Raul_SJ
G.O.A.T.
What does this mean? A guy that was 0-15 in major finals would have the same score as a guy that was 1-0 in Slam finals?
Best Tennis Players of the Open Era Ranked
- Thread starter zvelf
- Start date
What does this mean? A guy that was 0-15 in major finals would have the same score as a guy that was 1-0 in Slam finals?
Jason Swerve
Hall of Fame
Something off with the women when Hingis has 25 titles and ranks below people with less than 25 titles.Here are the women:
zvelf
Hall of Fame
What does this mean? A guy that was 0-15 in major finals would have the same score as a guy that was 1-0 in Slam finals?
Yes. It's interesting how one person seems to think slam finals are underrated in this system and another seems to think they are overrated. But being 1-0 means you only have to win 7 matches in a row. Being 0-15 in finals means you have to win 6 matches in a row 15 times. Only 8 men and 6 women have achieved the latter in the entire open era.
AlexM
Rookie
F A C T
zvelf
Hall of Fame
Huh? Hingis isn't ranked below anyone who has less than 25 titles. And Hingis didn't win 25 titles but 43.
Raul_SJ
G.O.A.T.
That is why the guy who is 0-15 in Finals should have a higher score than the guy who is 1-0 in Finals.
It should not be equal!
zvelf
Hall of Fame
I would tend to agree, but like I said earlier, I weighted slam wins far more than any other metric because of the general opinion that slam wins matter far more than anything else. So when slam wins are weighted at 50% and almost every other metric is weighted evenly, the math says 1 slam win = 15 final losses. So to be clear, I did not choose the number 15. The math worked out that way given the premises.
Raul_SJ
G.O.A.T.
Do not blame the model for its output. You agree that reaching 15 Finals and losing should be rated higher than reaching one Final and winning.
When the output of the model do not correspond to reality, it is time to go back and reevaluate the model and its assumptions.
Last edited:
ridiculously exaggerated with the value of slams. and all nole records are greatly undermined and even make no points. masters are not worth anything nor to win all of them twice! you get points for being top10 but no points for YE # 1 (muzza sacrificed his career for this, sampras sacrificed AO the following year for his last YE # 1 and fed himself has said that he would always take one more YE # 1 over one more slam) !!! to hold all 4 slams at once? positive h2h with biggest rivals? and so you should be no1 for a year and a half to reach the value of 1 slam. ok everyone has the right to think what they want I guess.
Last edited:
zvelf
Hall of Fame
Sorry, I thought it was obvious that this model does not take doubles and mixed doubles into account. If it did, Martina Navratilova would be on top for 1968-present.
Jason Swerve
Hall of Fame
It should, is the point. Take the SAT, and tell your colleges they should only count the math section. See where that gets you admitted.
zvelf
Hall of Fame
First, there's no set reality here. We have the numbers and we have people's gut feelings. To show you all why I can’t impose much bias on this model, I am showing you what the model results are if slams are weighted at 40% as opposed to 50%. I removed the players whose careers span across pre-1968 to make it less confusing. Because there are 9 metrics and so many calculations at work, as soon as I change the weighting, players switch rankings all over the place. Maybe you only have one single favorite player and you psychologically project that you would manipulate the model to favor just that one player. Well, I love more than a dozen players and if I try to help one with weighting manipulation, I hurt several others and there’s simply no way around that.
So ask yourselves this, should Navratilova be higher than Serena? Should Hingis be higher than Venus and Goolagong? Should Davenport be higher than Sharapova? Should Kerber drop below Mauresmo, Austin, Halep, and Sabatini? Should Novotna and Martinez be ahead of Kuznetsova? If you answer yes, then a 40% slam weighting works better for you. If you answer no, 50% slam weighting works better for you. But more likely, you will have mixed feelings.
The men don’t change as much with the weighting change (though McEnroe comes to the verge of surpassing Borg), but should Murray be higher than Vilas? Should Nastase be higher than Courier? Should Roddick be higher than Kafelnikov? Should Muster be higher than Bruguera? Should Chang and Stich be higher than Rafter and Safin? If you answer yes, then a 40% slam weighting works better for you. If you answer no, 50% slam weighting works better for you. But again, more likely, you will be mixed on these players’ movements.
So you’ll understand that I’m mixed on the weighting as well. I actually personally feel the 40% slam weighting fits my idealization of where players should be ranked better than the 50% slam weighting, but I decided to go with 50% as a concession to the people who say ONLY slams matter and there are a lot of them.
itrium84
Hall of Fame
@Mileta Cekovic What do you think about this model/tables and the underlying principles? It all sounds pretty reasonable to me.
zvelf
Hall of Fame
Berrettini would be ranked around 116, essentially tied with Sebastien Grosjean, but Grosjean has played his entire career whereas Berrettini still has a ways to go and potentially collect a lot more achievements.
jussumman
Hall of Fame
The effort is made to base opinion on objective benchmarks. I've searched all types of GOAT rankings and top 10 are unanimous with usually 5 of the 10 in the same spot. So the differences are very minimal. The Big 3 are very close to each other ATM and only thing separating them are tweaks on weighing certain benchmarks over others and including some over others.
I have Novak as#1 GOAT on my objective-opinion based on Weighted Big Titles.
This guys ranking system, comprehensive, but is a bit too complex and somewhat confusing tbh
"1 major win = reaching 15 major finals = 3.67 Year-End Championships or 3.67 Olympic Golds = winning 53.6 tournaments = 76.3 weeks at #1 = 8.85 years in the top 10 = half the points of achieving a calendar year grand slam"
Mileta Cekovic
Rookie
Seriously guys, please stop the FUD about Elo inflation in Tennis. Some of you have probably heard or read somewhere that Elo is subject to inflation and now you seem to be spreading FUD without deeper understanding.
First to define what Elo Inflation means:
Elo Inflation means that average Elo rating of top N players increase over time, making it useless for cross-era comparisons.
It is true that under specific circumstances Elo rating computations CAN LEAD to inflation, but it DOES NOT apply to ATP-level tennis.
Elo can inflate if number of players and matches increase over time. Simply, if more new players are entering the sport competitions then the number of retired players and there is an increase in the total number of matches played over the seasons, then there are more Elo points in the flow, and there is a chance for Elo inflation.
However, this does not apply to ATP-level tennis in any way:
- Number of official matches per season on ATP-level is pretty much constant over time since Open Era begun in 1968: https://www.ultimatetennisstatistics.com/seasons
Number of matches per season lately is even less then in 70-ties, 80-ties or 90-ties!
- Average number of matches a single player played over a season is constant in the Open Era. Simply a top player can play 50-90 matches per season, sometimes more, sometimes less, depending on how deep he/she goes into the tournaments, but average number of matches a player plays per season is constant, one cannot play more because of physical limitations.
- Average Elo of top players is pretty much constant over time in the Open Era: https://www.ultimatetennisstatistics.com/dominanceTimeline?averageElo=true
As can be seen on the graph above, there are periods of greater dominance of for example Top 3 or Top 5 players, but Top10, Top 20, Top 50 and Top 100 average Elo is pretty much constant over time.
So there is NO Elo Inflation in ATP-level Tennis and that will be true for Any Elo formula used for tennis, not only UTS Elo formula (that has tweaked Elo K-factor for tennis), but really any Elo formula.
Last edited:
Mileta Cekovic
Rookie
This model seems reasonable, very good effort. Thank you @zvelf for the women part, it cannot be found on a lot of places.
Main point here is that the method is applied systematically and consistently to all players.
Looking at this effort, as well as many other efforts to model a 'systematic and consistent numerical value of tennis players' greatness' (however silly this sounds
), one interesting thing can be observed:They all lead to, if not the same rankings, then to the very, very similar rankings of tennis players!
For me, that can mean only one thing, they are all right!
At the end, regarding the model above itself, as some members noted, it weights Major wins pretty heavy (compared to other tennis achievements) and correlates very much to the number of major wins.
I.e. 1 major win = 15 major finals, not sure this is completely right. Or 1 major win = 53.6 ordinary titles. Why exactly 53.6, why not 50 or 60?
Not necessarily a downside, but a personal (maybe subjective) observation.
Last edited:
mike danny
Bionic Poster
Grosjean is much better than Berrettini though.
Wander
Hall of Fame
I think these models are pretty good and I, like @Mileta Cekovic particularly appreciate the OP for making a statistics based model for ranking the women because I can't recall seeing one before.
The 40% Slam weighing seems slightly better to me personally than the 50% model. Also, I think 1 major being equal to 71.2 titles on the women's model and 53.6 on the men's seems to me like undervaluing regular tournaments a little bit too much.
Can you explain the differences in the weighing of the men's and women's models? Any particular reason?
The 40% Slam weighing seems slightly better to me personally than the 50% model. Also, I think 1 major being equal to 71.2 titles on the women's model and 53.6 on the men's seems to me like undervaluing regular tournaments a little bit too much.
Can you explain the differences in the weighing of the men's and women's models? Any particular reason?
Jason Swerve
Hall of Fame
Enough respect due to the OP for this. This board is way too ATP focused when the WTA is more interesting in every way and has been since the early 90's.
zvelf
Hall of Fame
The reason women's titles are worth so much less than men's is because the highest 1968 and onward achievement by women is Navratilova's 167 tournament wins whereas for men, it's Connor's 109 tournament wins, therefore the men and women's tournament wins scale differently. In the major counts though, Serena's 23 is much closer to Federer and Nadal's 20. So the relationship is between 23:167 and 20:109.
zvelf
Hall of Fame
My personal preference of which tour was more fun to watch is WTA 1991-2007, 2008-2009 are about even between the two tours, but then ATP 2010-2016, then back to WTA 2017-present.
Last edited:
Jason Swerve
Hall of Fame
That early '10s was a real disappointment, wasn't it. That's what we were warning people would happen after Hingis retired. Some saw the writing on the wall, but it didn't make a difference in the end.
itrium84
Hall of Fame
No rush, thank you for your efforts
zvelf
Hall of Fame
Here are my updates to these stats after the U.S. Open. This time I decided to exclude players whose careers straddled between pre- and post-Open Era until I figure out a better way to mesh them.
After Wimbledon, Djokovic moved to #1 over Federer as the best man in the Open Era. The other big movers are the other major winners this year - Barty to #22 best woman of the Open Era and #40 for Medvedev on the men's side. Fellow Next Gens Zverev is at #51 (interestingly sandwiched between Davydenko and Nalbandian) and Tsitsipas is at #65. It may seem strange that Raducanu is just ahead of Krejcikova already with them being #58 and #59, respectively, but Krejcikova having 2 more titles does not surpass Raducanu having a raw 11% higher career win/loss percentage, which is huge, but also quite possibly an artifact of the small sample size of her short career. Raducanu's 75.8% win percentage places her between Swiatek's 77.5% and Venus Williams' 75.5%, so approaching ATG levels. Of course it remains to be seen whether Raducanu can maintain or improve upon that. You don't want to extrapolate too much from a small sample size.
Again, my methodology is detailed here:
Settling GOAT using a mostly objective method
With all of the debating that goes on over who is the Greatest of All Time (GOAT) in tennis, there has to be a more objective way of measuring the best. So I developed a relatively simple, fairly objective way to determine who is the GOAT, not that I expect everyone to agree with my methodology...
tt.tennis-warehouse.com
What I did not detail before is how I awarded Grand Slam points. I'll do that here now. I simply took the number of slams played on both sides, men and women, which in the past 53 years has been 422 times (2020 Wimbledon was not played and so is not included) and divided it by the number of times some form of grand slam variation occurred. So:
| Sample period is 53 years: 1968-2021 minus 2020 Wimbledon that was not played | 422 | Points | |
| 3 instances of a Grand Slam winners in the Open Era (Laver, Court, Graf) | 3 | 140.67 | Out of every 140.67 majors, one can expect a player to win 4 in a calendar year once |
| 8 instances of at least 4 majors in a row in the Open Era (Laver, Djokovic, Court, Navratilova, Graf x2, Williams x2) | 8 | 52.75 | Out of every 52.75 majors, one can expect a player to win 4 in a row once (therefore the probability to win 1 major = 1/(4th root of 52.75) = 37.1%) |
| 2 instances of 6 majors in a row in the Open Era (Court, Navratilova) | 2 | 211.00 | Out of every 212 majors, one can expect a player to win 6 in a row once |
| 11 players have completed a career grand slam 19 times in the Open Era (Laver, Agassi, Federer, Nadal, Djokovic x2, Court, Evert x2, Navratilova x2, Graf x4, Williams x3, Sharapova) | 19 | 22.21 | Out of every 22.21 majors, one can expect a player to win 1 of each one once |
If a player gets higher points for something that would have also awarded lower points (for example, a CYGS automatically means a career GS), then only the higher points are awarded. Players don't get points for both the CYGS and the career GS for winning the CYGS. So for Graf, she won the CYGS (140.67 points) + a NCYGS (52.75 points) + 2 additional career grand slams (22.21 x 2) = 237.8 points. Now, I could have separated the men and women's stats to derive different points for each side, but because these instances of variations of grand slams are so few, it would skew the numbers in drastic ways. The more years that pass, the higher these numbers will become but at the same time, the more people who achieve these, the smaller the numbers will become, so they adjust every year.
#itrium84
Last edited:
Where is Court and King on the latest list? AGAIN, the Open Era does NOT apply to Women's Tennis as ALL top players competed on the tour and played in Slams before 1968, as there was NO women's Pro Tour as there was in Men's tennis where the top players competed before 68.
Here are my updates to these stats after the U.S. Open. This time I decided to exclude players whose careers straddled between pre- and post-Open Era until I figure out a better way to mesh them.
After Wimbledon, Djokovic moved to #1 over Federer as the best man in the Open Era. The other big movers are the other major winners this year - Barty to #22 best woman of the Open Era and #40 for Medvedev on the men's side. Fellow Next Gens Zverev is at #51 (interestingly sandwiched between Davydenko and Nalbandian) and Tsitsipas is at #65. It may seem strange that Raducanu is just ahead of Krejcikova already with them being #58 and #59, respectively, but Krejcikova having 2 more titles does not surpass Raducanu having a raw 11% higher career win/loss percentage, which is huge, but also quite possibly an artifact of the small sample size of her short career. Raducanu's 75.8% win percentage places her between Swiatek's 77.5% and Venus Williams' 75.5%, so approaching ATG levels. Of course it remains to be seen whether Raducanu can maintain or improve upon that. You don't want to extrapolate too much from a small sample size.
Again, my methodology is detailed here:
Settling GOAT using a mostly objective method
With all of the debating that goes on over who is the Greatest of All Time (GOAT) in tennis, there has to be a more objective way of measuring the best. So I developed a relatively simple, fairly objective way to determine who is the GOAT, not that I expect everyone to agree with my methodology...
What I did not detail before is how I awarded Grand Slam points. I'll do that here now. I simply took the number of slams played on both sides, men and women, which in the past 53 years has been 422 times (2020 Wimbledon was not played and so is not included) and divided it by the number of times some form of grand slam variation occurred. So:
Sample period is 53 years: 1968-2021 minus 2020 Wimbledon that was not played 422 Points 3 instances of a Grand Slam winners in the Open Era (Laver, Court, Graf) 3 140.67 Out of every 140.67 majors, one can expect a player to win 4 in a calendar year once 8 instances of at least 4 majors in a row in the Open Era (Laver, Djokovic, Court, Navratilova, Graf x2, Williams x2) 8 52.75 Out of every 52.75 majors, one can expect a player to win 4 in a row once (therefore the probability to win 1 major = 1/(4th root of 52.75) = 37.1%) 2 instances of 6 majors in a row in the Open Era (Court, Navratilova) 2 211.00 Out of every 212 majors, one can expect a player to win 6 in a row once 11 players have completed a career grand slam 19 times in the Open Era (Laver, Agassi, Federer, Nadal, Djokovic x2, Court, Evert x2, Navratilova x2, Graf x4, Williams x3, Sharapova) 19 22.21 Out of every 22.21 majors, one can expect a player to win 1 of each one once
If a player gets higher points for something that would have also awarded lower points (for example, a CYGS automatically means a career GS), then only the higher points are awarded. Players don't get points for both the CYGS and the career GS for winning the CYGS. So for Graf, she won the CYGS (140.67 points) + a NCYGS (52.75 points) + 2 additional career grand slams (22.21 x 2) = 237.8 points. Now, I could have separated the men and women's stats to derive different points for each side, but because these instances of variations of grand slams are so few, it would skew the numbers in drastic ways. The more years that pass, the higher these numbers will become but at the same time, the more people who achieve these, the smaller the numbers will become, so they adjust every year.
#itrium84
zvelf
Hall of Fame
As I noted, "This time I decided to exclude players whose careers straddled between pre- and post-Open Era until I figure out a better way to mesh them." Even if the Open Era doesn't technically apply to the women, 1968 is a convenient cut-off because the stats going beyond that point in time get fuzzy. Games from that far back get unearthed here and there and the numbers change pretty frequently. Tennis Abstract just recently uploaded a bunch of women's matches from that era and I'm still sorting through that data.
zvelf
Hall of Fame
@zvelf Any updates maybe?
Here is another update after another year has passed. It's also a fitting time since Serena and Roger, both in the top 3 here, just retired, not to mention Barty earlier in the year. I've added players' winning percentages in slams as well as against top 10 players to the stats. As a reminder, this is only for players whose careers started in or after 1968 (I made one exception for Goolagong).
Djokovic has reinforced his numbers at the top due to adding a lot more weeks at #1. With her U.S. Open win, Swiatek breaks into the top 35. For this rendition, I halved the importance of weeks at #1, which slightly affected the rankings.
Now for the men, 1 major win = reaching 12.9 major finals = 3.0 Year-End Championships or 3.0 Olympic Golds = winning 44.0 tournaments = 150.7 weeks at #1 = 7.7 years in the top 10
For the women, 1 major win = reaching 12.5 major finals = 2.8 Year-End Championships or 2.8 Olympic Golds = winning 70.7 tournaments = 138.8 weeks at #1 = 7.7 years in the top 10
There will probably be some controversy that Federer is still higher than Nadal despite Rafa taking a 2 slam lead over Roger. However, a poll I took on the importance of the YEC in this thread (https://tt.tennis-warehouse.com/ind...-compared-to-a-year-end-atp-final-win.731659/) showed that on average, TTW values a YEC to be about 1/3 of a slam. The statistical median of the poll was 3, which just happened to be the actual worth of a YEC in this formula. So Federer's 6 YEC =Nadal's 2 slam win lead, tying them up there. What about Nadal's Olympic Gold? Well, Federer counters that with 11 more overall tournament wins and 101 more weeks at #1. Nadal could have cut into the latter at the U.S. Open, but Nadal's early loss and Alcaraz and Ruud making the final kept Rafa from reclaiming #1.
Even post 1968 Court won 92 tournaments, 11 slams and had the highest winning % of any other female player which dwarfs most of what other female players accomplished post 1968
Similar threads
