ELO Weaklings of the Weak Era & Delpo 10th!

Meles

Meles

Bionic Poster
[The following is based on Jeff Slackman's current ELO ratings. See footnote for more technical details*:
http://tennisabstract.com/reports/atp_elo_ratings.html ]

The ELO rankings list a peak rating and the time of that rating. The following list has been filtered down for the weaklings in the current ELO ranking system. If a player achieved their peak rank in 2015 and 2016 they've been filtered from this list. If their current rating was within 98% of peak they have been filtered from this list. Some of the omissions like 1 and 2 are worth commenting on, as ELO rates Murray and Djokovic as at their peak performance this year. Fedal definitely down from peak. Brief comments after some of the players.
2016ELOfilter.png


This is a first look and more ELO threads to follow. I'm surprised at Cilic's decline. Its surprising that his 2014 US Open win and play after that event did not beat his 2010 ELO rating. Cilic has been losing a lot of matches in 2016 and had minor injury issue (of course he looked awesome at Wimbledon.)

Del Potro is the cool one in the ELO system. He was ranked just 36 points behind Wawrinka coming into Wimbledon so its little surprise that he won their match. ELO says Delpo is back!

Is it a weak era? Well, you really can't quite tell from this.

*ELO rating is named after its creator Arpad Elo, a Hungarian-born American physics professor, who designed the system to rate Chess players. Its been adapted for tennis and the rating is very much like a ranking, but a player whose rating is 100 points greater than their opponent's is expected to win 64% of the time; if the difference is 200 points, then the expected score for the stronger player wins 76% of the time. Lets keep it simple and leave ELO as a superior ranking system based on matchups. The numbers referenced are not surface specific.
 
Last edited:
Meles said:
I'm surprised at Cilic's decline. Its surprising that his 2014 US Open win and play after that event did not beat his 2010 ELO rating. Cilic has been losing a lot of matches in 2016 and had minor injury issue (of course he looked awesome at Wimbledon.)
Click to expand...
Shoulder injury that forced him out of the game for more than 2 months.
 
Before commenting further, this from the site you linked to:
Unlike the official rankings, Elo ratings give credit for who you play, not the round or tournament in which you play them.
Click to expand...
This could be a weakness. An example that immediately reminds me of this:

In 2015 Federer lost to Novak in the WTF, but not in a match that was critical. Then Novak won when the match was critical and thus won the WTF.

Now, if all matches in a H2H are measured, I would expect that such events would eventually be overwhelmed by the general data.

It's a bit like people who say that % of games won will be boosted for players who don't make it far in tournaments, or that any such stats will be boosted in this way. In the end the best of the best keep winning those stats.

But this did catch my eye.
 
Let's say that Novak's rating of 2565.1 is currently the gold standard. He is just a few points shy of a 200 lead over Murray. That should give him something like a 75% chance of winning against Murray.

Questions:

What would the peak be yearly so that we can compare? I think the data you showed means that Federer was very high in 2007 - 2524.3. Do absolute values from different years mean anything? Or just relative values at a time when different players are facing either other?

What about surface? (Nadal's elo during his reign on clay should have been unbelievable.)

How often per year are these numbers run? I'd love to see a running record of Fed, Nadal and Novak going back a few years.

Can elo be computed retroactively?

How does elo go up and down when competition is stronger at the top? Stronger for the top 5, top 10, top 20?
 
Gary Duane said:
Before commenting further, this from the site you linked to:

This could be a weakness. An example that immediately reminds me of this:

In 2015 Federer lost to Novak in the WTF, but not in a match that was critical. Then Novak won when the match was critical and thus won the WTF.

Now, if all matches in a H2H are measured, I would expect that such events would eventually be overwhelmed by the general data.

It's a bit like people who say that % of games won will be boosted for players who don't make it far in tournaments, or that any such stats will be boosted in this way. In the end the best of the best keep winning those stats.

But this did catch my eye.
Click to expand...
I've updated the OP with image of spreadsheet data for better readability as your requested.

This waits all matches the same. Its good to understand how its done. Djokovic saves his best for majors, so perhaps his ELO is low when being used for determining if how strong he would be in major and high for other times. Federer could currently be the reverse as some of his fans seem to think he's not handling the biggest matches as well. Another example is Thiem who actually has an ELO of 7 and Zverev at 20. They both have wins over Federer while he was in questionable form, so there current ELOs may be overstated.

I really like the general feel I've seen from the ELO numbers so far. Nobody is currently putting out surface specific ELO ratings, but there are some from the end of 2012 and as you and I know the hard court and clay court numbers have a chance of being usable (grass and carpet probably shakey.)
 
TupeloDanger said:
ELO breaks in a shallow talent pool with top heavy dominance, making it a useless tool for cross-era comparisons. It is known.
Click to expand...
I've not run into that. Its not surface specific, so it definitely has limitations. The cross era numbers I've seen don't look far off the mark. Of course comparing players like Federer to Laver is ultimately a foolish exercise if you are trying to say which player would win over the other.
 
Meles said:
I've updated the OP with image of spreadsheet data for better readability as your requested.

This waits all matches the same. Its good to understand how its done. Djokovic saves his best for majors, so perhaps his ELO is low when being used for determining if how strong he would be in major and high for other times. Federer could currently be the reverse as some of his fans seem to think he's not handling the biggest matches as well. Another example is Thiem who actually has an ELO of 7 and Zverev at 20. They both have wins over Federer while he was in questionable form, so there current ELOs may be overstated.

I really like the general feel I've seen from the ELO numbers so far. Nobody is currently putting out surface specific ELO ratings, but there are some from the end of 2012 and as you and I know the hard court and clay court numbers have a chance of being usable (grass and carpet probably shakey.)
Click to expand...
The elo numbers rank people about the same as my own statistics when we use players from the last 21 years. Nadal right at the top on clay, Nadal and Borg on top. That matches results I have from slams. Rating HCs and grass together for both would probably put Borg on top. I would expect Borg to be right at the top of all players taking into consideration those two surfaces.

Sampras falls down on all surfaces because of his weakness on clay. If you measure elo for Novak from 2011 to now, most likely he will top everyone on HCs.

Clutch can't be measured. There is still the Sampras-factor, the ability to raise the game for a tournament or slam.

Intuitively I think peak Agassi on HCs is way too low.

I think elo is something to be thrown into the mix. It is interesting to be able to compare between eras.

The biggest weakness for me is the inability to gauge strength of competition. That remains a subjective thing and is the weakness of all stats comparing eras.
 
Meles said:
[The following is based on Jeff Slackman's current ELO ratings. See footnote for more technical details*:
http://tennisabstract.com/reports/atp_elo_ratings.html ]

The ELO rankings list a peak rating and the time of that rating. The following list has been filtered down for the weaklings in the current ELO ranking system. If a player achieved their peak rank in 2015 and 2016 they've been filtered from this list. If their current rating was within 98% of peak they have been filtered from this list. Some of the omissions like 1 and 2 are worth commenting on, as ELO rates Murray and Djokovic as at their peak performance this year. Fedal definitely down from peak. Brief comments after some of the players.
2016ELOfilter.png


This is a first look and more ELO threads to follow. I'm surprised at Cilic's decline. Its surprising that his 2014 US Open win and play after that event did not beat his 2010 ELO rating. Cilic has been losing a lot of matches in 2016 and had minor injury issue (of course he looked awesome at Wimbledon.)

Del Potro is the cool one in the ELO system. He was ranked just 36 points behind Wawrinka coming into Wimbledon so its little surprise that he won their match. ELO says Delpo is back!

Is it a weak era? Well, you really can't quite tell from this.

*ELO rating is named after its creator Arpad Elo, a Hungarian-born American physics professor, who designed the system to rate Chess players. Its been adapted for tennis and the rating is very much like a ranking, but a player whose rating is 100 points greater than their opponent's is expected to win 64% of the time; if the difference is 200 points, then the expected score for the stronger player wins 76% of the time. Lets keep it simple and leave ELO as a superior ranking system based on matchups. The numbers referenced are not surface specific.
Click to expand...
Much better.

There should be someone close to 100% each year. That would be close for Djokovic and Murray. But one of the lower ranked players should be peaking right now, guys ranked lower and not on this list...
 
Gary Duane said:
Let's say that Novak's rating of 2565.1 is currently the gold standard. He is just a few points shy of a 200 lead over Murray. That should give him something like a 75% chance of winning against Murray.

Questions:

What would the peak be yearly so that we can compare? I think the data you showed means that Federer was very high in 2007 - 2524.3. Do absolute values from different years mean anything? Or just relative values at a time when different players are facing either other?

What about surface? (Nadal's elo during his reign on clay should have been unbelievable.)

How often per year are these numbers run? I'd love to see a running record of Fed, Nadal and Novak going back a few years.

Can elo be computed retroactively?

How does elo go up and down when competition is stronger at the top? Stronger for the top 5, top 10, top 20?
Click to expand...
These are all great questions. I just started looking at the numbers and rather try to put up some overly complicated numbers I elected to do this post and see how it goes, TTW reactions, and just left myself digest it more.

Your first sentence is correct. If you factor in Djokovic's ability to up his level at slams and Murray's flop at the French, then maybe we are talking a 300 point difference. This pretty much reflects the general sentiment on TTW though I'm afraid many would prefer Djokovic to be rated much lower and Murray too. Again these aren't surface specific so it is what it is.

Their are surface numbers for clay, grass, carpet, and hard courts through the end of 2012. Those are the fun ones as they have like the top 200 of the open era.

Lots of stuff here Gary from Slackman:
https://github.com/JeffSackmann/tennis_atp

CSV files can easily be taken into excel. He has all the match results since 1969 in CSV form.:p I'm a little dissapointed, but I don't seem to see historical ELO ratings on his github.

Again this thread is a foray into ELO and I've not decided if I'd like to do more.

I would like to take the clay and hard court peak ratings and then sort of order those somehow so we might see the strength near the top of the field (top 10). Again its only peak ratings. In theory it would be possible to generate these rating coming into every slam and spit out some top contenders. I'd sure like to avoid having to do these calculations from the raw csv files; that would take a lot of doing. Its enough work to pull these peak numbers and I'm not sure that anything remotely useful will come of it. If Kuerten is peak in 1998 on clay, that rating is going to be useless for rating the field at the French in 2002. Its a very crude thing. I've enjoyed seeing the strength and depth of some of the clay players in early 2000. JC Ferrero was around 7th in this list which is where I would rate him on points, but quite a few others are quite strong. For me this is like a 4th tool to go with points stats, games stats, and DR type stats. I thinks its a very strong tool and very simple; just one number.:D
 
Gary Duane said:
Much better.

There should be someone close to 100% each year. That would be close for Djokovic and Murray. But one of the lower ranked players should be peaking right now, guys ranked lower and not on this list...
Click to expand...
I filtered for the weaklings.;) The missing rankings spots are all players near peak or peaking. Murray, Ninja, Djoko, Thiem, Raonic, Goffin, Gasquet, and even Wawa etc. all around peak level. Anyone who was peak in 2015 and 2016 was removed from my list. Anyone within 98% of peak was removed from the list. The list is just the weak era weaklings though I"m not so sure this is a horribly weak era except that 3 and 4 are down from their peaks.

There is another site that that has another type of ELO ranking and it covers the last couples years. Sadly the numbers aren't on the same scale. I posted this in another thread and here are the top ratings of recent times:
BFRDqHN.png

The numbers here are very close to the peak numbers in the other graphic. I don't trust the number on Laver in this one. FFS Roddick is higher than Laver.:D Again this is not surface specific. Sampras is at 9, but perhaps if you realize that this includes clay numbers then his rating is about right. The Laver 1970 number could be right. He won all the slams, but was not dominate outside of that. Of course this doesn't go back any further than 1969, so maybe 31 year old Laver is about right. Some scholars might argue that this list favors recent players as the current string technology seems to allow most of the top players to play reasonably well on all surfaces. At least we have simple numbers here and we can say Murray is overrated lop 50 points off and 50 to Laver, etc. Its one number which is cool; albeit a very complicated number that should be looked at as a black box by the user (lots of luck hand calculating.o_O)
 
  • Like
Reactions: gn
tennisaddict said:
You don't need ELO to decipher this is a weak era.
Click to expand...
LOL. Well ELO puts a number on Nadal and Federer and they are definitely weaker. Its pretty cool that it ranks them still as 3 and 4 despite some missed play. Feder is 9th in the ATP points race and Nadal is 4th. ELO still puts Federer over Nadal which is likely correct except of course on clay.

Delpo is 10th, but way down from his peak rating still. You gotta like that TA.;) LOL. It had him about even with Stanimal and look who won.:D
 
NatF said:
Well in the tennisabstract description it says, "It awards points based on who you beat, not when you beat them." Which is BS.
Click to expand...
Well it has issues, but its better than rankings in many ways, so it has promise. I'd guess that it under rates peak Fed and Djoko. It may over rate Murray for instance. Murray's peak number was 2009; was that really his peak playing? I think its a very interesting scale when you know how to factor things. Its rating Delpo highly and we'll see if he lives up to his rating in the next few months. Its rating Fed at 3 which is probably about right. It certainly adjusts much more quickly than rankings for players returning from long injury layoffs. I'm not ready to judge it yet. I searchedd TTW and it looks like its come up before, but not much discussion.
 
Meles said:
Well it has issues, but its better than rankings in many ways, so it has promise. I'd guess that it under rates peak Fed and Djoko. It may over rate Murray for instance. Murray's peak number was 2009; was that really his peak playing? I think its a very interesting scale when you know how to factor things. Its rating Delpo highly and we'll see if he lives up to his rating in the next few months. Its rating Fed at 3 which is probably about right. It certainly adjusts much more quickly than rankings for players returning from long injury layoffs. I'm not ready to judge it yet. I searchedd TTW and it looks like its come up before, but not much discussion.
Click to expand...

It has merits but it seems to me it's best at rating how players stack up within a small time frame, across several years and decades it seems like you would get inflation where you get points for beating the reputation.
 
NatF said:
Well in the tennisabstract description it says, "It awards points based on who you beat, not when you beat them." Which is BS.
Click to expand...
People make the same point about % of games, but over a long period of time it seems to still show the best of the best when you are looking at the top players and guys who make it to around 60% of higher.

There seems to be a certain self-correcting effect. Otherwise players who are playing lower ranked players and winning matches that aren't as important would be way up, but they aren't.

I have a bit of an intuitive approach.

To me the best players of the last 25 years in terms of careers are:

Sampras/Agassi (mostly 90s)

During that period a lot of players had interesting short peaks on various surfaces, so the 90s was complicated, mostly off grass though.

Federer/Nadal/Djokovic (after 2003)

Nadal ruled clay, Federer mostly ruled grass but more competitive on grass, and complicated on HCs.

Murray has been a lot like Agassi, but with less winning success.

So I look for those names to stand out in stats.

Before 1991 it gets hard because we are missing the stats.

Individual years with elo seems a bit off to me. It does not seem to pinpoint most dominant years very well. I'm doubtful about highest elo ever for Novak this year meaning much. It's a dominant year, but I do believe the competition in 2016 has been a bit weak on HCs. Not terribly so. Not "weak era". But certainly not what it was in perhaps 2011.
 
Djokovic with the highest ELO rating of all time? This is blasphemous! A travesty! Burn it! :D In all seriousness, there will always be flaws with a system like this which is reflected by how low Laver is ranked. I still found it interesting though. I see they ranked Federer 2007 the highest peak on hardcourt when I would say he was better in 2005 probably level wise. However, AO 2007 is arguably his peak in any one tournament on the surface so that is kind of in sync. Based on this though, Djokovic hit his peak in Miami of this year. I was most surprised to see Djokovic ranked at #3 all time on clay behind Nadal and Borg even before he won RG and ahead of Lendl, Wilander and Guga. Nadal has an astronomical dominance on clay reflected in the chart and that is not surprising. But where the hell is the grass chart? That's the main one everyone wants to see.
 
Last edited:
Elo is a nice tool to go more in depth to a team/player's performance. From what I know, its less accurate the more you go back so its probably not the best tool to analyse Laver/Rosewall etc. If people are confused about Murray's peak being the Monte Carlo Masters QF 2009; that's not reflective of an entire year (of course). If I recall correctly, elo would also take into account the strength of players he had played recently at that time, so if a player(s) with a strong elo at that time lost to Murray he would have a higher rating coming out of that match and you guys already know how players' performances can vary day to day anyway.

This stuff isn't my strong suit and I only followed elo in the nba earlier this year, feel free to call me an idiot if I'm wrong.
 
Gary Duane said:
People make the same point about % of games, but over a long period of time it seems to still show the best of the best when you are looking at the top players and guys who make it to around 60% of higher.

There seems to be a certain self-correcting effect. Otherwise players who are playing lower ranked players and winning matches that aren't as important would be way up, but they aren't.

I have a bit of an intuitive approach.

To me the best players of the last 25 years in terms of careers are:

Sampras/Agassi (mostly 90s)

During that period a lot of players had interesting short peaks on various surfaces, so the 90s was complicated, mostly off grass though.

Federer/Nadal/Djokovic (after 2003)

Nadal ruled clay, Federer mostly ruled grass but more competitive on grass, and complicated on HCs.

Murray has been a lot like Agassi, but with less winning success.

So I look for those names to stand out in stats.

Before 1991 it gets hard because we are missing the stats.

Individual years with elo seems a bit off to me. It does not seem to pinpoint most dominant years very well. I'm doubtful about highest elo ever for Novak this year meaning much. It's a dominant year, but I do believe the competition in 2016 has been a bit weak on HCs. Not terribly so. Not "weak era". But certainly not what it was in perhaps 2011.
Click to expand...
Its the best I've seen Novak plays so it doesn't surprise me its his peak ELO. 50 points is not much of a difference with Federer. I don't think you can take that to the GOAT bank. It means they are both close at their peak. Its really pointless to try to go beyond that. If your agenda is to say Djokovic is horrible and the rest of the tour is even worse then ELO is a stick in the eye.
laugh_above.gif
 
NoleFam said:
Djokovic with the highest ELO rating of all time? This is blasphemous! A travesty! Burn it! :D In all seriousness, there will always be flaws with a system like this which is reflected by how low Laver is ranked. I still found it interesting though. I see they ranked Federer 2007 the highest peak on hardcourt when I would say he was better in 2005 probably level wise. However, AO 2007 is arguably his peak in any one tournament on the surface so that is kind of in sync. Based on this though, Djokovic hit his peak in Miami of this year. I was most surprised to see Djokovic ranked at #3 all time on clay behind Nadal and Borg even before he won RG and ahead of Lendl, Wilander and Guga. Nadal has an astronomical dominance on clay reflected in the chart and that is not surprising. But where the hell is the grass chart? That's the main one everyone wants to see.
Click to expand...
Grass is very suspect because so few events in a year. Grass statistics outside of career or long periods are dubious.

There are different ELO ratings from differenct sources. Djokovic is a great clay court player and lists I've seen have Federer higher. If you throw the beast out, both of these players would have won a lot of French Opens.

The Laver and Borg ratings are low. Laver peaked at the slams and ELO doesn't rate those higher than ordinary matches. Laver was far from dominant outside of the majors in 1969. It was the first full year of quasi-open events and its little surprise that the veteran pros had an avantage. The ELO for Laver is a little low because it doesn't preference majors. I don't think its far off the mark for Laver at 31. His peak play should have been years before that and ELO doesn't cover that period, nor should it.

Borg surprised me a bit on clay, but his peak ELO rating on all surfaces is one of the highest. I suspect clay may be underrated for him because many of his top ranked opponents were not playing much clay so he had wins over lower ranked clay specialists. The ELO is a crude tool, but for me the disparity on clay pretty much cements Rafa as the king. The French Open was hardly open in the 1970s with their stance against team tennis and other ridiculous actions. All this probably hurt Borgs rating, but it does make one wonder if Connors had played the French more during the early part of his career, might he have managed to win one. (Doubt it myself.)

Nobody has posted surface specific ELO ratings in this thread so I'd rather stay off the subject unless someone posts some data sources.
 
NatF said:
It has merits but it seems to me it's best at rating how players stack up within a small time frame, across several years and decades it seems like you would get inflation where you get points for beating the reputation.
Click to expand...
Your Elo rating is only maintained if you keep winning...
 
jm1980 said:
Your Elo rating is only maintained if you keep winning...
Click to expand...
There is another site that has been doing ELO type ratings pretty consistenly over the last two years. I'd be curious to see how Fed's ELO was in late 2015 versus now after two losses to Thiem and one to Zverev. That should have dropped his rating some would be my guess.
.....
https://tenniseloranking.blogspot.cz/

Fed's rating hasn't moved that much over the last year. I don't like the numbers on the above site. The difference between Murray and Djokovic is far too close.
 
Last edited:
You must log in or register to reply here.

Similar threads

money_ball
Peak tennis ELO ratings: Djokovic at #1
2
Replies
80
Views
7K
NoleFam
NoleFam
Meles
ELO Strong Era Wimbledon Finals
Replies
34
Views
3K
Meles
Meles
Lord Anomander
ELO Ranking instead of ATP
Replies
37
Views
4K
MatsPhilander
MatsPhilander
Meles
Weakeraerer's Delight: ELO ranks the Wimbledon Fields from 1978
2
Replies
74
Views
4K
DreddyTennis45
DreddyTennis45
nolefam_2024
ATP Finals in last 21 years
Replies
26
Views
1K
Rafa4LifeEver
Rafa4LifeEver
Back
Top