage of the OP as he found out these ELO numbers...

 

[ak24alive]

Quoted for posterity.

Leaked footage of the OP as he found out these ELO numbers...

@Lew II ‘s world crashing down around him...

Lol I laughed so hard myself at the "it can't be real". It's simply funny. The cute thing is that Lew himself was shocked. He didn't know that Elo had so many inconsistencies. I like his honesty in admitting that it has to be wrong. His ways might be wrong but it seems to me that he really wants to find the truth although being so sure about numbers doesn't help.
Tennis has so many variables that numbers rarely tell the full picture.

 

[Tennis_Hands]

Lol I laughed so hard myself at the "it can't be real". It's simply funny. The cute thing is that Lew himself was shocked. He didn't know that Elo had so many inconsistencies. I like his honesty in admitting that it has to be wrong. His ways might be wrong but it seems to me that he really wants to find the truth although being so sure about numbers doesn't help.
Tennis has so many variables that numbers rarely tell the full picture.

Pretty sure that that was not a genuine reaction, but a reaction forced by the situation, because he just couldn't say anything else (which is corroborated by his repeated insistence that there is a mistake, despite of being pretty easy to find out by himself).

Like I said, if he was anywhere as consistent with handling stats as his cherry-picking suggests, he would have noticed the problem a long time ago.

 

[ak24alive]

Pretty sure that that was not a genuine reaction, but a reaction forced by the situation, because he just couldn't say anything else (which is corroborated by his repeated insistence that there is a mistake, despite of being pretty easy to find out by himself).

Like I said, if he was anywhere as consistent with handling stats as his cherry-picking suggests, he would have noticed the problem a long time ago.

You might very well be right. It definitely is a possibility. But I would like to believe that it was an honest reaction. It leaves more room for discussion. But it's totally understandable why you wouldn't believe him. I have been in similar situations when I assumed the worst of people and I was right too but because Lew doesn't bother me I am willing to assume that he is being honest.

 

[TMF]

Anyone who ever use ELO in a debate/argument from other threads you can just give him/her a link of this thread.

 

[Lew II]

I honestly didn't know well how ELO works, but something that puts 2003-07 so low and 2011-16 so high was too tempting

 

[Tornes]

You must be trolling. Never seen Lew post anything insightful or analytical, he simply posts numbers he's filtered from ultimatetennisstatistics...In fact I've never seen him comment on actual match play. I've seen him compare Federer's defensive abilities to Isner and Karlovic before - for sure he's watched a lot of the game

Personal insight is overrated when you have objective stats.

Never compared Federer's defence to Isner's. I just wrote that the least defensive players are, the highest winner/error ratio they will have, posting Isner-Karlovic as an example.

The stats are the stats, your interpretation of them is not objective.

You compared hitting winners past Federer to hitting them past Isner/Karlovic

Find the post then.

You deleted your old account

If you can't prove I wrote what you say, don't mention it

I can confirm he really wrote it. And that he "forgot" it and use his deleted account as an excuse is just another proof of his "honesty" and "integrity".

It is hard to discuss anything with poster who is not only seriously biased but also don't remember what he wrote...

 

[Sysyphus]

I honestly didn't know well how ELO works, but something that puts 2003-07 so low and 2011-16 so high was too tempting

this is great

 

[Lew II]

I can confirm he really wrote it. And that he "forgot" it and use his deleted account as an excuse is just another proof of his "honesty" and "integrity".

It is hard to discuss anything with poster who is not only seriously biased but also don't remember what he wrote...

Not true, I didn't write it.

People don't believe him, he is known to be one of my haters.

 

[Tennis_Hands]

Not true, I didn't write it.

People don't believe him, he is known to be one of my haters.

At this point it is very probable that someone who you call a "hater" is a reasonable person that doesn't let your BS fly.

I mean, you just admitted that you trolled the whole forum for days on end with your nonsense.

Are the people that objected to that also "haters"?

 

[Lew II]

At this point it is very probable that someone who you call a "hater" is a reasonable person that doesn't let your BS fly.

I mean, you just admitted that you trolled the whole forum for days on end with your nonsense.

Are the people that objected to that also "haters"?

Find the post or gtfo. Don't put words in my mouth, it's very rude.

 

[Tennis_Hands]

Find the post or gtfo. Don't put words in my mouth, it's very rude.

When was the last time you posted with your old account?

 

[tennisaddict]

Everything that puts 2003-07 low and 2011-16 high is correct to me.

Now, since I don't trust ELO anymore, let's move on. Take a look at this:

https://tt.tennis-warehouse.com/index.php?threads/goat-points-by-year.634341/

Did I see an apology ?

 

[NatF]

Not true, I didn't write it.

People don't believe him, he is known to be one of my haters.

I'm afraid you did write it. It was part of the exchange where you tried to prove Federer's defensive game was way poorer than Djokovic and Murray's by posting their career return game stats - I then pointed out that Federer in 2003 won more return games on grass than 2013 Djok and Murray. I'm not sure if you responded much after that...

 

[Lew II]

Plenty of sports with actual objective measures have world records that are at least a decade old, where the sport evolved with slight peaks and valleys up to that point and with no one having surpassed that level of performance since.

Nothing inherently inconsistent in thinking that this could be the case in tennis as well. Though I personally don't consider the evidence for this to be terribly overwhelming, just like the evidence you present for your views isn't. The different weak/stronk era theorists are ultimately on similar ground in this regard.

In tennis you can't prove evolution stopped as in athletics, so to be consistent you have to believe it followed the normal line of evolution.

Also, if Federer's generation was so good it wouldn't have had a terrible drop in results in their mid-late 20s. Bolt before retiring was still the fastest man alive. If there's a case of evolution stopping, it is for the Big4 who are still the best in the world in their 30s.

 

[Lew II]

I'm afraid you did write it. It was part of the exchange where you tried to prove Federer's defensive game was way poorer than Djokovic and Murray's by posting their career return game stats - I then pointed out that Federer in 2003 won more return games on grass than 2013 Djok and Murray. I'm not sure if you responded much after that...

That's true. Not true that I compared his defensive game to Isner's and Karlovic's.

But who cares about the past? What I write now, remember it, is that winner-unforced ratio alone is useless, especially when comparing matches of different players.

 

[Tennis_Hands]

Find the post or gtfo. Don't put words in my mouth, it's very rude.

This the thread, boy?

https://tt.tennis-warehouse.com/ind...-federer-in-grass.619156/page-3#post-12393364

 

[Lew II]

This the thread, boy?

https://tt.tennis-warehouse.com/ind...-federer-in-grass.619156/page-3#post-12393364

It is, thanks.

 

[NatF]

That's true. Not true that I compared his defensive game to Isner's and Karlovic's.

But who cares about the past? What I write now, remember it, is that winner-unforced ratio alone is useless, especially when comparing matches of different players.

You implied it with your posts. We were talking about Philippoussis' great numbers W/UE and you commented that Isner/Karlovic would have great numbers against each other...

Match stats (not just winners/errors) are more useful than the factoids you produce.

 

[NatF]

This the thread, boy?

https://tt.tennis-warehouse.com/ind...-federer-in-grass.619156/page-3#post-12393364

Not sure if that's it. I can't see the post with the return game stats?

 

[Lew II]

You implied it with your posts. We were talking about Philippoussis' great numbers W/UE and you commented that Isner/Karlovic would have great numbers against each other...

Match stats (not just winners/errors) are more useful than the factoids you produce.

I think you don't understand what I wrote. Can we now move on from old posts?

God now I want to delete my account again...

 

[Tennis_Hands]

I think you don't understand what I wrote. Can we now move on from old posts?

God now I want to delete my account again...

I will find posts even if you delete this one as well.

Certainly.

 

[NoleFam]

Elo is all about exchange of points between opponents. It doesn't even penalize a player for missing a huge chunk of the season.
For Example Nole's Elo just after Wimbley 2017 was 2421.
He didn't play the next half of the season after that.
His Elo at year end 2017 was 2417.

Lololol the guy was penalized 4 Elo points for missing so much ****.
Reason: He didn't lose to any players in this time.
His ATP ranking fell down a lot in that time.
His Elo rating however didn't.
Which is why Elo doesn't go hand in hand with the ATP rankings mate.
If you want to know how Nole had such high Elo when he was no. 22 you should check Elo week by week and you will see that how less he was penalized for losing to lesser players in the first half of 2018. That's the flaw of Elo. It doesn't adjust well.
Also it doesn't tell you the level of a particular player. It is an indicator of a player's competition but as you can see here it doesn't work well even there.
Dude it's just a mathematical model. Believe me there are so many variables in tennis that these models can't incorporate. It isn't utter **** I will give you that but it isn't perfect either.

I believe this is actually a glitch on the site because no way was Murray and Djokovic #3 and #4 in Elo rankings on 4/30/18. However, this is what it shows if you go to that date and have 'Elo Ranking' selected. However, if you change that column to 'Recent Elo Ranking', Djokovic was #8 and Murray was not in the top 170. So I think you are thinking that is a legit list of current rankings and it's not. It has to be an explanation for what it is but it can't possibly be recent. Elo has flaws but is not off by margins that huge.

 

[NoleFam]

So many people in here are crapping on Elo and don't even understand. You can't crap on something that is the official measure for multiple sports and been around forever. I mean you can try but you can't really succeed. Lew likes pulling you guys' chains and getting you riled up, and you basically fall for it everytime, but trying to invalidate Elo to in turn invalidate Lew falls short. Elo has flaws as far as tennis but so many of you in here are so wrong about what you are talking about. You don't even know that what you are posting is not even the way to interpret it.

 

[Lew II]

I knew something was wrong.

 

[Eren]

I believe this is actually a glitch on the site because no way was Murray and Djokovic #3 and #4 in Elo rankings on 4/30/18. However, this is what it shows if you go to that date and have 'Elo Ranking' selected. However, if you change that column to 'Recent Elo Ranking', Djokovic was #8 and Murray was not in the top 170. So I think you are thinking that is a legit list of current rankings and it's not. It has to be an explanation for what it is but it can't possibly be recent. Elo has flaws but is not off by margins that huge.

Nadal
Jan 31st 2011 - 2379
Feb 2nd 2015 - 2383
In the former, Nadal had just won 3 slams and was world number 1. In the latter, Nadal had just one slam to his name (RG) and was world number 3 after being beaten by Berdych at AO. Let's compare the Nadal that Berdych faced in 2015 to the Nadal that Djokovic faced in 2012:

Jan 16th 2012 - 2377
Jan 19th 2015 - 2397
Apparently, Berdych faced the stronger Nadal.

What about this part? IF all of it is a glitch that website is not reliable and should never be used anyway.

 

[Lew II]

I will find posts even if you delete this one as well.

Certainly.

Sure, you're obsessed.

I don't give a damn about your posts instead. All I can remember is the smiley LOL

 

[Tennis_Hands]

Sure, you're obsessed.

I don't give a damn about your posts instead. All I can remember is the smiley LOL

I am sure that you will remember a couple more things as a result of our interactions besides that smiley.

Oh, you will!

 

[NoleFam]

What about this part? IF all of it is a glitch that website is not reliable and should never be used anyway.

It has to be an explanation. The guy that runs the site is very thorough and on his game. We will have to ask him what it means because I am not even sure, however, I am sure how to get an accurate recent Elo ranking/rating.

Jan. 31st, 2011 - Nadal 2456
Feb. 2nd, 2015 - Nadal 2365

Jan. 16th, 2012 - Nadal 2446
Jan 19th, 2015 - Nadal 2380

What you quoted was incorrect. I gave you an accurate reading for those specific dates.

 

[Eren]

It has to be an explanation. The guy that runs the site is very thorough and on his game. We will have to ask him what it means because I am not even sure, however, I am sure how to get an accurate recent Elo ranking/rating.

Jan. 31st, 2011 - Nadal 2456
Feb. 2nd, 2015 - Nadal 2365

Jan. 16th, 2012 - Nadal 2446
Jan 19th, 2015 - Nadal 2380

What you quoted was incorrect. I gave you an accurate reading for those specific dates.

The difference still seems small tbh. But it's more accurate with regards to reality than in the OP.

 

[NoleFam]

The difference still seems small tbh. But it's more accurate with regards to reality than in the OP.

Well it's about 90 points between 2011 and 2015, which in Elo, is quite a lot. This says Nadal was much better in Jan 2011 than he was in Feb 2015. The OP is mistaken and many others in this thread are as well.

 

[Tornes]

It

I believe this is actually a glitch on the site because no way was Murray and Djokovic #3 and #4 in Elo rankings on 4/30/18. However, this is what it shows if you go to that date and have 'Elo Ranking' selected. However, if you change that column to 'Recent Elo Ranking', Djokovic was #8 and Murray was not in the top 170. So I think you are thinking that is a legit list of current rankings and it's not. It has to be an explanation for what it is but it can't possibly be recent. Elo has flaws but is not off by margins that huge.

It is no glitch, it is just how it works.

Recent ELO is something different than normal ELO. I gives way much more emphasis on recent matches than any normal ELO system.

It has to be an explanation. The guy that runs the site is very thorough and on his game. We will have to ask him what it means because I am not even sure, however, I am sure how to get an accurate recent Elo ranking/rating.

Jan. 31st, 2011 - Nadal 2456
Feb. 2nd, 2015 - Nadal 2365

Jan. 16th, 2012 - Nadal 2446
Jan 19th, 2015 - Nadal 2380

What you quoted was incorrect. I gave you an accurate reading for those specific dates.

No, he is correct. You quated "recent ELO", he quated ELO. 2 different systems.

 

[NoleFam]

It

It is no glitch, it is just how it works.

Recent ELO is something different than normal ELO. I gives way much more emphasis on recent matches than any normal ELO system.



No, he is correct. You quated "recent ELO", he quated ELO. 2 different systems.

Well I am sure the guy that runs the site can explain what it means and what his formula is, because until then we don't know.

No he is not correct. He gave specific dates meaning he was looking for a reading at that time frame and he quoted the wrong rating. Elo is supposed to give emphasis on recent matches and that is the way it is done.

 

[TheGhostOfAgassi]

this is also ELO

@NatF I dedicate this song to you. So you can think about the woman in this forum you think is the devil incarnated: me.

 

[Towny]

Well it's about 90 points between 2011 and 2015, which in Elo, is quite a lot. This says Nadal was much better in Jan 2011 than he was in Feb 2015. The OP is mistaken and many others in this thread are as well.

Recent Elo and Elo are not the same. No one has been using recent Elo, they have been using Elo, which is why I used those figures. I'm not sure exactly how recent Elo is calculated, but it seems to be better at gauging level than normal Elo. However, it still has some egregious issues:

Federer
Oct 1st 2004 2538
Feb 21st 2011 2551
Nov 16th 2014 2548
Fed was slamless in the latter 2 and yet was considered a stronger opponent than his 2004 self who had recently won the USO

Nadal
Jan 16th 2012 2446
May 23rd 2016 2446
I think we can clearly see Nadal was a far stronger opponent in 2012 than 2016

Djokovic
Feb 4th 2008 2428
Jul 3rd 2017 2450
Djokovic had just won the AO and previously made USO final. Djokovic 2017 was slamless at this point. Apparently, he was tougher

The point is that these stats can't be blindly quoted as their are too many inconsistencies. I'm not saying that Elo is necessarily a total waste of time, it's just that it's clearly not completely objectively.