PDA

View Full Version : the Grand Slam Record Book vol. 2


chris whiteside
Aug 17th, 2011, 05:32 PM
We're being very well catered for at the moment for the Grand Slam Record Book vol. 2 has just been issued.

This gives the complete women's draws for all the Slams from inception (RG 1925).

Each page is split down the middle with half the draw on each side which I find very reader friendly.

These draws are in English - unfortunately perhaps for the majority but certainly for me any pages of dialogue is in Italian!

There is a table which would appear to be some statistical attempt to work out who the GOAT is, the top 10 being:

01 Steffi Graf 2760
02 Monica Seles 2751
03 Martina Navratilova 2728
04 Helen Wills 2721
05 Chris Evert 2699
06 Suzanne Lenglen 2680
07 Margaret Court 2670
08 Martina Hingis 2668
09 Maureen Connolly 2650
10 Billie Jean King 2643

The list goes down to #20 then weirdly gives the #s 22, 23, 24, 26, 28, 34, 35, 37, 38, 41, 46 and 52 positions. :confused:

If anyone does get the book and can understand Italian maybe they could tell us what criteria they used and whether any reason is given for this baffling listing.

Rollo
Aug 17th, 2011, 05:52 PM
Interesting Chris!

Does it also provide doubles and mixed draws?

chris whiteside
Aug 18th, 2011, 01:16 PM
Interesting Chris!

Does it also provide doubles and mixed draws?

Yes!!!!!!!!!!!!!! although It doesn't have the seedings for those events from what I have seen.

aleale
Sep 7th, 2011, 11:48 AM
I would like to thank you
This forum has been VERY helpful for me.

My Friend and co-autor Andrea Carta will write the explanation of this method ( ELO )
http://en.wikipedia.org/wiki/Elo_rating_system

The Grand Slam Record Book Vol 2 included ALL grand Slam tournaments
Aus, RG, Wimbledon, US (Woman) Single, double and mixed double.

if you have any questions, please write to me.

There is a "Grand Slam Record Book Vol.1" too.
with all the MAN draws ( single and double)

Ciao
Alessandro Albiero

Fantasio
Sep 7th, 2011, 01:24 PM
My Friend and co-autor Andrea Carta will write the explanation of this method ( ELO )
http://en.wikipedia.org/wiki/Elo_rating_system
Here I am.
Chris, Rollo, if you are still interested in the method, I shall provide full details.
In the book, I only listed top 20 + most known players (that's the "weird" reason). BTW, I'll provide full rankings as soon as the US Open will come to an end (let's hope before Tuesday!).

Andy

chris whiteside
Sep 7th, 2011, 05:31 PM
Great book Andy and Alessandro - just wish I could understand Italian!

I particularly like the lay-out.

I would be very interested for any SIMPLE explanation you could give regarding the method - I did try to read the Wiki article but it was way over my head!

On page 23 you mention Rino Tommasi's statistical method of ranking and 1961 is mentioned. Is it possible that this is saying that he began them then?

If so do you have access to any of his lists from 1961-1970?

Thanks and looking forward to full list at end of USO.

Fantasio
Sep 8th, 2011, 06:20 PM
On page 23 you mention Rino Tommasi's statistical method of ranking and 1961 is mentioned. Is it possible that this is saying that he began them then?
No, that's another matter - we only mentioned his method.

I would be very interested for any SIMPLE explanation you could give regarding the method - I did try to read the Wiki article but it was way over my head!
I'll try. As you highlighted SIMPLE, I'll stay "simple" as most as possibile.

1) The Elo method was invented in the sixties, and applies to chess. Today it provides reliable rankings in chess, but we must remember these rankings are always related to the whole career, not the last year or couple of years. So the method is unreliable if you want "real-time" rankings. For the same reason - whole career - it's very good for GOAT analysis instead.

2) Nobody worries, in chess, about the "whole career" problem, because decline is slow. Sometimes, especially if a player comes back after long inactivity, it IS a problem.

3) As there are many similarities between chess and tennis, Mazak suggested to apply ELO method to tennis.

4) As there are some differences between chess and tennis, I modified some of the method's parameters according to these differences.

5) The method works this way: the matches are all the same, tournaments do not matter (that's one of the thing I modified). Only matches count. Players start the professional career at 2000 points' level. When two players meet, the winner gains points, the loser loses the same points. If difference between them is great, few points are gained (if the stronger player wins). If difference is great, many points are gained (if the weaker player wins).

6) Mathematical formulas rule the method. In chess, as the players start at 2000 points, these formulas imply that elite players are ranked between 2600-2700, sometimes up to 2800; null players are ranked at 1300-1400, with absolute starters at 1200. As I confirmed the 2000 points rule, the tennis's rankings are similar to the chess ones.

7) Now, the important thing.

8) This is NO GOAT. This is dominance level achieved, opponents considered. Certanly a strong indicator of who the GOAT may be, but not the only one. I myself believe this to be the most important indicator, but there are also longevity, number of victories in majors, H2H against main opponents, polyvalence, number of victories in minors, as we all know.

9) So that's the reason why Seles and Graf are almost equal. Seles has less victories, less longevity, little polyvalence, but dominance level is similar. It doesn't matter if dominance lasted 2-3 years or much more. If dominance lasted enough to become unquestionable, that's good for the ELO method (in chess there are a lot of significant examples).

10) Last but not least, forgive me if my english is far from perfect. At least I don't use Babelfish!

Andy

GeeTee
Sep 8th, 2011, 08:44 PM
Ciao Andy.

What is 'polivalence'?

Fantasio
Sep 8th, 2011, 08:56 PM
What is 'polivalence'?
Winning on all surfaces, basically. What is the correct term?

Andy

GeeTee
Sep 8th, 2011, 11:38 PM
Winning on all surfaces, basically. What is the correct term?

Andy
I'm not sure there is one single word in English really..

'All-court ability (or record)' maybe? Although 'all-court' tends to mean a versatile player who can mix serve/volley and baseline play.

Any other suggestions guys??

Andy T
Sep 9th, 2011, 01:18 PM
I'm not sure there is one single word in English really..

'All-court ability (or record)' maybe? Although 'all-court' tends to mean a versatile player who can mix serve/volley and baseline play.

Any other suggestions guys??

I agree GeeTee that all-court tends to refer to the style of game a player has as opposed to his or her ability across all surfaces. Maybe "success on all surfaces" is a better term. The quality is the player's adaptability.

Sam L
Sep 9th, 2011, 01:56 PM
No, that's another matter - we only mentioned his method.


I'll try. As you highlighted SIMPLE, I'll stay "simple" as most as possibile.

1) The Elo method was invented in the sixties, and applies to chess. Today it provides reliable rankings in chess, but we must remember these rankings are always related to the whole career, not the last year or couple of years. So the method is unreliable if you want "real-time" rankings. For the same reason - whole career - it's very good for GOAT analysis instead.

2) Nobody worries, in chess, about the "whole career" problem, because decline is slow. Sometimes, especially if a player comes back after long inactivity, it IS a problem.

3) As there are many similarities between chess and tennis, Mazak suggested to apply ELO method to tennis.

4) As there are some differences between chess and tennis, I modified some of the method's parameters according to these differences.

5) The method works this way: the matches are all the same, tournaments do not matter (that's one of the thing I modified). Only matches count. Players start the professional career at 2000 points' level. When two players meet, the winner gains points, the loser loses the same points. If difference between them is great, few points are gained (if the stronger player wins). If difference is great, many points are gained (if the weaker player wins).

6) Mathematical formulas rule the method. In chess, as the players start at 2000 points, these formulas imply that elite players are ranked between 2600-2700, sometimes up to 2800; null players are ranked at 1300-1400, with absolute starters at 1200. As I confirmed the 2000 points rule, the tennis's rankings are similar to the chess ones.

7) Now, the important thing.

8) This is NO GOAT. This is dominance level achieved, opponents considered. Certanly a strong indicator of who the GOAT may be, but not the only one. I myself believe this to be the most important indicator, but there are also longevity, number of victories in majors, H2H against main opponents, polyvalence, number of victories in minors, as we all know.

9) So that's the reason why Seles and Graf are almost equal. Seles has less victories, less longevity, little polyvalence, but dominance level is similar. It doesn't matter if dominance lasted 2-3 years or much more. If dominance lasted enough to become unquestionable, that's good for the ELO method (in chess there are a lot of significant examples).

10) Last but not least, forgive me if my english is far from perfect. At least I don't use Babelfish!

Andy


Very interesting method. Welcome to the board, ciao! :) Where can I get this book?

Fantasio
Sep 12th, 2011, 12:47 PM
Very interesting method. Welcome to the board, ciao! :) Where can I get this book?
Alessandro - I hope - will know that!

After US 2011, that's the new all-time ELO ranking, top 50 (but no change occurred at this level):

1) Graf 2760
2) Seles 2751
3) Navratilova 2728
4) Wills-Moody 2721
5) Evert 2699
6) Lenglen 2680
7) Smith-Court 2670
8) Hingis 2668
9) Connolly 2650
10) Moffitt-King 2643
11) S. Williams 2637
12) V. Williams 2612
13) Henin 2610
14) Goolagong-Cawley 2601
15) Cljisters 2577
16) Bueno 2566
17) Marble 2560
18) Sanchez-Vicario 2555
19) Davenport 2549
20) Austin 2548
21) Jacobs 2540
22) Mandlikova 2529
23) Sabatini 2525
24) Bjurstedt-Mallory 2525
25) Gibson 2520
26) Novotna 2517
27) Hart 2514
28) Sharapova 2513
29) Osborne-DuPont 2510
30) Richey-Gunter 2509
31) Krahwinkel-Sperling 2507
32) Brough 2504
33) Round 2499
34) Douglass-Lambert-Chambers 2498
35) Pierce 2493
36) Haydon-Jones 2489
37) Mauresmo 2489
38) Capriati 2481
39) Fry 2477
40) Martinez 2469
41) Betz 2468
42) McKane-Godfree 2461
43) Hard 2461
44) Passemard-Mathieu 2459
45) Sukova 2456
46) Ivanovic 2453
47) Turner-Bowrey 2448
48) Jaeger 2441
49) Shriver 2437
50) Li 2436

20- points' difference is not that important, BTW.

aleale
Sep 12th, 2011, 01:20 PM
Alessandro - I hope - will know that!



The grand slam record book Vol.2
by amazon.it
http://www.amazon.it/grand-slam-record-book/dp/8860020239/


The grand slam record book Vol.1
http://www.amazon.it/grand-slam-record-book/dp/8860020174/


In UK
http://www.tennisbookshop.com/news.php


or write to Our editor:

effepilibri@email.it

Mr. Pietro Farro

Sam L
Sep 12th, 2011, 01:43 PM
The grand slam record book Vol.2
by amazon.it
http://www.amazon.it/grand-slam-record-book/dp/8860020239/


The grand slam record book Vol.1
http://www.amazon.it/grand-slam-record-book/dp/8860020174/


In UK
http://www.tennisbookshop.com/news.php


or write to Our editor:

effepilibri@hotmail.it

Mr. Pietro Farro
Thanks. Tennis Bookshop mails worldwide so might get it from there.

chris whiteside
Sep 13th, 2011, 12:03 PM
Thanks for the explanation regarding the Elo method, Fantasio.

The one thing I don't understand is exactly how the number of points a player loses or gains is determined.

Interesting that in these analyses whatever method is used Ann Jones always finishes ahead of Virginia Wade and to finish inside the Top 40 of an all-time list is no mean feat.

I'm assuming that #45 on the up-to-date list is Helena. Can you confirm it is she and not Vera?

Fantasio
Sep 13th, 2011, 04:51 PM
The one thing I don't understand is exactly how the number of points a player loses or gains is determined.
It depends on the rating of the two players involved. The difference in rating determines chances of victory/defeat for each one of them. Points gained/losed are determined comparing the actual result with these chances. If you prefer, I can give you the exact formula. :-)

in these analyses whatever method is used Ann Jones always finishes ahead of Virginia Wade and to finish inside the Top 40 of an all-time list is no mean feat.
Three victories and 6 more finals during 9 years are a lot. Between 1966 and 1969 Haydon-Jones had a little dominance (not like King's and Court's, of course). Wade only has 3 victories during 11 years, no finals, no dominance at all. Wade is ranked 53th. Haydon-Jones 36th.

I'm assuming that #45 on the up-to-date list is Helena. Can you confirm it is she and not Vera?
Of course it's Helena!

chris whiteside
Sep 14th, 2011, 11:49 AM
It depends on the rating of the two players involved. The difference in rating determines chances of victory/defeat for each one of them. Points gained/losed are determined comparing the actual result with these chances. If you prefer, I can give you the exact formula. :-)


Three victories and 6 more finals during 9 years are a lot. Between 1966 and 1969 Haydon-Jones had a little dominance (not like King's and Court's, of course). Wade only has 3 victories during 11 years, no finals, no dominance at all. Wade is ranked 53th. Haydon-Jones 36th.


I would like to see the exact formula because I am totally confused.

For example if Graf and Seles played surely a victory for one of them over a major opponent must be worth something yet there is minimal differential.

Maybe it's too simplistic but to me more points should be gained by beating a higher ranked player i.e. say if a player (A) on 2458 plays someone (B) on 1958 then someone (C) on 2168 it is more of a feat to beat the second player yet there is a bigger differential in points with the first player?

Also are we saying that if A beats B she will gain points and B lose a similar number but will this be the same number of points if it happened that B beat A?

My understanding is that all tournaments are equal and only matches count so why would Jones having 6 extra Slam finals than Wade be particularly relevant? I think I'm becoming more confused.

Fantasio
Sep 14th, 2011, 12:29 PM
I would like to see the exact formula because I am totally confused.

Maybe it's too simplistic but to me more points should be gained by beating a higher ranked player i.e. say if a player (A) on 2458 plays someone (B) on 1958 then someone (C) on 2168 it is more of a feat to beat the second player yet there is a bigger differential in points with the first player?

The formula is: <points gained/lost = K * (1 / (1 + 10 ^ (difference / 400)))>, that is, when A is rated 2458 and B 1958, if A wins he/she gains 0.8 points (we assume K = 15), B loses 0.8 points. If B wins he/she gains 14.2 points, A loses 14.2.
When A and C play, difference is not so high, so if A wins he/she gains 2.4 points, C loses 2.4; if C wins he/she gains 12.6 points, and C loses the same.

Now it should be clear (I hope).

why would Jones having 6 extra Slam finals than Wade be particularly relevant? I think I'm becoming more confused.
Theoretically, it's possible that Jones reached all her finals beating only extremely low-ranked players, while Wade - also theoretically - could have reached her semifinals beating Court, King, Evert and so on. If such were the case, Wade would be ranked higher. In one tournament, I guess it's a real possibility. Maybe even two tournaments. Six times, I guess not. If Jones reached 6 finals and Wade did not, I guess most times Jones defeated someone stronger than Wade did, so in the end... Jones is ranked higher.

Andy

Sam L
Sep 24th, 2011, 12:09 AM
The formula is: <points gained/lost = K * (1 / (1 + 10 ^ (difference / 400)))>, that is, when A is rated 2458 and B 1958, if A wins he/she gains 0.8 points (we assume K = 15), B loses 0.8 points. If B wins he/she gains 14.2 points, A loses 14.2.
When A and C play, difference is not so high, so if A wins he/she gains 2.4 points, C loses 2.4; if C wins he/she gains 12.6 points, and C loses the same.

Now it should be clear (I hope).


Theoretically, it's possible that Jones reached all her finals beating only extremely low-ranked players, while Wade - also theoretically - could have reached her semifinals beating Court, King, Evert and so on. If such were the case, Wade would be ranked higher. In one tournament, I guess it's a real possibility. Maybe even two tournaments. Six times, I guess not. If Jones reached 6 finals and Wade did not, I guess most times Jones defeated someone stronger than Wade did, so in the end... Jones is ranked higher.

Andy

I spoke to someone who plays chess and really understands this rating system and he explained to me in layman's terms how it works. I think I have a better understanding how combined with your explanation. Basically, it works as a statistical probability that should player A and B meet, given their performances in the past the one with the higher rating should prevail. Ultimately, it is an assessment of skill level judging by your head-to-head against all others. If you have a lot of wins against someone rated very highly, the system will rate you higher. It's a test of dominance against your peers and against those whom you've never met. Court and Graf never played but judging by the players that they played and they ultimately played, in a random match up Graf should prevail.

I think it's a great system for measuring GOAT at least on a statistical level and then you can add intangibles to it. It is a starting point certainly.

But I have a question. Why does this list differ from the one posted here? http://www.tennisforum.com/showthread.php?t=330900

Is that purely because the latter only consisted of open era players?

DennisFitz
Sep 24th, 2011, 11:30 PM
No, that's another matter - we only mentioned his method.


I'll try. As you highlighted SIMPLE, I'll stay "simple" as most as possibile.

1) The Elo method was invented in the sixties, and applies to chess. Today it provides reliable rankings in chess, but we must remember these rankings are always related to the whole career, not the last year or couple of years. So the method is unreliable if you want "real-time" rankings. For the same reason - whole career - it's very good for GOAT analysis instead.

2) Nobody worries, in chess, about the "whole career" problem, because decline is slow. Sometimes, especially if a player comes back after long inactivity, it IS a problem.

3) As there are many similarities between chess and tennis, Mazak suggested to apply ELO method to tennis.

Apart from both being one-on-one competitions, I don't think there is anything remotely similar. Chris vs Martina on clay and then Chris vs Martina on grass? Indoors? Or at what stage in their career? It does make a difference, and helps inform opinions and decisions when rating players across eras, and for a career. Chess doesn't have anything like that.

4) As there are some differences between chess and tennis, I modified some of the method's parameters according to these differences.

5) The method works this way: the matches are all the same, tournaments do not matter (that's one of the thing I modified). Only matches count. Players start the professional career at 2000 points' level. When two players meet, the winner gains points, the loser loses the same points. If difference between them is great, few points are gained (if the stronger player wins). If difference is great, many points are gained (if the weaker player wins).

Wondering if the part about tournaments do not matter, but you modified, means the majors were weighted more than regular tour events?
Again, if H2H results are the determining facto, I can't fathom how Graf and Seles are #1 and #2 overall, with Graf holding what seems like a slim edge, when she had 2-1 advantage in H2H ?! Also, a winner gaining points and loser losing points sounds more like a boxing formula. Tennis is played as a tournament. And there's something to be said about winning several rounds before having to play against your rival.

9) So that's the reason why Seles and Graf are almost equal. Seles has less victories, less longevity, little polyvalence, but dominance level is similar. It doesn't matter if dominance lasted 2-3 years or much more. If dominance lasted enough to become unquestionable, that's good for the ELO method (in chess there are a lot of significant examples).

Andy

I respectfully, but LOUDLY disagree. You say the reason Graf and Seles are almost equal, yet you point out the number of ways in which Seles is less than Graf. But their dominance level is similar. Seles had 2 dominating years (even though she wasn't even ranked #1 for all of 1991). Graf had 5, and you could even say 6 dominating years: 1988, 1989, 1993, 1995-1996 (and don't forget 1987 w/only 2 losses). That's more than twice as much as Seles, at her best. Graf matched and doubled Seles' best years in the majors. And Seles' two best years overall in terms of W-L aren't even in the top 15 of the Open era. I absolutely disagree that just because Seles achieved two years of dominance (and her 1991-1992 were impressive), it absolutely doesn't guarantee any future similar success. That's the beauty of the sport. It happens in real time. Not based on statistical probability of what should or might happen. It's also what makes a system for rating players over different eras interesting. You really do have to look at the big picture.

So for me, system that rates Graf #1 and Seles #2 all-time is deeply flawed system that wants to take into account all the wouldas, coulda, shouldas. And doesn't acknowledge or take into account the incredible career achievements of Navratilova, Court, Evert, Connolly, etc. who rank behind Seles.

Fantasio
Sep 25th, 2011, 12:18 AM
But I have a question. Why does this list differ from the one posted here? http://www.tennisforum.com/showthread.php?t=330900

Is that purely because the latter only consisted of open era players?
Yes. Mainly because it consisted - as you guessed - of open era players only. The method works correctly if ALL matches are counted. You cannot start "in the middle" because "truncated" players (let's say, King and Court) are not correctly rated, and such an error propagates further on.
Of course, it's also likely that Wuornos used different parameters, so it's unlikey the two rankings will ever be exactly the same.

Fantasio
Sep 25th, 2011, 12:32 AM
Wondering if the part about tournaments do not matter, but you modified, means the majors were weighted more than regular tour events?
Only majors are counted. To count "minors" is not possible, for a lot of reasons.

Graf matched and doubled Seles' best years in the majors. And Seles' two best years overall in terms of W-L aren't even in the top 15 of the Open era. I absolutely disagree that just because Seles achieved two years of dominance (and her 1991-1992 were impressive), it absolutely doesn't guarantee any future similar success.
You did not understand how the method works, but that's not your fault - I explained that in the book. ELO ONLY rates dominance level, NOT dominance duration. It doesn't matter if dominance lasts two, or 20, or 2000 years, if dominance is clear (Seles' was).
In chess, the problem does not exist. If you ask people who the chess GOAT was, some will say "Fischer", some "Kasparov". Fischer, as Seles did, only dominated two/three years, between 1970 and 1972, then disappeared. Kasparov dominated 20 years, between 1984 and 2004, yet that's not enough. For chess fans, used to the ELO method, only dominance level counts, and this is the case.
Of course, you may think this is wrong, but I'm not saying the opposite: ELO does NOT claim the most rated player is the GOAT, it only shows who dominated most. Of course, there are other things to count, if you want to find out the GOAT.

Sam L
Sep 25th, 2011, 12:37 AM
Apart from both being one-on-one competitions, I don't think there is anything remotely similar. Chris vs Martina on clay and then Chris vs Martina on grass? Indoors? Or at what stage in their career? It does make a difference, and helps inform opinions and decisions when rating players across eras, and for a career. Chess doesn't have anything like that.



Wondering if the part about tournaments do not matter, but you modified, means the majors were weighted more than regular tour events?
Again, if H2H results are the determining facto, I can't fathom how Graf and Seles are #1 and #2 overall, with Graf holding what seems like a slim edge, when she had 2-1 advantage in H2H ?! Also, a winner gaining points and loser losing points sounds more like a boxing formula. Tennis is played as a tournament. And there's something to be said about winning several rounds before having to play against your rival.



I respectfully, but LOUDLY disagree. You say the reason Graf and Seles are almost equal, yet you point out the number of ways in which Seles is less than Graf. But their dominance level is similar. Seles had 2 dominating years (even though she wasn't even ranked #1 for all of 1991). Graf had 5, and you could even say 6 dominating years: 1988, 1989, 1993, 1995-1996 (and don't forget 1987 w/only 2 losses). That's more than twice as much as Seles, at her best. Graf matched and doubled Seles' best years in the majors. And Seles' two best years overall in terms of W-L aren't even in the top 15 of the Open era. I absolutely disagree that just because Seles achieved two years of dominance (and her 1991-1992 were impressive), it absolutely doesn't guarantee any future similar success. That's the beauty of the sport. It happens in real time. Not based on statistical probability of what should or might happen. It's also what makes a system for rating players over different eras interesting. You really do have to look at the big picture.

So for me, system that rates Graf #1 and Seles #2 all-time is deeply flawed system that wants to take into account all the wouldas, coulda, shouldas. And doesn't acknowledge or take into account the incredible career achievements of Navratilova, Court, Evert, Connolly, etc. who rank behind Seles.

I think Fantasio did a better job of explaining but this isn't a complete measure of greatness. It's a measure of dominance during a period of time and Seles was as dominant as anybody in history of tennis.

Yes. Mainly because it consisted - as you guessed - of open era players only. The method works correctly if ALL matches are counted. You cannot start "in the middle" because "truncated" players (let's say, King and Court) are not correctly rated, and such an error propagates further on.
Of course, it's also likely that Wuornos used different parameters, so it's unlikey the two rankings will ever be exactly the same.

Ok thanks. I agree that it must be considered from the beginning and not just open era.

DennisFitz
Sep 25th, 2011, 06:23 PM
Only majors are counted. To count "minors" is not possible, for a lot of reasons.


You did not understand how the method works, but that's not your fault - I explained that in the book. ELO ONLY rates dominance level, NOT dominance duration. It doesn't matter if dominance lasts two, or 20, or 2000 years, if dominance is clear (Seles' was).
In chess, the problem does not exist. If you ask people who the chess GOAT was, some will say "Fischer", some "Kasparov". Fischer, as Seles did, only dominated two/three years, between 1970 and 1972, then disappeared. Kasparov dominated 20 years, between 1984 and 2004, yet that's not enough. For chess fans, used to the ELO method, only dominance level counts, and this is the case.
Of course, you may think this is wrong, but I'm not saying the opposite: ELO does NOT claim the most rated player is the GOAT, it only shows who dominated most. Of course, there are other things to count, if you want to find out the GOAT.

OK, sort of.

So it's all, and only about dominance, right! In the majors. Noting else.

I wonder why Serena isn't higher. She wen undefeated in majors in 2002. Won 2 of 3 in 2003. And again in 2010. Each time losing to the eventual champ in the French. Three years of domination.

Navratilova won 3 of the 4 majors in 1983-1984. Her losses were in the SF and 4R, which I realize is not quite the same as Seles 1991-1992 seasons. Martina also won 2 of 4 in 1985, losing in the finals of the other two, 2 of the 3 majors in 1986, losing in the final of the other. And winning 2 of 4 in 1987, losing in the finals of the other two. And that doesn't rate as high as Seles' ?!?!?!?!?!?!?!?! :eek:

I would think that if you have comparable levels of dominance, the one who does it more rates higher? Sounds like a system that perhaps punishes players with longer careers, like Serena, Chris, Martina, because they had some "off" years where they "only" won 1 major, and didn't dominate.

As far as dominance in the majors, I'd easily take:
Court, and her 1970 Grand Slam year, plus 1963, 1965, 1969, and 1973 seasons (and even if you only count Open era, Court's dominance in the majors outranks Seles').
Navratilova's 1983-1984, plus 1985-1987 seasons
Serena's 2002-2003, 2010 seasons.
Chris Evert's 1974-1975-1976, 1980 seasons
Connolly's 1952-1953 seasons

OVER Seles' 1991-1992 seasons, when looking at it from a dominance perspective. And in this case, less is not more, in terms of # of dominating seasons. I mean if you have 3 dominating years in the majors, doesn't that outrank two, or one! And if you win the Grand Slam, as Connolly did, doesn't that outrank someone who skipped a major!

Not even sure if decisive wins, i.e., 6-0,6-0, are factored in. But in tennis, it only matters who wins the last point. Seles' 1991-1992 seasons in the majors were great. She didn't win any major without losing a set. The other players above all did. That's dominance!

austinrunner
Sep 26th, 2011, 07:40 AM
Only majors are counted. To count "minors" is not possible, for a lot of reasons.

Which are?

Fantasio
Sep 26th, 2011, 09:41 AM
As far as dominance in the majors, I'd easily take:
Court, and her 1970 Grand Slam year, plus 1963, 1965, 1969, and 1973 seasons (and even if you only count Open era, Court's dominance in the majors outranks Seles').
Navratilova's 1983-1984, plus 1985-1987 seasons
Serena's 2002-2003, 2010 seasons.
Chris Evert's 1974-1975-1976, 1980 seasons
Connolly's 1952-1953 seasons

OVER Seles' 1991-1992 seasons, when looking at it from a dominance perspective. And in this case, less is not more, in terms of # of dominating seasons. I mean if you have 3 dominating years in the majors, doesn't that outrank two, or one!
I will repeat myself.
In Elo method, dominance duration does NOT count. Only dominance level. Of course Court dominated 5 years, Navratilova also five. Also great was Evert, not mentioning Lenglen and Wills, that you forgot.
All these players could have dominated 20 years or more, there's no difference.

IF you are only looking at dominance level, it's easy to notice that between 1991 and 1993 Seles played 8 majors, won 7, only losing one match to Graf (who is - without any doubt - one of the most likey GOAT candidates). She defeated Graf (twice), Sanchez-Vicario (3 times), old Navratilova (twice), Novotna (once), Sabatini (3 times).
If this is NOT supreme dominance, I don't know how to define such a thing.
Well, Navratilova 1983-1984 dominated. But she "only" won 6 majors, losing to Horvath and Sukova. Losing to a low-ranked player such as Horvath is a disaster, comparing to losign to Graf (in Elo method).

But there's Court in 1969-1971. Well, she's better than Seles, having won 8 majors, losing only to Haydon-Jones - better than Horvath, still no Graf. And Court only defeated King twice, Goolagong twice. Much, bur not as much as Seles. Plus, don't forget she only played 49 matches during 9 majors (winning 48), while Seles played 56, winning 55, despite competing in 8 events. That's why Seles is higher than Court.

Serena played 6 majors, won 5. No match.

Evert played 8, won 6. Very good, and she defeated Goolagong 3 times, Navratilova also 3 times. She only lose to Goolagong and King. That's very good, and she's very close to Seles, but still not equal.

Connolly extraordinary. 9 over 9! Her problem, she did not defeat strong players. Hart, Fry, Brough are top 30 (alltime) at best. And she played 51 matches only.

So, why is Graf ranked higher (more or less, she's equal)? Because between 1988 and 1990 she won 8 majors, played 9, only lose to Sanchez-Vicario (better than Haydon-Jones), defeated Navratilova (3 times), old Evert (3 times), Sabatini (4 times), young Seles (twice), young Sanchez-Vicario (once). I think Seles defeated better opponents, but Graf won more, and more matches (62 over 63).

Now the difference between "dominance level" and "dominance duration" should be clear. Of course I know that Graf dominated more years (there are 1995 and 1996, of course), won much more than Seles, so she's a better GOAT candidate than unlucky Monica. But dominance level is the same.

Or, we can also say that "dominance level" = "strength at best" (if consistent). So, the old question, "was Seles really better than Graf or was Graf not at her best during 1991-1992?" could be settled, as the two players were of equal strength (according to Elo).

Sam L
Sep 26th, 2011, 10:04 AM
Fantasio, thanks for that explanation. And now I am really understanding this. It confirms my suspicions that the reason why Seles is so high is because of her success against Graf.

The reason why Connolly who had same Grand Slam success as Seles is, comparatively, so low is because she didn't have high quality competition.

I think, and this is just my opinion and you can call biased if you like, this is a terrific rating system for determining GOAT.

The problem with other statistics and arguments is that ultimately it isn't objective and it doesn't consider the skill level of players. This determines a player's skill level, not just against her peers but also against all those who ever played in a single Grand Slam tennis match.

If Grand Slams are to be used as the only arbiters for greatness I think this should be the system used. Of course, there's the argument for other tournaments and No. 1 ranking etc.. too.

To sum up, Seles dominated an era that included a GOAT and her losses came to that GOAT. Comparatively, it was harder for her to have dethroned Graf than anyone else in history and for dethroning Graf, she's rewarded heavily here.

tennisvideos
Sep 26th, 2011, 11:29 AM
All very interesting but highly subjective IMO esp when comparing opposition - Doris Hart and Anne Jones, for example, were both great players IMO and probably highly underrated by most. And to only factor in the slams is very limiting - years at #1, career singles winning % and tournament wins should also factor in there somewhere.

Happy to see it gives Seles her due but even I, a huge fan, could never rate her above Connolly, Court, Navratilova and Evert.

Still, an interesting perspective.

chris whiteside
Sep 26th, 2011, 11:53 AM
At least I now understand this including the formula.

Slams only - I was under the misconception it was all tournaments which is why I couldn't see why Jones' Slam finals should have been particularly significant in relation to Wade. Makes sense now as Ann was more consistent in the Slams than Virginia.

Certainly an interesting and worthwhile system although personally I believe there are many factors to take in in determining the GOAT including Tier 1/Major Championships.

Margaret Court for example went through Seasons where she won numerous events losing only a handful of matches all year. Admittedly some events were weaker than others but she still beat all her closest rivals - that's dominance.

The figures don't appear to show it but after losing early at her first two Slams Maureen Connolly then went on to win 9 Slams in a row she entered before retiring on the way beating the leading players of the day - you can't do any more than that so to me that's the ultimate.

Anyway you could go on ad infinitum arguing as to what should constitute the criteria.

It is as it is.

Sam L
Sep 26th, 2011, 11:59 AM
The figures don't appear to show it but after losing early at her first two Slams Maureen Connolly then went on to win 9 Slams in a row she entered before retiring on the way beating the leading players of the day - you can't do any more than that so to me that's the ultimate.


But her competition didn't include another like her. If she came immediately after Wills/Lenglen and beat Wills/Lenglen a couple of times, I think she would top this list.

Which goes to show that it's not just about your dominance but the dominance of your competition.

This is exactly how it works in chess, apparently. I'm not a chess nerd but I found out from someone who is. :tape:

tennisvideos
Sep 26th, 2011, 12:13 PM
But her competition didn't include another like her. If she came immediately after Wills/Lenglen and beat Wills/Lenglen a couple of times, I think she would top this list.

Which goes to show that it's not just about your dominance but the dominance of your competition.

This is exactly how it works in chess, apparently. I'm not a chess nerd but I found out from someone who is. :tape:

But it's not Connolly's fault who her opponents were ... or that they weren't up to her level - and that is subjective anyway. I think some of her opponents are underrated.

Sam L
Sep 26th, 2011, 12:22 PM
But it's not Connolly's fault who her opponents were ... or that they weren't up to her level - and that is subjective anyway. I think some of her opponents are underrated.
I agree it's not her fault BUT the fact that her opponents didn't dominate at any other period in history is objective.

I think if anything WWII being a factor is bigger. If it weren't for WWII, there certainly might've been someone dominating in the late 40s and she would've had to overcome her.

Fantasio
Sep 26th, 2011, 01:53 PM
Fantasio, thanks for that explanation. And now I am really understanding this. It confirms my suspicions that the reason why Seles is so high is because of her success against Graf.
You're right. Of course, defeating Graf was important because Seles defeated all the other players (that is, was consisent). Sanchez-Vicario also defeated Graf, but was not consistent!

The reason why Connolly who had same Grand Slam success as Seles is, comparatively, so low is because she didn't have high quality competition.
Right again. It was not Connolly's fault, and maybe she could have defeated Court/King/everyone else, but there's no way to prove it. We only know she was stronger than Fry/Hart/Brough, but that's not enough.

To sum up, Seles dominated an era that included a GOAT and her losses came to that GOAT. Comparatively, it was harder for her to have dethroned Graf than anyone else in history and for dethroning Graf, she's rewarded heavily here.
You understood everything. :-)

And to only factor in the slams is very limiting - years at #1, career singles winning % and tournament wins should also factor in there somewhere.
I said many times that dominance level is just one factor. I never said ELO ranking=GOAT ranking and I myself think Navratilova to be the GOAT (nor Graf neither Seles).

Which are? (reasons for not counting "minors")
There are two main reasons.

First one, in tennis there's no official ELO ranking, in chess there is. This is one of the main reasons why top chess players do not compete in minor events: playing weak players is risky, because losing to just one of them would mean losing a lot of points. Beating all the others (in chess most tournaments are round-robin) could not compensate such a loss. In tennis, majors are valued much more than "minors", so top players do compete, often lose, but without consequences. Our Schiavone, for example, did not play seriously in minor events after winning in Paris last year, still never exited top 10. In chess that would not have been possible, and had she been a chess player, she would only have played in "majors". As many top players are Schiavone-like (i.e Williams sisters), it's better to count only majors: weak players will be ranked badly (as they play seriously in minor events) but top players will not. As we are only interested in top players, that's enough.

Second one, we've got no complete results for "minors", and it's nonsense counting them starting from 1968 (or 1974). Or we count them all, either we don't. And even counting them all, the Schiavone/Williams problem will stand. You surely do not want Serena to be ranked lower!

austinrunner
Sep 26th, 2011, 08:13 PM
Results are results. How do you know that Schiavone or the Williams sisters didn't try hard in the events that were not Grand Slam tournaments?

I don't care how Serena Williams (or any other player) is ranked. I'm purely interested in an accurate rating. Ignoring non-Grand Slam tournaments is a huge shortcoming of your system, especially given the fact that at times the "minor" tournaments were considered more important than certain Grand Slam tournaments.

DennisFitz
Sep 27th, 2011, 04:26 AM
I will repeat myself.
In Elo method, dominance duration does NOT count. Only dominance level. Of course Court dominated 5 years, Navratilova also five. Also great was Evert, not mentioning Lenglen and Wills, that you forgot.
All these players could have dominated 20 years or more, there's no difference.

IF you are only looking at dominance level, it's easy to notice that between 1991 and 1993 Seles played 8 majors, won 7, only losing one match to Graf (who is - without any doubt - one of the most likey GOAT candidates). She defeated Graf (twice), Sanchez-Vicario (3 times), old Navratilova (twice), Novotna (once), Sabatini (3 times).
If this is NOT supreme dominance, I don't know how to define such a thing.
Well, Navratilova 1983-1984 dominated. But she "only" won 6 majors, losing to Horvath and Sukova. Losing to a low-ranked player such as Horvath is a disaster, comparing to losign to Graf (in Elo method).

But there's Court in 1969-1971. Well, she's better than Seles, having won 8 majors, losing only to Haydon-Jones - better than Horvath, still no Graf. And Court only defeated King twice, Goolagong twice. Much, bur not as much as Seles. Plus, don't forget she only played 49 matches during 9 majors (winning 48), while Seles played 56, winning 55, despite competing in 8 events. That's why Seles is higher than Court.

Serena played 6 majors, won 5. No match.

Evert played 8, won 6. Very good, and she defeated Goolagong 3 times, Navratilova also 3 times. She only lose to Goolagong and King. That's very good, and she's very close to Seles, but still not equal.

Connolly extraordinary. 9 over 9! Her problem, she did not defeat strong players. Hart, Fry, Brough are top 30 (alltime) at best. And she played 51 matches only.

So, why is Graf ranked higher (more or less, she's equal)? Because between 1988 and 1990 she won 8 majors, played 9, only lose to Sanchez-Vicario (better than Haydon-Jones), defeated Navratilova (3 times), old Evert (3 times), Sabatini (4 times), young Seles (twice), young Sanchez-Vicario (once). I think Seles defeated better opponents, but Graf won more, and more matches (62 over 63).

Now the difference between "dominance level" and "dominance duration" should be clear. Of course I know that Graf dominated more years (there are 1995 and 1996, of course), won much more than Seles, so she's a better GOAT candidate than unlucky Monica. But dominance level is the same.

Or, we can also say that "dominance level" = "strength at best" (if consistent). So, the old question, "was Seles really better than Graf or was Graf not at her best during 1991-1992?" could be settled, as the two players were of equal strength (according to Elo).

OK, I get it now. This is just a dominance rating. If you had a certain level of dominance in the majors, you are ranked high.

I'd have to say it's a nice little statistic. But in the grand scheme if things IMHO, it's not a big deal. I think it's super impressive to have a dominating year or two, the way Djokovic has had this year. I just think that Martina Navratilova's 1983-1984 seasons in the majors, where she won 6 in a row (should she be penalized because the Australian had smaller draw than when Seles played?) is more impressive than Seles 1991-1992 seasons. Martina didn't duck out on one of the majors either.

It's also like saying Chris Evert dominated on clay, winning 125 matches in a row on clay. Erego, she had the top level of dominance in the history of the game. Which is true. Just doesn't mean Chris is the best ever, because she was the best ever on clay.

And majors were majors back in the mid 1970s. But most experts will tell you the VS Championships counted more than a win in the French or Australian.

Fantasio
Sep 27th, 2011, 09:16 AM
How do you know that Schiavone or the Williams sisters didn't try hard in the events that were not Grand Slam tournaments?
No comment.

Ignoring non-Grand Slam tournaments is a huge shortcoming of your system, especially given the fact that at times the "minor" tournaments were considered more important than certain Grand Slam tournaments.
If you consider dominance in majors to be an important factor, you'll appreciate this ranking. If not, you're welcome.
BTW, I agree that the "only majors" approach is a shortcoming, and that's the reason I introduced a lot of minor changes to the parameters of the Elo system (it's no use going into details). Also, I "promoted" (and counted) to majors Fed Cup finals, Masters tournament and old Hard Court (clay) World Championships. For men, I "promoted" other tournaments, such as WCT finals and Irish Championships, but the general idea is to count all finals played best of 5 - if and when they will be available. Women only play best of 3, so things are more complicated. Even if we'd get complete results, to include every single tournament will only complicate matters, and to decide which tournaments are important (for GOAT researches) and which are not is highly subjective.
This is not as easy as it looks, and required many months of hard work before I felt satisfied with the tuning of the (many) parameters. Could I do better? We need complete results before trying anything, as - I said that before - or you count them all either you don't. Only majors is nevertheless a good approach (I know many people firmly stating "minors" count nothing), although it works better for men - best of 5 makes a big difference!
The real problem is: will we ever get complete results? I'm afraid this won't happen. Even old almanacs as Wright/Ditson or Spalding are far from exaustive. So, what to do? One way or another we will be forced to choose which tournaments are important and which are not. But now... the "only majors" approach is the only one available.

Sam L
Sep 27th, 2011, 10:13 AM
Results are results. How do you know that Schiavone or the Williams sisters didn't try hard in the events that were not Grand Slam tournaments?

I don't care how Serena Williams (or any other player) is ranked. I'm purely interested in an accurate rating. Ignoring non-Grand Slam tournaments is a huge shortcoming of your system, especially given the fact that at times the "minor" tournaments were considered more important than certain Grand Slam tournaments.

This has already been talked about but I just want to say again the lack of complete records. Rollo and others are still coming up with research about tournament results nobody knew about or weren't available in public. How will you include everything? :confused: Also, Davis Cup was the most important event for players for a long time. Are we to include those ties? World Team Tennis? Where does it begin and end?

I, for one, am glad that this system only took into account the Grand Slams. The books are called "Grand Slam Record Book", they're about the 4 majors so it's appropriate that it considers dominance at the level and shows a rating at level. Domination at majors is a major factor, if not, the No. 1 factor for greatness anyway. Other things are just secondary.

Whatever you think about this, you can't deny it's not objective.


I'd have to say it's a nice little statistic. But in the grand scheme if things IMHO, it's not a big deal.


I think it's better than any other subjective rating/ranking we have going at the moment. And ask anyone, I'm by no means a fan of Graf. I think Navratilova should be No. 1 too but this is at least an objective way of looking at things without bringing in personal biases.

austinrunner
Sep 28th, 2011, 01:14 AM
Every ranking or "dominance" system ever devised is subjective, including this one. From the decision about which tournaments to include (i.e., what is a "major" tournament?) to the decision about which statistical method to use (e.g., how to tweak ELO), it's entirely subjective. Note the author's admission in this thread that he (or she) "promoted" certain tournaments and events to "major" status, e.g., Federation Cup and Fed Cup finals (but curiously not the semifinals).

The mere fact that we don't (yet) have all the results from every tournament is no reason to ignore 95 percent of the tournaments ever held. That's neither objective nor scientific.

Davis Cup never included women.

Sam L
Sep 28th, 2011, 10:25 AM
Every ranking or "dominance" system ever devised is subjective, including this one. From the decision about which tournaments to include (i.e., what is a "major" tournament?) to the decision about which statistical method to use (e.g., how to tweak ELO), it's entirely subjective. Note the author's admission in this thread that he (or she) "promoted" certain tournaments and events to "major" status, e.g., Federation Cup and Fed Cup finals (but curiously not the semifinals).

The mere fact that we don't (yet) have all the results from every tournament is no reason to ignore 95 percent of the tournaments ever held. That's neither objective nor scientific.

Davis Cup never included women.


Where did he say that Fed Cup results were included? Sorry I couldn't find it anywhere. I thought it was only based on Grand Slam results.

Also, what do you mean what is a "major" tournament? Obviously the Grand Slams. And before you start, yes I know that the Canadian in 1900 may have been more prestigious or the Italian Open in the 60s or whatever else but at the end of the day, these are the four majors that tennis historians and fans have unanimously agreed upon.

That doesn't mean, of course, that other results should count. And the author never said that. But what he is trying to achieve here is to objectively measure dominance in slams all-time and this has been achieved.

It's objective because the elo rating system isn't even unique to tennis. It's created by someone who doesn't even play or like tennis and that's as objective as you can get. More objective than any tennis fan trying to push their favourite by arguing for certain stats to used against others.

This was never meant to be a be-all-end-all solution to GOAT discussions and the author clearly stated that at beginning.

austinrunner
Sep 28th, 2011, 11:55 AM
Also, what do you mean what is a "major" tournament? Obviously the Grand Slams. And before you start, yes I know that the Canadian in 1900 may have been more prestigious or the Italian Open in the 60s or whatever else but at the end of the day, these are the four majors that tennis historians and fans have unanimously agreed upon.

There is no such unanimity.

Where did he say that Fed Cup results were included? Sorry I couldn't find it anywhere. I thought it was only based on Grand Slam results.But what he is trying to achieve here is to objectively measure dominance in slams all-time and this has been achieved.

It's objective because the elo rating system isn't even unique to tennis. It's created by someone who doesn't even play or like tennis and that's as objective as you can get. More objective than any tennis fan trying to push their favourite by arguing for certain stats to used against others.

Refer to the following quotation. Also note that the author's system is not the same as used in chess. The author tweaked it in undisclosed ways.

BTW, I agree that the "only majors" approach is a shortcoming, and that's the reason I introduced a lot of minor changes to the parameters of the Elo system (it's no use going into details). Also, I "promoted" (and counted) to majors Fed Cup finals, Masters tournament and old Hard Court (clay) World Championships.

chris whiteside
Sep 28th, 2011, 12:00 PM
Where did he say that Fed Cup results were included? Sorry I couldn't find it anywhere. I thought it was only based on Grand Slam results.

Also, what do you mean what is a "major" tournament? Obviously the Grand Slams. And before you start, yes I know that the Canadian in 1900 may have been more prestigious or the Italian Open in the 60s or whatever else but at the end of the day, these are the four majors that tennis historians and fans have unanimously agreed upon.

That doesn't mean, of course, that other results should count. And the author never said that. But what he is trying to achieve here is to objectively measure dominance in slams all-time and this has been achieved.

It's objective because the elo rating system isn't even unique to tennis. It's created by someone who doesn't even play or like tennis and that's as objective as you can get. More objective than any tennis fan trying to push their favourite by arguing for certain stats to used against others.

This was never meant to be a be-all-end-all solution to GOAT discussions and the author clearly stated that at beginning.



Previous post from Fantasio.

I agree with most of what you say but funnily enough was going to ask what you thought now that it transpires it is not just Slams alone.


No comment.


If you consider dominance in majors to be an important factor, you'll appreciate this ranking. If not, you're welcome.
BTW, I agree that the "only majors" approach is a shortcoming, and that's the reason I introduced a lot of minor changes to the parameters of the Elo system (it's no use going into details). Also, I "promoted" (and counted) to majors Fed Cup finals, Masters tournament and old Hard Court (clay) World Championships. For men, I "promoted" other tournaments, such as WCT finals and Irish Championships, but the general idea is to count all finals played best of 5 - if and when they will be available. Women only play best of 3, so things are more complicated. Even if we'd get complete results, to include every single tournament will only complicate matters, and to decide which tournaments are important (for GOAT researches) and which are not is highly subjective.
This is not as easy as it looks, and required many months of hard work before I felt satisfied with the tuning of the (many) parameters. Could I do better? We need complete results before trying anything, as - I said that before - or you count them all either you don't. Only majors is nevertheless a good approach (I know many people firmly stating "minors" count nothing), although it works better for men - best of 5 makes a big difference!
The real problem is: will we ever get complete results? I'm afraid this won't happen. Even old almanacs as Wright/Ditson or Spalding are far from exaustive. So, what to do? One way or another we will be forced to choose which tournaments are important and which are not. But now... the "only majors" approach is the only one available.

Sam L
Sep 28th, 2011, 12:11 PM
There is no such unanimity.



Refer to the following quotation. Also note that the author's system is not the same as used in chess. The author tweaked it in undisclosed ways.

Previous post from Fantasio.

I agree with most of what you say but funnily enough was going to ask what you thought now that it transpires it is not just Slams alone.

God I was searching for Fed Cup on every page but this one. Someone's not thinking outside the square today.

Anyway, I think it proves there's subjectivity in the parameters but that was obvious from the beginning. If you did only open era, you'll get a different result.

I still don't think it's possible to count all minors with the information available since we can't even rely on some of the information. For example, how many sources actually can prove some of these results from a century ago?

I would've preferred to have seen one just on Grand Slams alone but I think in the case of the very top, not sure how big a change Fed Cup, WHCC and YEC(?) were going to make anyway.

chris whiteside
Sep 28th, 2011, 12:12 PM
Every ranking or "dominance" system ever devised is subjective, including this one. From the decision about which tournaments to include (i.e., what is a "major" tournament?) to the decision about which statistical method to use (e.g., how to tweak ELO), it's entirely subjective. Note the author's admission in this thread that he (or she) "promoted" certain tournaments and events to "major" status, e.g., Federation Cup and Fed Cup finals (but curiously not the semifinals).

The mere fact that we don't (yet) have all the results from every tournament is no reason to ignore 95 percent of the tournaments ever held. That's neither objective nor scientific.

Davis Cup never included women.

I have to agree with austinrunner here - although personally I wish you'd just left it at Slams. I think you've probably devalued it slightly by going further because then it does not become as objective.

Of course, I originally thought it was almost events which when you think of it is almost impossible - further results keep turning up all the time.

I am uneasy about including Fed Cup because you can advance to the final even though you lose your matches but your team-mates win e.g. Billie-Jean King lost to Ann Jones in the 1966 semi but advanced because Julie Heldman won her match and then they won the doubles.

Rollo awarded ranking points for the Fed Cup and as a result of this it put BJK at #1 on his list and for the overall panel ranking otherwise Maria Bueno would have finshed as #1 - I always felt this was dubious.

I know you're not claiming it to be definitive but I do believe it would have had more relevance based on Slams only rather than adding some other events as the Slams are the cornerstone of the game.

Sam L
Sep 28th, 2011, 12:16 PM
I have to agree with austinrunner here - although personally I wish you'd just left it at Slams. I think you've probably devalued it slightly by going further because then it does not become as objective.

Of course, I originally thought it was almost events which when you think of it is almost impossible - further results keep turning up all the time.

I am uneasy about including Fed Cup because you can advance to the final even though you lose your matches but your team-mates win e.g. Billie-Jean King lost to Ann Jones in the 1966 semi but advanced because Julie Heldman won her match and then they won the doubles.

Rollo awarded ranking points for the Fed Cup and as a result of this it put BJK at #1 on his list and for the overall panel ranking otherwise Maria Bueno would have finshed as #1 - I always felt this was dubious.

I know you're not claiming it to be definitive but I do believe it would have had more relevance based on Slams only rather than adding some other events as the Slams are the cornerstone of the game.
But hold on, you're not getting points for reaching the Fed Cup final, you're getting points for beating the player in the Fed Cup final.

In your example, the fact that they advanced to the final has no bearing on BJK's ranking.

The fact that she lost to Ann Jones is all that matters.

The important thing is about match ups against other players. You don't get points for achieving something.

So your example about Rollo's results changing won't happen here.

Fantasio
Sep 28th, 2011, 03:11 PM
But hold on, you're not getting points for reaching the Fed Cup final, you're getting points for beating the player in the Fed Cup final.

In your example, the fact that they advanced to the final has no bearing on BJK's ranking.
The fact that she lost to Ann Jones is all that matters.

The important thing is about match ups against other players. You don't get points for achieving something.
I could not say that better.
(btw, where's Keira gone? I liked her much more!)

Now I will address the points.

First, the "only-slam-ELO-ranking":

1) Graf 2748
2) Seles 2713
3) Wills 2709
4) Evert and Navratilova beautifully paired 2696
6) Smith-Court 2659
7) Lenglen 2652
8) Connolly 2650
9) Hingis 2646
10) Moffitt-King 2643
11) Serena 2630
12) Venus 2608
13) Goolagong 2601
14) Henin 2594
15) Bueno 2566
16) Sanchez-Vicario 2564
17) Cljisters 2564
18) Marble 2560
19) Jacobs 2538
20) Mandlikova 2535
and so on.

Everyone should be able to evaluate the differences, and decide for him/herself if they are significant or not.

Second, the "minors" problem.
What amuses me, every time I read someone debating this issue, is the impossibility to compromise. Last year Wozniacki was number 1, according to ATP. But many people I know firmly state that Serena was number 1, followed by Cljisters, then by Schiavone (who also won Fed Cup). Poor Caroline just 4th, given the fact that not only she didn't win any major, but not even reached a single final - and just one semifinal.
You can say that ATP rankings are too minors-oriented, but no matter what your ranking system is, sooner or later a Wozniacki-like case will occur. And in such a case, compromise is not possible. Or you count minors and accept Wozniacki number 1, either you don't. Some of you will say that Wozniacki number 1 is an abomination, some others will say that minors cannot be discarded. No compromise.

So we can discuss until the end of the world if minors should count or not. It's no use. Instead I chose: first, there are not (and never will) complete results for every tennis tournament; second, in chess top players do not compete in minor events - never - and I think this is exactly what tennis top players would do IF Elo ranking system were in use. So I chose the approach "only majors". But of course I cannot discharge what some of you said, that in the past "majors" and "slams" were not the same. If I want to do some serious GOAT research, the approach has to be different. How could I evaluate professionals, if I only counted "slams"? Gonzales would drop to 40-50th, for example (and we're lucky women - almost - did not deal with such a problem). A GOAT research, even if limited to "majors", has to count all "majors", slam or not. Otherwise it would just be a funny joke (men only-slam-Elo-ranking has Tilden as number 1, followed by Borg, then Lacoste).
That's why I decided to "promote" to "majors" some historical tournaments, Davis Cup (and Fed Cup) included. I was helped by Mazak, who already has compiled a list of major-like tournaments for each year, and his list (at least 90% of it) was of the greatest importance for me. Without "promotion" there wouldn't have been a serious men ranking; without men ranking there wouldn't have been a women ranking.

BTW, we addressed this historical problem in the books, too. For most "major-like" tournaments we reported the winner and in the future we hope to include complete draws for such events (if there will be other editions).

So, at least for myself, the only doubt concerns the "major-like" tournaments. Did I include them all? Did I include too many? I'm not sure of that, and maybe never will.

Sam L
Sep 30th, 2011, 03:41 PM
I could not say that better.
(btw, where's Keira gone? I liked her much more!)


Grazie. It's Fanny Tacchinardi Persiani, your compatriot. ;) But I like Keira too, maybe I will bring her back. :cool:

Thanks also for the grand slam only list. Very interest.

thrust
Oct 8th, 2011, 01:03 AM
Alessandro - I hope - will know that!

After US 2011, that's the new all-time ELO ranking, top 50 (but no change occurred at this level):

1) Graf 2760
2) Seles 2751
3) Navratilova 2728
4) Wills-Moody 2721
5) Evert 2699
6) Lenglen 2680
7) Smith-Court 2670
8) Hingis 2668
9) Connolly 2650
10) Moffitt-King 2643
11) S. Williams 2637
12) V. Williams 2612
13) Henin 2610
14) Goolagong-Cawley 2601
15) Cljisters 2577
16) Bueno 2566
17) Marble 2560
18) Sanchez-Vicario 2555
19) Davenport 2549
20) Austin 2548
21) Jacobs 2540
22) Mandlikova 2529
23) Sabatini 2525
24) Bjurstedt-Mallory 2525
25) Gibson 2520
26) Novotna 2517
27) Hart 2514
28) Sharapova 2513
29) Osborne-DuPont 2510
30) Richey-Gunter 2509
31) Krahwinkel-Sperling 2507
32) Brough 2504
33) Round 2499
34) Douglass-Lambert-Chambers 2498
35) Pierce 2493
36) Haydon-Jones 2489
37) Mauresmo 2489
38) Capriati 2481
39) Fry 2477
40) Martinez 2469
41) Betz 2468
42) McKane-Godfree 2461
43) Hard 2461
44) Passemard-Mathieu 2459
45) Sukova 2456
46) Ivanovic 2453
47) Turner-Bowrey 2448
48) Jaeger 2441
49) Shriver 2437
50) Li 2436

20- points' difference is not that important, BTW.

Sorry, but I dont't understand what this is about. Hingis above Connolly, just under Court?

Andy T
Oct 8th, 2011, 12:22 PM
Small favour: please could someone who has the book do a quick look-up to see if any mention is made of who was the #3 seed at the Australian Open of 1971 (between Goolagong & Chanfreau) but withdrew? Thanks! Andy

Rollo
Oct 10th, 2011, 12:55 PM
Small favour: please could someone who has the book do a quick look-up to see if any mention is made of who was the #3 seed at theAustralianOpen (http://www.tennisforum.com/showthread.php?t=442899&page=4#) of 1971 (between Goolagong & Chanfreau) but withdrew? Thanks! Andy


Hi Andy-I have the book (someone just gifted me with it) and checked. There is no mention of the #3 seed.

The book is a wonderful resource. There has never been a printed source for all the slam results for men or women. Now we have one. I'm sure it will add a lot to what we don't have here.:worship:

Despite the comprehensive nature of the book there is still a place for our results here in the Blast. For one thing new information allows us to edit all the time. I have quietly been working on reconfiguring our slam draws.

And wonderful as the book is, it has errors. Not a lot from what I can tell, but they are there.

Andy T
Oct 10th, 2011, 01:57 PM
Hi Andy-I have the book (someone just gifted me with it) and checked. There is no mention of the #3 seed.

The book is a wonderful resource. There has never been a printed source for all the slam results for men or women. New we have one. I'm sure it will add a lot to what we don't have here.:worship:

Despite the comprehensive nature of the book there is still a place for our results here in the Blast. For one thing new information allows us to edit all the time. I have quietly been working on reconfiguring our slam draws.

And wonderful as the book is, it has errors. Not a lot from what I can tell, but they are there.

Thanks for looking, Rollo. It's wonderful such a book has been published and I will have to invest in a copy for myself, I think. Shame that this particular mystery was not cleared up, though!

I totally agree that the resource being built up on BFTP, containing info regarding slams and both high and low profile regular tournaments is invaluable. Keep up the good work!

Fantasio
Oct 10th, 2011, 04:44 PM
There is no mention of the #3 seed.
I suspect - but am far from sure - there could be a mistake in the "official" draws, and #3 seed was Turner-Bowrey. It's just a suspect, but all the 1971 draws I see over the Internet obviously come from a same source. A mistake in that source, and the legend arises.

And wonderful as the book is, it has errors. Not a lot from what I can tell, but they are there.
If you are certain, please notify me or Alessandro of such errors. I hope a new, error-proof, software inclusive edition will be printed sooner or later.

Andy T
Oct 10th, 2011, 07:38 PM
I suspect - but am far from sure - there could be a mistake in the "official" draws, and #3 seed was Turner-Bowrey. It's just a suspect, but all the 1971 draws I see over the Internet obviously come from a same source. A mistake in that source, and the legend arises.
Ciao!
This would seem unlikely to me. The Victorian champs in late January had Court-Goolagong-Harris-Hunt-Gourlay-Young as top 6 domestic seeds and Shaw-Morozova-Hogan-Chanfreau-Walsh & Stove - Sawamatsu and Roussow as top 8 foreign seeds.
Comparing this with the "received" version of the Aussie Open line-up (Court-Goolagong-?-Chanfreau-Harris-Hogan-Shaw-Gourlay-Bowrey, with Walsh, Hunt & Young unseeded), I have an inkling that Sawamatsu (who was a SF in Victoria), Stove or, most probably, Morozova, (who had made the NSW finals earlier in January), pulled out.

chris whiteside
Oct 12th, 2011, 11:40 AM
But hold on, you're not getting points for reaching the Fed Cup final, you're getting points for beating the player in the Fed Cup final.

In your example, the fact that they advanced to the final has no bearing on BJK's ranking.

The fact that she lost to Ann Jones is all that matters.

The important thing is about match ups against other players. You don't get points for achieving something.

So your example about Rollo's results changing won't happen here.


Sorry, I do understand the system. This was me going off at a tangent and really nothing to do with the parameters of the system besides a very obscure point about choosing to include some matches outside the Slams which perhaps shouldn't have occurred.

However, after reflection I have a few reservations about the method used.

If you win a Slam without beating the top players this will not reflect your achievement in your score.

I have oft heard said that someone's Slam didn't count as much because they only had to beat "so and so".

There may be some merit to this if top players were missing from the field.

However take 1962 Wimbledon for example.

Probably one of the best Slam fields ever assembled:

Margaret Smith already dominating the game - but surely her score would not be reflecting that yet? - Maria Bueno, Darlene Hard, Billie-Jean King, Angela Mortimer, Ann Haydon, Rennee Schuurman, Sandra Reynolds the finalist from 1960 and semi-finalist from 1961 still among the top players but who couldn't get a place in the seedings.

Karen Susman won by "only" beating Schuurman, Haydon and in the final Vera Sukova. That wouldn't advance her points total spectacularly.

The fact is that players like Smith, Bueno and Hard were not good enough to get to the semis or final - there were better players than them at this tournament.

However, at the end Karen Susman was the last player standing in an event which included all the best players at the time so to me she has beaten all the players in the field - her victory is the same as if she HAD beaten them personally.

austinrunner
Oct 12th, 2011, 12:11 PM
Every tennis rating system I've ever seen has insurmountable flaws, is incomplete, and is riddled with subjectivity. Perhaps Jeff Sagarin, the U.S. college basketball rating guru, should be hired for tennis. He's about as good as it gets.

Fantasio
Oct 12th, 2011, 12:26 PM
However, at the end Karen Susman was the last player standing in an event which included all the best players at the time so to me she has beaten all the players in the field - her victory is the same as if she HAD beaten them personally.
Most people I know strongly disagree. I debated this issue many years ago, when I first tried something in this field. At the time I agreed with you, and thought that "indirect victory" was as important as a "direct" one, but this idea was largely mocked.
In the end I compromised. The ELO system is an acceptable - far from perfect, of course - compromise.

Sam L
Oct 24th, 2011, 10:26 AM
The book is a wonderful resource. There has never been a printed source for all the slam results for men or women. Now we have one. I'm sure it will add a lot to what we don't have here.:worship:


I too have this now. The best thing about this is that you can see the results from the 19th century. I didn't know that in the early years before the final they only played to 6-5. There was no advantage! I didn't know that. That is exactly like how real tennis still has it. Well, you learn something new every day.

Every tennis rating system I've ever seen has insurmountable flaws, is incomplete, and is riddled with subjectivity. Perhaps Jeff Sagarin, the U.S. college basketball rating guru, should be hired for tennis. He's about as good as it gets.

But that's the problem. ELO is NOT a tennis rating system. It's used for several sports like Football and Basketball (so they tell me). It's not perfect but it's the best we're going to get. And it is very logical.

The only flaw perhaps with this particular list is that it includes the slams and some other matches. I think it would've been better to have either slams only or everything. But seeing as it's impossible to have everything, I think it should've been only slams. However, I do appreciate the work the authors have put in to create a more complete system. Also, the results didn't change by much.

austinrunner
Oct 24th, 2011, 11:22 AM
Now you're basing your argument on semantics. Is it a "rating system" or isn't it. Maybe we would agree that it's a "ranking system" based on dominance. Aside from that, ELO can be used as a power rating system, which is how it is used in college football and basketball. There's no reason that ELO could not be used similarly for tennis.

And I still disagree that a whatever-you-want-to-call-it system should exclude all known non-Grand Slam tournament matches just because the results of perhaps 50 percent of those matches are not yet known. Every historical database grows with time, as does scientific knowledge. That does not stop statisticians and scientists from saying "here is what we know today."

Sam L
Oct 24th, 2011, 11:39 AM
Now you're basing your argument on semantics. Is it a "rating system" or isn't it. Maybe we would agree that it's a "ranking system" based on dominance. Aside from that, ELO can be used as a power rating system, which is how it is used in college football and basketball. There's no reason that ELO could not be used similarly for tennis.

And I still disagree that a whatever-you-want-to-call-it system should exclude all known non-Grand Slam tournament matches just because the results of perhaps 50 percent of those matches are not yet known. Every historical database grows with time, as does scientific knowledge. That does not stop statisticians and scientists from saying "here is what we know today."

It is a rating system but it's not developed for tennis or any sport unlike other systems you were talking about?

In any case, it's better than looking down the list of GS winners and arguing for our favourites?

Some of those results are never going to be known. Also, how can you even determine what are actual tournaments or not before the advent of pro tennis? And do you include pro tennis results from before the pro era? How about events like World Tennis Tennis? Where do you draw the line between tournament and exhibition? Do we include everything? Where do you draw the line?

austinrunner
Oct 24th, 2011, 11:44 AM
It is a rating system but it's not developed for tennis or any sport unlike other systems you were talking about?

In any case, it's better than looking down the list of GS winners and arguing for our favourites?

Some of those results are never going to be known. Also, how can you even determine what are actual tournaments or not before the advent of pro tennis? And do you include pro tennis results from before the pro era? How about events like World Tennis Tennis? Where do you draw the line between tournament and exhibition? Do we include everything? Where do you draw the line?

I don't understand your first question. Please rephrase it or explain.

We don't include World Team Tennis. We include all sanctioned tournaments by, e.g., the USLTA, the FTF, and other national governing bodies. With women's tennis, there were few professional matches. We are constantly finding new match results. The percentage of unknown results is going down steadily. We will never get to 100 percent coverage, but we can get very close to that.

Sam L
Oct 24th, 2011, 11:54 AM
I don't understand your first question. Please rephrase it or explain.

We don't include World Team Tennis. We include all sanctioned tournaments by, e.g., the USLTA, the FTF, and other national governing bodies. With women's tennis, there were few professional matches. We are constantly finding new match results. The percentage of unknown results is going down steadily. We will never get to 100 percent coverage, but we can get very close to that.

Sorry I meant it is a rating system but it's not developed specifically for tennis or any sport. This is unlike other systems you were talking about in your above post which were specifically developed for tennis?

Well there you go. Didn't Chrissie skip the Aussie and French Opens to play in World Team Tennis? Wouldn't this be disadvantaging her?

What were all sanctioned tournaments and by whom? Didn't the Americans and the Europeans have a big argument over the World Championships? WHCC, WCCC?

austinrunner
Oct 24th, 2011, 12:32 PM
Sorry I meant it is a rating system but it's not developed specifically for tennis or any sport. This is unlike other systems you were talking about in your above post which were specifically developed for tennis?

Well there you go. Didn't Chrissie skip the Aussie and French Opens to play in World Team Tennis? Wouldn't this be disadvantaging her?

What were all sanctioned tournaments and by whom? Didn't the Americans and the Europeans have a big argument over the World Championships? WHCC, WCCC?

I'm still not understanding you or you are still not understanding me. There already is in use an ELO rating system for college football and basketball. There is no reason that the same system could not be used for tennis players.

Sure, but the objective is not to protect your favorite players from the consequences of their decisions.

They agreed at what was sanctioned. What they didn't agree about is whether they should remained sanctioned in the future. The US position prevailed.

Sam L
Oct 24th, 2011, 12:44 PM
I'm still not understanding you or you are still not understanding me. There already is in use an ELO rating system for college football and basketball. There is no reason that the same system could not be used for tennis players.

Sure, but the objective is not to protect your favorite players from the consequences of their decisions.

They agreed at what was sancioned. What they didn't agree about is whether they should remained sanctioned in the future. The US position prevailed.

No but the point is who makes the call to include WTT or not? And we talk about women's tennis here but this is about men's tennis also. I mean the issue of greatness is also there so how about pro tennis events before 1968?

I don't understand your point. This is an ELO rating system that we are talking about. I am advocating for this system. My point is that this is probably the fairest system out there. I don't know of any other systems but you mentioned there were others and I'm guessing they were created specifically for tennis.

austinrunner
Oct 24th, 2011, 04:50 PM
No but the point is who makes the call to include WTT or not? And we talk about women's tennis here but this is about men's tennis also. I mean the issue of greatness is also there so how about pro tennis events before 1968?

I don't understand your point. This is an ELO rating system that we are talking about. I am advocating for this system. My point is that this is probably the fairest system out there. I don't know of any other systems but you mentioned there were others and I'm guessing they were created specifically for tennis.

This is a women's tennis forum. What is done for women doesn't have to correspond with what is done for men.

There is not "one type" of ELO rating system. The authors of The Grand Slam Book tweaked the ELO system used for chess to make it fit what they wanted for tennis. As I have already said several times, an ELO system underlies some college football and basketball rankings. The football-basketball rankings count every game and score and provide a ranking where one can easily see the quality differences between teams. That is the system I think should be used for tennis.

Sam L
Oct 25th, 2011, 11:26 AM
This is a women's tennis forum. What is done for women doesn't have to correspond with what is done for men.

There is not "one type" of ELO rating system. The authors of The Grand Slam Book tweaked the ELO system used for chess to make it fit what they wanted for tennis. As I have already said several times, an ELO system underlies some college football and basketball rankings. The football-basketball rankings count every game and score and provide a ranking where one can easily see the quality differences between teams. That is the system I think should be used for tennis.

This may be a women's forum but those books and those lists were developed with men's tennis in mind also.

The simple fact of the matter is that there is clear way to draw a line between what should be included and what shouldn't be. At least, what is a major is and what are important events are clearer.

Football and Basketball count every game because they're using that for scores right now. There's no dispute about what are the official games and what are not.

As has been said by one of the authors already counting every single match will push someone like Serena Williams even down further in the list since her GS/big match record is far superior to her minor tournament record.

This is about a measure of GOAT and dominance of the biggest matches and titles, not a WTA ranking system.

It is about the quality differences between players but only limited to the biggest matches and biggest events. That I think is fairer.

Did you know that some Chess players refuse to play in small or unimportant events because of the risk of losing to someone with a low ELO ranking? The system heavily punishes losses to those with a low ranking. That's not possible for tennis players and therefore it is perfectly reasonable to only count the biggest matches because it is a measure of their best against their peers' best.

chris whiteside
Oct 25th, 2011, 12:26 PM
Most people I know strongly disagree. I debated this issue many years ago, when I first tried something in this field. At the time I agreed with you, and thought that "indirect victory" was as important as a "direct" one, but this idea was largely mocked.
In the end I compromised. The ELO system is an acceptable - far from perfect, of course - compromise.

In the main I agree that the ELO is about as good a system as we are liable to get. I would have a few things different but that's just personal.

I appreciate I am a minority on my views regarding "direct" and "indirect" victories. I strongly stick by it though.

You win a Slam at which all the main competiton is playing...................

Another point I ahve been thinking about and I hope my reasoning is correct.

Is the players score not in arrears?

For example, take Margaret Court. From the beginning of 1962 she had a dominant period when she only lost a handful of matches. One of these was, of course, to BJK at Wimbledon. But her ELO rating would not be that big since she had only been competing from 1961 at the Slams (Australian in 1960) and hadn't done that well to date.

In 1968 when she came out of retirement I assume her rating will continue from where she left off in 1966 which would surely be higher than in 1962 since she had had significant success in the ensuuing years.

She took some while to get back into her swing and didn't do particularly well that year so her rating would drop back.

Then from the beginning of 1969 she was again dominant winning 7 of the 8 Slams in 69/70.

So while it was a much better victory to beat her in 1962 or 1969 than in 1968, beating her in 1968 actually brings more reward to the victor? Her actual from at the time is not being taken into account? Or is my reasoning wrong?

chris whiteside
Oct 25th, 2011, 12:35 PM
The argument on ranking/rating could go on ad infinitum.

I think the ELO system has a lot to commend it but instinctively agree with SamL that it would have more merit in sticking just to the Slams rather than including selective other matches.

However, I can also see where austinrunner is coming from in saying that all matches should be included.

The problem here would surely be that you would have to start making adjustments which would immediately make it subjective. It is unanimously accepted that a victory over a player at a Slam is a much bigger deal than one in most other events.

However

Sam L
Oct 25th, 2011, 12:41 PM
For example, take Margaret Court. From the beginning of 1962 she had a dominant period when she only lost a handful of matches. One of these was, of course, to BJK at Wimbledon. But her ELO rating would not be that big since she had only been competing from 1961 at the Slams (Australian in 1960) and hadn't done that well to date.

In 1968 when she came out of retirement I assume her rating will continue from where she left off in 1966 which would surely be higher than in 1962 since she had had significant success in the ensuuing years.

She took some while to get back into her swing and didn't do particularly well that year so her rating would drop back.

Then from the beginning of 1969 she was again dominant winning 7 of the 8 Slams in 69/70.

So while it was a much better victory to beat her in 1962 or 1969 than in 1968, beating her in 1968 actually brings more reward to the victor? Her actual from at the time is not being taken into account? Or is my reasoning wrong?

From what I understand, that is correct. It doesn't consider "form" for sure. And if a player doesn't play their rating is frozen. So if a player beats Margaret in 1968 she would have a higher reward than if a player beats her in 1970 IF Margaret lost to players with bad ratings in 1968. If they didn't have really bad ratings Margaret's rating probably wouldn't be affected so much. The theory is that considering Margaret's pre-1968 success she should have continued success in 1968 and if she doesn't, and if she doesn't especially against bad players, she will be lose her rating.

Fantasio
Nov 10th, 2011, 09:49 AM
Sorry for the delay.

Football and Basketball count every game because they're using that for scores right now. There's no dispute about what are the official games and what are not.
This is a very important point, often underestimated and misunderstood.
When I apply the Elo system to tennis, I'm not trying to build some alternate ranking system. It would be impossibile because, unlike Football and Basketall (and Chess, of course) the Elo system is NOT in use in tennis, and players did NOT play according to it.
As I wrote before, top chess players do not compete in minor events because Elo system is in use, and there's no difference between "majors" and "minors": as top players are more interested in "majors", they cannot be (and are not) in top form if and when playing "minors", so if they play "minors" are likey to lose against weak players (who know they cannot win "majors", so are in top form in "minors" instead). As a loss against a weak player would be a disaster, and in minor tournaments there are a lot of them, top players choose not play such events and concentrate their efforts on big events.
In tennis there is no Elo system, and Williams sisters can play "minors" without serious consequences. As I wrote before, Schiavone staied in top 10 for a very long time despite very poor results in "minors". In chess that would have been impossible.
So, why do I apply Elo to tennis? Just in order to evaluate dominance, which is an important parameter for GOAT research. But I can't forget the problem of top players playing "minors" instead, and with poor results, so a line must be drawn somewhere between "majors" and "minors": we lose accuracy for not-top-players, of course, but we gain accuracy for top ones - and these are the players we are interested!
The real problem, as you correctly notice, is to "where" draw the line. For men's tennis a possibile solution would be "only best of five", but for women's tennis it isn't.

So we have to live with doubts, sorry. :-)

Fantasio
Nov 10th, 2011, 10:08 AM
So while it was a much better victory to beat her in 1962 or 1969 than in 1968, beating her in 1968 actually brings more reward to the victor? Her actual from at the time is not being taken into account? Or is my reasoning wrong?
You're right. That's the reason Elo system only works if we count a lot of matches: the system is able to compensate for over and under-rating, but not within a couple of matches.

austinrunner
Nov 10th, 2011, 11:54 AM
But I can't forget the problem of top players playing "minors" instead, and with poor results, so a line must be drawn somewhere between "majors" and "minors": we lose accuracy for not-top-players, of course, but we gain accuracy for top ones - and these are the players we are interested!
The real problem, as you correctly notice, is to "where" draw the line. For men's tennis a possibile solution would be "only best of five", but for women's tennis it isn't.

So we have to live with doubts, sorry. :-)

Precisely. That's why your system is not any better than any other that's been devised to date.

Fantasio
Nov 10th, 2011, 01:18 PM
Precisely. That's why your system is not any better than any other that's been devised to date.
You forget "my" system is not a GOAT calculation, but only a "dominance in majors" calculation. I have written many times that I myself reckon Navratilova, not Graf, to be the women's GOAT.

austinrunner
Nov 10th, 2011, 10:30 PM
You are wrong. I didn't forget. Your system is still riddled with subjectivity and uncertainty.

Fantasio
Nov 11th, 2011, 12:03 AM
You are wrong. I didn't forget. Your system is still riddled with subjectivity and uncertainty.
Well, in this case you are welcome to try and do better than me. Once you achieve objectivity and certainty, I'll bow down and call you master (c).

Fantasio
Dec 30th, 2011, 12:52 AM
And wonderful as the book is, it has errors. Not a lot from what I can tell, but they are there.
I'm still waiting. If there are errors, I will correct all of them, but at this moment not a single one has been notified to me.

Fantasio
Jan 29th, 2012, 04:37 PM
In the new Elo rankings after Australian 2012, Kvitova went from 59th to 47th, just after Ivanovic. No other variations among the first 60 players.

Rollo
Jan 30th, 2012, 05:35 PM
Originally Posted by Rollo http://imgsrv2.tennisuniverse.com/wtaworld/images/buttons/blue/viewpost.gif (http://www.tennisforum.com/showthread.php?p=20345225#post20345225)
And wonderful as the book is, it has errors. Not a lot from what I can tell, but they are there.


Posted by Fantasio I'm still waiting. If there are errors, I will correct all of them, but at this moment not a single one has been notified to me.



Sorry not to respond earlier Fantasio. I've been gone for a while

IAlso, note my intentions here (especially post 1).
http://www.tennisforum.com/showthread.php?t=447550


With that in mind I'll start with the 1922 Australian Championships.

We have

Singles (Draw=15)

1st Round

Mall Molesworth d. Annie Gray 7-5 6-3
Marjorie Mountain d. Mrs GB Stevens 6-4 6-3
Syd Carr d. Miss Rock 6-2 1-6 8-6
Gwen Utz d. Esme Doddemeade 6-4 6-2

Esna Boyd d. Floris St.George 6-4 6-1
Jessie Watson d. Mrs M Grant 6-3 6-2
Sylvia Lance d. Lorna Bull 6-3 6-1
Mary Elliott-bye

Quarterfinals

Molesworth d. Mountain 6-4 6-4
Utz d. Carr default
Boyd d. Watson 6-3 6-2
Lance d. Elliot 6-1 6-2

Semifinals

Boyd d. Lance 6-4 10-8
Molesworth d. Utz 6-2 6-3

Final
Mall Molesworth d. Esna Boyd 6-3 10-8
-------------------------------------------------

Doubles (Draw=9)

Preliminary Round

Annie Gray/Nellie Lloyd d. Lorna Bull/Mary Elliott 6-4 6-3

Quarterfinals

Esna Boyd/Marjorie Mountain d. Mrs Docker/Miss Schultze 6-1 6-4
Mall Molesworth/Jessie Watson d. Syd Carr/Esme Doddemeade 7-5 4-6 6-3
Irene Forbes-Smith/Sylvia Lance d. Gray/Lloyd 5-7 13-11 6-2
Floris St.George/Gwen Utz d. Daisy Garland/D Menzies 6-3 6-2

Semifinals

Boyd/Mountain d. Molesworth/Watson 7-5 6-3
St.George/Utz d. Forbes Smith/Lance default

Final
Esna Boyd/Marjorie Mountain d. Floris St.George/Gwen Utz 1-6 6-4 7-5
-------------------------------------------------

Mixed Doubles (Draw=15)

1st Round

Gwen Utz/Harold Utz-bye
Sylvia Lance/Stewart Henderson d. Mary Elliott/Jack Clemenger 6-0 6-0
Marjorie Mountain/Rupert Wertheim d. Lorna Bull/Keith Poulton 3-6 6-4 6-3
Hazel Webber/Arthur Yencken d. Minnie Goodman/Bruce Dive 6-0 9-7

Esna Boyd/John Hawks d. Syd Carr/Dudley Bullough 6-4 6-0
Mall Molesworth/Leslie Baker d. Mitchell/Prosper Sandral 6-1 7-5
Jessie Watson/E Jordan d. Grace Paterson/William Sheehan 6-4 6-3
Annie Gray/Percy Lane d. Floris St. George/Frank Peach 8-6 5-7 10-8

Quarterfinals

Utz/Utz d. Lance/Henderson default
Mountain/Wertheim d. Webber/Yencken 6-0 6-2
Boyd/Hawkes d. Molesworth/Baker 6-2 5-7 6-2
Watson/Jordan d. Gray/Love ????? (probable default)

Semifinals

Utz/Utz d. Mountain/Wertheim 6-3 6-3
Boyd/Hawkes d. Watson/Jordan 6-3 6-3

Final
Esna Boyd/Jack Hawkes d. Gwen Utz/Harold Utz 6-1 6-1
-------------------------------------------------

Draw notes and resolves:

The official site does not have the complete draw for any of the five events. It also lists our 2 missing Mixed QF results as byes, clearly an oversight. It also has an incorrect doubles semifinal result. All results stemming from this site should be double-checked.

Wednesday 6th and Thursday 7th washed out and tournament didn't finish until Saturday 9th December. The backup in play explains the defaults in the doubles events.

The Sydney Morning Herald, in its Dec 9 edition, indicated that Miss Lance had to default in the women's doubles and the mixed with her partner Henderson. The reason given was her departure for New Zealand. Note that the rand Slam book and official Aussie sites have a score for this semi.However, unless these scores can be verified in another source (such as anewspaper) the default in the semis of the doubles is the likely score. See the following:
http://trove.nla.gov.au/ndp/del/arti...ore*%7C*ignore*

Rollo
Jan 30th, 2012, 05:47 PM
In comparing that to your published results for 1922 I see the following:


For singles you are missing these first names:

Annie Gray 7-5 6-3
Mary Elliott-bye

There are errors in your doubles scores

(missing)
Preliminary Round

Annie Gray/Nellie Lloyd d. Lorna Bull/Mary Elliott 6-4 6-3

(missing or incorrect first names)
Quarterfinals

Floris St.George/Gwen Utz d. Daisy Garland/D Menzies 6-3 6-2 (*not Peg Menzies)

(Wrong score)
Semifinals

St.George/Utz d. Forbes Smith/Lance default (*and not 6-3 6-2-see link to newspaper article mentioning default.)


Missing first names for the following in mixed doubles:

Gwen Utz/Harold Utz-bye
Sylvia Lance/Stewart Henderson d. Mary Elliott/Jack Clemenger 6-0 6-0
Marjorie Mountain/Rupert Wertheim d. Lorna Bull/Keith Poulton 3-6 6-4 6-3
Hazel Webber/Arthur Yencken d. Minnie Goodman/Bruce Dive 6-0 9-7

Esna Boyd/John Hawks d. Syd Carr/Dudley Bullough 6-4 6-0
Mall Molesworth/Leslie Baker d. Mitchell/Prosper Sandral 6-1 7-5
Jessie Watson/E Jordan d. Grace Paterson/William Sheehan 6-4 6-3
Annie Gray/Percy Lane d. Floris St. George/Frank Peach 8-6 5-7 10-8

and finally note:

The official site has Lorna Utz. It is wrong. It is Gwen (Gwendolyn). There were two Utz's. Lorna Utz (nee Bull-listed in the 1922 singles draw) was maried to Leslie Utz in 1924. Gwen (nee Chiplin) was married to her mixed partner H. S. "Barney" Utz. This is the pair that traveled to England and played Wimbledon in 1925. Major thanks go out to GeeTee for pointing this out.

Rollo
Jan 30th, 2012, 05:52 PM
At any rate I think that's more than a single error/addition. Take a look and see what you think. I don't doubt that there will be errors/additions in what we are doing too-this is all ongoing work. The more we work together the more we will get closer to complete and accurate results.

I'm trying to go over results for one slam a week. You can look for other results here.

http://www.tennisforum.com/showthread.php?t=224499

Thanks Fantasio.

Fantasio
Jan 31st, 2012, 07:27 PM
IAlso, note my intentions here (especially post 1).
http://www.tennisforum.com/showthread.php?t=447550
With that in mind I'll start with the 1922 Australian Championships.
Extremely good job, I see. I stand corrected, "D. Menzies" included.

Rollo
Jan 31st, 2012, 08:44 PM
Extremely good job, I see. I stand corrected, "D. Menzies" included.

Thanks Fantasio. Your own work has corrected of a lot of my errors, so keep up the good job. :)

GeeTee
Feb 2nd, 2012, 09:33 PM
Jessie Watson/EDWARD Jordan d. Grace Paterson/William Sheehan 6-4 6-3
Another detail.

Fantasio
Jun 15th, 2012, 12:49 AM
In the new Elo rankings after Australian 2012, Kvitova went from 59th to 47th, just after Ivanovic. No other variations among the first 60 players.
No variations in the Elo rankings after Roland Garros.

thrust
Jun 23rd, 2012, 01:50 AM
All very interesting but highly subjective IMO esp when comparing opposition - Doris Hart and Anne Jones, for example, were both great players IMO and probably highly underrated by most. And to only factor in the slams is very limiting - years at #1, career singles winning % and tournament wins should also factor in there somewhere.

Happy to see it gives Seles her due but even I, a huge fan, could never rate her above Connolly, Court, Navratilova and Evert.

Still, an interesting perspective.

I AGREE! I like stats to, but lately here and on the men's forum, things have become a bit too much. One has to rate eras separately and judge a players greatness on their total career stats. Therefore, the greatest dominate players are: Lenglen, Wills, Marble and Connolly-for a few years, Court, Evert-Nav, Graf, Seles- for 2-3 years, then Graf again after the stabbing. Wills and Court probably had the longest span of domination. Lenglen was basically unbeatable for just about all of her career. Connolly dominated at the end of the peak years of: Hart,du Pont, Brough and Betz until her unfortunate accident. For me the eras are basically: 1915-1926, 1927-1939, 1940-1952, 1953-1960,1961-1973, 1974-1984,1985-1995, 1996-2000, 2001-2010.

thrust
Jun 23rd, 2012, 02:14 AM
Stats that should be included are: Number of sanctioned tournaments won, slams won, % of tournaments won that a player competed in, the same for slams. The rest is really not that important, except for H-H against top players of an era.

Fantasio
Jul 10th, 2012, 10:26 AM
No variations in the Elo rankings after Roland Garros.
And no variations, again, after Wimbledon. But of course, after seven different players won the last seven major events, it's obvious there is no dominance at all, by any player. Unless Kvitova or Azarenka rise again, there won't be any variation for a long time.

thrust
Jul 14th, 2012, 12:24 AM
But it's not Connolly's fault who her opponents were ... or that they weren't up to her level - and that is subjective anyway. I think some of her opponents are underrated.

Connoly's top opponents were Brough and Hart, who were just past their peak, especially Brough. I think Hart was the only player to beat her in an important tournament final, the Italian, during her best years. Perhaps she was a bit lucky in tennis, but very unlucky in life.

tennisvideos
Jul 14th, 2012, 08:41 AM
I AGREE! I like stats to, but lately here and on the men's forum, things have become a bit too much. One has to rate eras separately and judge a players greatness on their total career stats. Therefore, the greatest dominate players are: Lenglen, Wills, Marble and Connolly-for a few years, Court, Evert-Nav, Graf, Seles- for 2-3 years, then Graf again after the stabbing. Wills and Court probably had the longest span of domination. Lenglen was basically unbeatable for just about all of her career. Connolly dominated at the end of the peak years of: Hart,du Pont, Brough and Betz until her unfortunate accident. For me the eras are basically: 1915-1926, 1927-1939, 1940-1952, 1953-1960,1961-1973, 1974-1984,1985-1995, 1996-2000, 2001-2010.

Yes great post Thrust :)

Sam L
Sep 21st, 2012, 12:18 PM
Any update on ELO rankings after the US Open? I wonder if Serena will creep into the Top ten as she deserves to be.

Rolling-Thunder
Dec 31st, 2013, 01:52 AM
I would like to see this updated with the 2012 then the 2013 seasons completed. How far up does Serena rise?

thrust
Jan 11th, 2014, 01:39 AM
I will repeat myself.
In Elo method, dominance duration does NOT count. Only dominance level. Of course Court dominated 5 years, Navratilova also five. Also great was Evert, not mentioning Lenglen and Wills, that you forgot.
All these players could have dominated 20 years or more, there's no difference.

IF you are only looking at dominance level, it's easy to notice that between 1991 and 1993 Seles played 8 majors, won 7, only losing one match to Graf (who is - without any doubt - one of the most likey GOAT candidates). She defeated Graf (twice), Sanchez-Vicario (3 times), old Navratilova (twice), Novotna (once), Sabatini (3 times).
If this is NOT supreme dominance, I don't know how to define such a thing.
Well, Navratilova 1983-1984 dominated. But she "only" won 6 majors, losing to Horvath and Sukova. Losing to a low-ranked player such as Horvath is a disaster, comparing to losign to Graf (in Elo method).

But there's Court in 1969-1971. Well, she's better than Seles, having won 8 majors, losing only to Haydon-Jones - better than Horvath, still no Graf. And Court only defeated King twice, Goolagong twice. Much, bur not as much as Seles. Plus, don't forget she only played 49 matches during 9 majors (winning 48), while Seles played 56, winning 55, despite competing in 8 events. That's why Seles is higher than Court.

Serena played 6 majors, won 5. No match.

Evert played 8, won 6. Very good, and she defeated Goolagong 3 times, Navratilova also 3 times. She only lose to Goolagong and King. That's very good, and she's very close to Seles, but still not equal.

Connolly extraordinary. 9 over 9! Her problem, she did not defeat strong players. Hart, Fry, Brough are top 30 (alltime) at best. And she played 51 matches only.

So, why is Graf ranked higher (more or less, she's equal)? Because between 1988 and 1990 she won 8 majors, played 9, only lose to Sanchez-Vicario (better than Haydon-Jones), defeated Navratilova (3 times), old Evert (3 times), Sabatini (4 times), young Seles (twice), young Sanchez-Vicario (once). I think Seles defeated better opponents, but Graf won more, and more matches (62 over 63).

Now the difference between "dominance level" and "dominance duration" should be clear. Of course I know that Graf dominated more years (there are 1995 and 1996, of course), won much more than Seles, so she's a better GOAT candidate than unlucky Monica. But dominance level is the same.

Or, we can also say that "dominance level" = "strength at best" (if consistent). So, the old question, "was Seles really better than Graf or was Graf not at her best during 1991-1992?" could be settled, as the two players were of equal strength (according to Elo).

To rate Seles over Court is only one reason this ranking formula is utter Nonsense!

Sam L
Jan 11th, 2014, 04:48 AM
To rate Seles over Court is only one reason this ranking formula is utter Nonsense!

That's why this ELO ranking list is good. It allows us to look at things differently and question things that are otherwise missed because we are so clouded by things like the record 24 Grand Slam singles titles that we can't look beyond other things like the competition they had, how long their careers were, how good were their opponents when they played them and when they came to be dominant.

This system punishes you if you are lose matches to players who are low on the list when you play them and I think that's why Serena was so low but she should be improved now and I'm keen to know where she sits.

thrust
Jan 15th, 2014, 11:29 PM
That's why this ELO ranking list is good. It allows us to look at things differently and question things that are otherwise missed because we are so clouded by things like the record 24 Grand Slam singles titles that we can't look beyond other things like the competition they had, how long their careers were, how good were their opponents when they played them and when they came to be dominant.

This system punishes you if you are lose matches to players who are low on the list when you play them and I think that's why Serena was so low but she should be improved now and I'm keen to know where she sits.

So Steffi Graf had tougher competition than Court did for all of her career? I think no! But then, I am obviously missing something here.

Sam L
Jan 19th, 2014, 09:31 AM
So Steffi Graf had tougher competition than Court did for all of her career? I think no! But then, I am obviously missing something here.
You are. You need to look at how these numbers are calculated. It's all about how many points your opponents have at the time when you beat them.

Bueno is tough competition, BJK is tough competition but neither would've had the points Chris or Martina would've had in the late 80s when Graf was beating them.

Likewise when Seles beat Graf and Navratilova.