Elo ranking for top 8 and what it tells us

[nolefam_2024]

Year end ranking vs Elo

Year	1	2	3	4	5	6	7	8
2013	2552	2538	2348	2405	2322	2431 * Federer	2258	2201
2014	2552	2408	2519	2291	2257	2365	2237	2186
2015	2603 * Djokovic	2398	2450	2276	2413	2263	2239	2279
2016	2500	2604 * Djokovic	2269	2285	2296	2199	2189	2190
2017	2375	2455	2197 * Dimitrov	2231	2201	2191	2165	2076 * Goffin
2018	2415	2419	2468	2220	2286	2170	2198	2172
2019	2436	2415	2378	2211	2243	2175	2210	2102
2020	2441	2412	2250	2245	2328	2176	2187	2197
2021	2435	2309	2303	2258	2227	2376	2173	2118
2022	2225 * Alcaraz	2403	2163 * Ruud	2205	2417	2140	2316	2151
2023	2454	2344	2287	2256	2168	2201	2207	2146

Looking at the Elo year end rating, these are the conclusions on elo ratings.

1. Each year there was at least 2400 once inside top 2 ranking. There has not been a top 2 under 2400.

2. Maximum ELO was in 2500s and very early 2600s. So someone in 2500 is glowing Crimson. Nadal and Djokovic in 2013 and Djokovic and Murray (barely) qualified for that.

3. Under 2200 is usually weak. We had under 2200 top 4 elo for only these years. 2017, 2022. Both years lacked Djokovic at the top coincidentally.

4. Number 8 had 2100 ELO for 9 out of last 11 times. The number 8 is barely winning anything.

5. Number 6,7 had 2100 ELO for just 5/6 out of 11 times. Much better than number 8 but still pretty weak.

6. Almost everytime the top 5 were at least at 2200. At 2200 they can at least have a chance to win vs number 1.

7. Based on this we can see this.
Number 8 is pretty weak, 6,7 are similar level. 4,5 are similar level. Top 3 are favorites for most events. Top 2 are mostly super heavyweights.

Current Number 8 - Rune (Weak)
Current Number 6,7 - Zverev and Tsitsipas
Current Number 4,5 - Sinner, Rublev
Current Number 3 - Medvedev favorite almost everywhere
Current Number 1,2 - Alcaraz super heavyweights

 

[Neptune]

2017 and 2022, absolute "gold".

2015 triggers too much butthurt.

 

[Jonas78]

Ruud #3 at 2163

 

[Aabye5]

Any system ranking Dimitrov so high is not worth its salt. ELO isn't designed for tennis, the actual rankings do a much better job.

 

[Neptune]

Any system ranking Dimitrov so high is not worth its salt. ELO isn't designed for tennis, the actual rankings do a much better job.

Kindly recheck your reasoning.

 

[Aabye5]

Kindly recheck your reasoning.

Why? Dimi really hasn't performed like a top 10 player in the past four years, yet ELO ranks him as one.

 

[Neptune]

Why? Dimi really hasn't performed like a top 10 player in the past four years, yet ELO ranks him as one.

Give it another try? LOL

 

[Jonas78]

Dunno exactly how ELO works, but a player like Medvedev is soon 28y old and has only managed to stay consistent once in his career, from USO2020 to AO2022, and he is highly mediocre outside HC.

 

[nolefam_2024]

Any system ranking Dimitrov so high is not worth its salt. ELO isn't designed for tennis, the actual rankings do a much better job.

Right so what you are saying is ELO did better job instead of ATP ranking here. ATP made him number 3. Bro, he had zero prayer vs fedal in any consequential match.

 

[nolefam_2024]

Dunno exactly how ELO works, but a player like Medvedev is soon 28y old and has only managed to stay consistent once in his career, from USO2020 to AO2022

What is consistent here. He is winning 80% matches for 2021 and 2023. Not bad at all. Especially after covid and Wimbledon ban.

 

[Jonas78]

What is consistent here. He is winning 80% matches for 2021 and 2023. Not bad at all. Especially after covid and Wimbledon ban.

Late 2020 - AO2022 was only period he consistently went deep in slams (QF+). He fell apart after the Nadal-loss, but 2023 has been better for sure. Still R3 at AO and R1 at RG isnt exactly ATG'ish. One final is all he managed in 2023 after all.

 

[Fabresque]

Why? Dimi really hasn't performed like a top 10 player in the past four years, yet ELO ranks him as one.

Well ELO ranked him as a top 10 player in 2017. When he was #3 in the world according to the ATP rankings.

C'mon, man.

 

[nolefam_2024]

Late 2020 - AO2022 was only period he consistently went deep in slams (QF+). He fell apart after the Nadal-loss, but 2023 has been better for sure. Still R3 at AO and R1 at RG isnt exactly ATG'ish

He should not be losing r3 at AO. Agreed. He is 2 time running finalist before that though. With hernia surgery he was still getting back in the form last year.

 

[ghostofMecir]

Year end ranking vs Elo

Year	1	2	3	4	5	6	7	8
2013	2552	2538	2348	2405	2322	2431 * Federer	2258	2201
2014	2552	2408	2519	2291	2257	2365	2237	2186
2015	2603 * Djokovic	2398	2450	2276	2413	2263	2239	2279
2016	2500	2604 * Djokovic	2269	2285	2296	2199	2189	2190
2017	2375	2455	2197 * Dimitrov	2231	2201	2191	2165	2076 * Goffin
2018	2415	2419	2468	2220	2286	2170	2198	2172
2019	2436	2415	2378	2211	2243	2175	2210	2102
2020	2441	2412	2250	2245	2328	2176	2187	2197
2021	2435	2309	2303	2258	2227	2376	2173	2118
2022	2225 * Alcaraz	2403	2163 * Ruud	2205	2417	2140	2316	2151
2023	2454	2344	2287	2256	2168	2201	2207	2146

Looking at the Elo year end rating, these are the conclusions on elo ratings.

1. Each year there was at least 2400 once inside top 2 ranking. There has not been a top 2 under 2400.

2. Maximum ELO was in 2500s and very early 2600s. So someone in 2500 is glowing Crimson. Nadal and Djokovic in 2013 and Djokovic and Murray (barely) qualified for that.

3. Under 2200 is usually weak. We had under 2200 top 4 elo for only these years. 2017, 2022. Both years lacked Djokovic at the top coincidentally.

4. Number 8 had 2100 ELO for 9 out of last 11 times. The number 8 is barely winning anything.

5. Number 6,7 had 2100 ELO for just 5/6 out of 11 times. Much better than number 8 but still pretty weak.

6. Almost everytime the top 5 were at least at 2200. At 2200 they can at least have a chance to win vs number 1.

7. Based on this we can see this.
Number 8 is pretty weak, 6,7 are similar level. 4,5 are similar level. Top 3 are favorites for most events. Top 2 are mostly super heavyweights.

Current Number 8 - Rune (Weak)
Current Number 6,7 - Zverev and Tsitsipas
Current Number 4,5 - Sinner, Rublev
Current Number 3 - Medvedev favorite almost everywhere
Current Number 1,2 - Alcaraz super heavyweights

Djoko‘s Wiki-Ultimate Tennis Statistics warrior fans are going through withdrawal with no new statistics being posted and the dopamine hits from reading the older stats just aren’t as strong hence the starting of threads like this one.

 

[Jonas78]

He should not be losing r3 at AO. Agreed. He is 2 time running finalist before that though. With hernia surgery he was still getting back in the form last year.

Yeah one slam final is all he got 2023 after all. He lost me after the Nadal AO loss, both because he should have finished Nadal off, and his statement after the match. I hope Sinner/Rune/Alcaraz rise and leave the Meddy/Tsits generation behind.

 

[Aabye5]

Right so what you are saying is ELO did better job instead of ATP ranking here. ATP made him number 3. Bro, he had zero prayer vs fedal in any consequential match.

No, I'm saying ELO did a worse job because it ranks Dimitrov too highly.

 

[nolefam_2024]

No, I'm saying ELO did a worse job because it ranks Dimitrov too highly.

Ok so you need to BACK OFF now.

ELO says he is not in even top 5. ATP ranking says he is world number 3.

We are talking facts here.

 

[Pheasant]

OP, do you have these top-10 ELO for other eras too? These are interesting stats. I'd be curious to see the 1980s and 1990s.

 

[nolefam_2024]

OP, do you have these top-10 ELO for other eras too? These are interesting stats. I'd be curious to see the 1980s and 1990s.

I am not sure how accurate Elo will be for 90s. There is huge shift in tennis consistency post 2000s. But I can go as far back as 2000.

 

[Aabye5]

Ok so you need to BACK OFF now.

ELO says he is not in even top 5. ATP ranking says he is world number 3.

We are talking facts here.

What the heck are you talking about? ELO has Dimitrov at number 7, which is way too high.

 

[nolefam_2024]

What the heck are you talking about? ELO has Dimitrov at number 7, which is way too high.

Who would you put in 2017 before him. Give him 10 players.

 

[Aabye5]

Who would you put in 2017 before him. Give him 10 players.

That's the problem with ELO, one good year is weighted too highly. At the moment, Hurkacz and even Ruud are more dangerous in any given match than Dimitrov but both are ranked lower than him in ELO today. Looking at the quality of an opponent is important, but consistency is also crucial in tennis. That's why his performance over the last 52-weeks should be weighted much higher than whatever Dimitrov accomplished three or four years ago.

 

[Neptune]

OP, do you have these top-10 ELO for other eras too? These are interesting stats. I'd be curious to see the 1980s and 1990s.

I did 2 years.

1994
Pete Sampras 2351
Andre Agassi 2292
Boris Becker 2233
Sergi Bruguera 2154
Goran Ivanisevic 2169
Michael Chang 2184
Stefan Edberg 2104
Alberto Berasategui 2091

1984
John McEnroe 2553
Jimmy Connors 2374
Ivan Lendl 2351
Mats Wilander 2346
Andres Gomez 2211
Anders Jarryd 2120
Henrik Sundstrom 2101
Eliot Teltscher 2121

 

[nolefam_2024]

I did 2 years.

1994
Pete Sampras 2351
Andre Agassi 2292
Boris Becker 2233
Sergi Bruguera 2154
Goran Ivanisevic 2169
Michael Chang 2184
Stefan Edberg 2104
Alberto Berasategui 2091

1984
John McEnroe 2553
Jimmy Connors 2374
Ivan Lendl 2351
Mats Wilander 2346
Andres Gomez 2211
Anders Jarryd 2120
Henrik Sundstrom 2101
Eliot Teltscher 2121

There is different criteria we should use for 90s.

 

[Pheasant]

I did 2 years.

1994
Pete Sampras 2351
Andre Agassi 2292
Boris Becker 2233
Sergi Bruguera 2154
Goran Ivanisevic 2169
Michael Chang 2184
Stefan Edberg 2104
Alberto Berasategui 2091

1984
John McEnroe 2553
Jimmy Connors 2374
Ivan Lendl 2351
Mats Wilander 2346
Andres Gomez 2211
Anders Jarryd 2120
Henrik Sundstrom 2101
Eliot Teltscher 2121

Awesome. Thanks!

 

[jm1980]

Why? Dimi really hasn't performed like a top 10 player in the past four years, yet ELO ranks him as one.

Elo had him at Top 10 this year, the first time since 2019.

This year it's easy to see why; Dimitrov has a better record against the Top 10 than basically everyone in the Top 10 below #4

ATP rank	Record vs Top 10 (2023)
5. Rublev	5-9 (36%)
6. Tsitsipas	1-8 (11%)
7. Zverev	4-14 (22%)
8. Rune	5-9 (36%)
9. Hurkacz	2-7 (22%)
10. Fritz	3-4 (43%)
11. Ruud (bonus)	1-4 (20%)
14. Dimitrov	7-11 (39%)

No one has more wins, and the only one with a better % is Fritz, but he only had 3 wins

 

[Aabye5]

Elo had him at Top 10 this year, the first time since 2019.

This year it's easy to see why; Dimitrov has a better record against the Top 10 than basically everyone in the Top 10 below #4

ATP rank	Record vs Top 10 (2023)
5. Rublev	5-9 (36%)
6. Tsitsipas	1-8 (11%)
7. Zverev	4-14 (22%)
8. Rune	5-9 (36%)
9. Hurkacz	2-7 (22%)
10. Fritz	3-4 (43%)
14. Dimitrov	7-11 (39%)

No one has more wins, and the only one with a better % is Fritz, but he only had 3 wins

But his 11 losses should matter. He had more opportunities than anyone other than Zverev. De Minaur and Cerundolo also don't see as big a bump. That run to the Paris final shouldn't be the biggest factor in Dimitrov's ranking.

 

[jm1980]

But his 11 losses should matter. He had more opportunities than anyone other than Zverev. De Minaur and Cerundolo also don't see as big a bump. That run to the Paris final shouldn't be the biggest factor in Dimitrov's ranking.

But his 11 losses do matter; they are the reason why he's not rated higher

He had more opportunities, but that's why percentages exist. Or do you expect 1-8 Tsitsipas or 2-7 Hurkacz would have a better record if they played 18 matches instead of 9?

 

[Aabye5]

But his 11 losses do matter; they are the reason why he's not rated higher

He had more opportunities, but that's why percentages exist. Or do you expect 1-8 Tsitsipas or 2-7 Hurkacz would have a better record if they played 18 matches instead of 9?

But why is there such a discrepancy between the 52-week rankings where Dimitrov is ranked 14 and ELO where he is ranked 7? His ranking is cut in half, and I don't believe he's really played better throughout 2023 to see such a substantial bump.

At the same time, why don't we see a similar ranking jump for Cerundolo or De Minaur who have even better records vs the top 10 than Dimitrov?

 

[King_olaf_the_hairy]

It's Elo, not ELO. It isn't an abbreviation; it was named after its creator, Arpad Elo.


Elo is calculated via a formula which can be (and is) adjusted by different sources. The numbers OP quoted are presumably from UTS? They are not definitive. For example according to Tennis Abstract which uses a different variation of the formula, Dimitrov's 2017 Elo was 2052, not 2197.

For me the biggest failing of Elo in tennis is that it doesn't properly account for stuff like differences in surfaces. Ultimately Elo represents the chance a player has to win a match, however this will obviously change depending on the surface (e.g. Kyrgios on grass, Ruud on clay, etc). You can measure performance by surface Elo, but you shouldn't then combine them. FIDE has different Elo ratings for classical, rapid, and blitz time controls in chess; it doesn't publish a combined rating because that would be meaningless.

To explain. Imagine a tennis player with a rating of 2200 on clay, 2200 on hard, and 2500 on grass. If he plays the same number of games on each surface, his "combined" or average rating would be 2300. But he would NEVER play at a 2300 level; he would either play at a 2200 or 2500 level. A good example of how it can be distorted is @Neptune's list above, where Alberto Berasategui managed to achieve the tenth best Elo of the year in 1994. Going into the Year End Championships he had a 65–22 match record. This broke down as 1–4 on carpet, 1–3 on hard, 0–0 on grass, and 63–15 on clay. At the YEC — on carpet — he won only eight games in three round-robin beatdowns. His second match was against Michael Chang; according to UTS their respective Elo ratings were 2185 (MC) and 2118 (AB). Yet Chang won 6–1 6–0 and nobody was surprised.

Elo has its uses, but it has its limits as well.


@Pheasant it's not the easiest to use, but the Tennis Abstract website has a fairly comprehensive dataset. I'm not sure if you can search by year though or only players. Sackmann used the numbers in his Tennis 128 list. The Ultimate Tennis Statistics site has Elo data as well.

 

[nolefam_2024]

It's Elo, not ELO. It isn't an abbreviation; it was named after its creator, Arpad Elo.


Elo is calculated via a formula which can be (and is) adjusted by different sources. The numbers OP quoted are presumably from UTS? They are not definitive. For example according to Tennis Abstract which uses a different variation of the formula, Dimitrov's 2017 Elo was 2052, not 2197.

For me the biggest failing of Elo in tennis is that it doesn't properly account for stuff like differences in surfaces. Ultimately Elo represents the chance a player has to win a match, however this will obviously change depending on the surface (e.g. Kyrgios on grass, Ruud on clay, etc). You can measure performance by surface Elo, but you shouldn't then combine them. FIDE has different Elo ratings for classical, rapid, and blitz time controls in chess; it doesn't publish a combined rating because that would be meaningless.

To explain. Imagine a tennis player with a rating of 2200 on clay, 2200 on hard, and 2500 on grass. If he plays the same number of games on each surface, his "combined" or average rating would be 2300. But he would NEVER play at a 2300 level; he would either play at a 2200 or 2500 level. A good example of how it can be distorted is @Neptune's list above, where Alberto Berasategui managed to achieve the tenth best Elo of the year in 1994. Going into the Year End Championships he had a 65–22 match record. This broke down as 1–4 on carpet, 1–3 on hard, 0–0 on grass, and 63–15 on clay. At the YEC — on carpet — he won only eight games in three round-robin beatdowns. His second match was against Michael Chang; according to UTS their respective Elo ratings were 2185 (MC) and 2118 (AB). Yet Chang won 6–1 6–0 and nobody was surprised.

Elo has its uses, but it has its limits as well.


@Pheasant it's not the easiest to use, but the Tennis Abstract website has a fairly comprehensive dataset. I'm not sure if you can search by year though or only players. Sackmann used the numbers in his Tennis 128 list. The Ultimate Tennis Statistics site has Elo data as well.

Thanks.

The Elo formula might be same but the numbers differ between tennis abstract and uts. I used uts due to it's ease of navigation. But it does provide higher Elo rating than tennis abstract

 

[Jonas78]

Elo had him at Top 10 this year, the first time since 2019.

This year it's easy to see why; Dimitrov has a better record against the Top 10 than basically everyone in the Top 10 below #4

ATP rank	Record vs Top 10 (2023)
5. Rublev	5-9 (36%)
6. Tsitsipas	1-8 (11%)
7. Zverev	4-14 (22%)
8. Rune	5-9 (36%)
9. Hurkacz	2-7 (22%)
10. Fritz	3-4 (43%)
11. Ruud (bonus)	1-4 (20%)
14. Dimitrov	7-11 (39%)

No one has more wins, and the only one with a better % is Fritz, but he only had 3 wins

Tsits is sooo overrated. Actually he has the return stats of a plain servebot, but lacks the serve. Raonic and Andersson would have a better shot at winning a slam.

 

[jm1980]

But why is there such a discrepancy between the 52-week rankings where Dimitrov is ranked 14 and ELO where he is ranked 7? His ranking is cut in half, and I don't believe he's really played better throughout 2023 to see such a substantial bump.

At the same time, why don't we see a similar ranking jump for Cerundolo or De Minaur who have even better records vs the top 10 than Dimitrov?

de Minaur and Cerundolo have too many really bad losses that tanked their Elos

 

[nolefam_2024]

But why is there such a discrepancy between the 52-week rankings where Dimitrov is ranked 14 and ELO where he is ranked 7? His ranking is cut in half, and I don't believe he's really played better throughout 2023 to see such a substantial bump.

At the same time, why don't we see a similar ranking jump for Cerundolo or De Minaur who have even better records vs the top 10 than Dimitrov?

Is there really?

The players from number 7 to 15 are all good. It's a very top heavy sport. If you are inside top 5, then you have great chances.

 

[Wander]

But why is there such a discrepancy between the 52-week rankings where Dimitrov is ranked 14 and ELO where he is ranked 7? His ranking is cut in half, and I don't believe he's really played better throughout 2023 to see such a substantial bump.

At the same time, why don't we see a similar ranking jump for Cerundolo or De Minaur who have even better records vs the top 10 than Dimitrov?

52-week ranking by its nature has a lag compared to Elo rating because it represents a kind of performance average of the last 52 weeks. For Elo, more recent results are more relevant. Dimitrov finished the season strong which is largely why his Elo ranking is higher than his ATP ranking.

 

[Dwell]

52-week ranking by its nature has a lag compared to Elo rating because it represents a kind of performance average of the last 52 weeks. For Elo, more recent results are more relevant. Dimitrov finished the season strong which is largely why his Elo ranking is higher than his ATP ranking.

His hair finished the season strong too. ATP breakthrough of the year

View this content on Instagram

View this content on Instagram

 

[Aabye5]

52-week ranking by its nature has a lag compared to Elo rating because it represents a kind of performance average of the last 52 weeks. For Elo, more recent results are more relevant. Dimitrov finished the season strong which is largely why his Elo ranking is higher than his ATP ranking.

But it doesn't really indicate his overall performance for 2023. That's why I think the ATP rankings are more useful overall.

 

[Wander]

But it doesn't really indicate his overall performance for 2023. That's why I think the ATP rankings are more useful overall.

I agree that ATP and WTA rankings are better for their intended purposes, such as deciding which tournaments you qualify for and tournament seeding etc.

What the 52 week rankings are not better at is predicting match results. You can test this for yourself and prediction models based on Elo ratings will perform better on average than a model based on ATP or WTA rankings.

 

[Jonas78]

Games won% on the given surface is a much better stat and predictor than Elo imo. Get your games won% to ~60% on the given surface over time and you will win slams. Meddy did that in 2021 and Alcaraz has done it, and they both won slams. Dimitrov was always highly mediocre and didnt pass 55-56% even in his best seasons.

 

[Poisoned Slice]

@Meles

 

[Aabye5]

I agree that ATP and WTA rankings are better for their intended purposes, such as deciding which tournaments you qualify for and tournament seeding etc.

What the 52 week rankings are not better at is predicting match results. You can test this for yourself and prediction models based on Elo ratings will perform better on average than a model based on ATP or WTA rankings.

But that's just it, in terms of match wins I haven't found it to be better. Let's follow Dimitrov, for example, and see whether it really makes sense to weight the newest matches more than matches earlier in the year. I suspect that his win over Alcaraz in Paris may not have come so easy earlier in the year. We can see how it pans out for him up to Australia.

 

[Neptune]

I agree that ATP and WTA rankings are better for their intended purposes, such as deciding which tournaments you qualify for and tournament seeding etc.

What the 52 week rankings are not better at is predicting match results. You can test this for yourself and prediction models based on Elo ratings will perform better on average than a model based on ATP or WTA rankings.

No ranking system is flawless, but both the Elo and ATP systems are well-developed and transparent, far superior to relying on biased feelings or eyetest.

Elo, in particular, excels even better than the ATP ranking in predicting match outcomes, something important to keep in mind.

That being said, a surface-specific ranking or Elo system would be more accurate. Currently, the general ranking or Elo system provides a better indication on hardcourt than on grass or clay.

 

[mike danny]

@Towny, the floor is yours.

 

[Towny]

@Towny, the floor is yours.

I would comment on the significant flaws in Elo but I must admit, I don't even get what the point of this thread is

 

[Meles]

Any system ranking Dimitrov so high is not worth its salt. ELO isn't designed for tennis, the actual rankings do a much better job.

You're catching on; ATP ranking is suspect.

 

[Meles]

@Meles

Truth be told Holger is overrated on all fronts. But with Big 3 suddenly looking like the Golden Boys(Girls) and Medly, Cramplitos, and Panick not playing until Auz Open, Holger has just enough A-bag in him to get the job done (Becker, Luthi, and Kenneth Carlsen; what can go wrong?)

 

[Jonas78]

Truth be told Holger is overrated on all fronts. But with Big 3 suddenly looking like the Golden Boys(Girls) and Medly, Cramplitos, and Panick not playing until Auz Open, Holger has just enough A-bag in him to get the job done (Becker, Luthi, and Kenneth Carlsen; what can go wrong?)

Lol

 

[TMF]

I would comment on the significant flaws in Elo but I must admit, I don't even get what the point of this thread is

Desperation

 

[Aabye5]

You're catching on; ATP ranking is suspect.

Dimitrov is suspect. Nice guy, but his mentality has been suspect his whole career.