How good is NADAL?

Discussion in 'General Pro Player Discussion' started by Wuornos, Nov 21, 2007.


Would you buy the book "The Rating of Tennis Players of the Open Era".

  1. Yes

    23 vote(s)
  2. NO

    11 vote(s)
  1. Wuornos

    Wuornos Professional

    Aug 2, 2007
    The problem that occurs when truly great player arrives on the scene and dominates the top of world tennis, is that those behind often end up being underrated because of the reduced levels of their achievement.

    Because the DOT ratings take into account the dominance of other players when ratings are calculated this problem is countered. See for definition of DOT rating and methodology.

    A number of people have said to me in the past that it’s impossible to know how good someone is until the great player has declined or stopped playing. This is clearly untrue.

    It is not such a difficult task to calculate domination at a lower level and when taking into account the shape of the overall population a reasonably accurate rating can be derived.

    Consider for example the following DOT ratings for the end of 1985:

    1. Ivan Lendl 2713
    2. Mats Wilander 2706
    3. John McEnroe 2701
    4. Jimmy Connors 2664
    5. Boris Becker 2614
    6. Kevin Curren 2609
    7. Yannick Noah 2579
    8. Johan Kriek 2578
    9. Pat Cash 2575
    10. Anders Järryd 2572
    11. Joakim Nyström 2565
    12. Heinz Günthardt 2561
    13. Henri Leconte 2559
    14. Stefan Edberg 2558
    15. Andrés Gómez 2554
    16. Ben Testerman 2551

    Then compare these with the DOT ratings for the present:

    1. Roger Federer 2841
    2. Rafael Nadal 2708
    3. Novak Đoković 2647
    4. Andy Roddick 2638
    5. Nikolay Davydenko 2630
    6. Fernando González 2602
    7. Tommy Haas 2596
    8. Marcos Baghdatis 2595
    9. Lleyton Hewitt 2583
    10. David Ferrer 2576
    11. Richard Gasquet 2576
    12. David Nalbandian 2571
    13. Carlos Moyà 2570
    14. Tommy Robredo 2568
    15. Marat Safin 2561
    16. Juan Carlos Ferrero 2557

    If we calculate the difference in rating between the position of current players and their counterparts in 1985, we get the following list.

    1. 128 Points
    2. 2 points
    3. -54 Points
    4. - 26 Points
    5. 16 Points
    6. – 7 Points
    7. 17 Points
    8. 17 Points
    9. 11 Points
    10. 4 Points
    11. 11 Points
    12. 10 Points
    13. 11 Points
    14. 10 Points
    15. 7 Points
    16. 6 Points

    Looking at this the pattern is reasonably clear. The lower you move in the DOT rankings by and large the smaller the discrepancy between players from different periods.

    This is because the underlying assumption is that the standard of the tennis population remains constant and while it is probable there may be larger discrepancies in rating between the top 2 or 3 players, as you move down the rating the sample size of the players being compared increases and therefore it tends towards the same value.

    In calculating the ratings we can only ever calculate the difference in dominance of various players in relation to each other and this is where most statistical analyses stall. Their conclusion is we can see what the domination difference is between say Davydenko and Nadal but we can’t tell how good they are because of the presence of Federer.

    Their mistake is by trying to relate the performance of players to the number 1 player where the standard can vary widely. However if we are assuming the overall standard remains constant overtime then we are much better to compare these differences in dominance with a much lower rated player. In doing this we are ensuring the consistency of population median and from their differential between players can be awarded an overall value.

    Now some people have said to me ‘Ah but we don’t know that the standard of tennis isn’t improving’ and this is true. Players may be getting better overtime. However, most experts and pundits agree that the level of a players greatness should be measured against their success against peers. After all the greats from other sports are still considered greats even if the sport has moved on considerably and the past giants records have been eclipsed. If we are taking this stance it is reasonable to make the assumption that the standard of the population remains the same as this in effect provides an accurate level of peer domination across various periods of time.

    If all the above is true then the DOT ratings that calculate performance based upon Domination, Opposition and Tournament should provide a reasonably accurate indicator of player performance against peers at the players peek.

    This would indicate that Rafael Nadal has proven to date that he is the 12th best player of the open era so far having achieved a peak rating of 2711 after Wimbledon in 2007.

    Finally as some of you will be aware I have been putting together a book based on the DOT ratings which explains how the calculations are made, how individual years and tournaments compare with each other and who were the top 100 male and female single players of the Open Era.

    If you would be interesting buying this publication please vote YES above as this will assist me in my negotiations with a publisher.

    Many thanks and kind regards

    Last edited: Nov 21, 2007
  2. David L

    David L Hall of Fame

    Apr 1, 2006
    This is interesting. I think the book is a good idea. Go for it.
  3. FiveO

    FiveO Hall of Fame

    Feb 18, 2005

    Alot of work and I appreciate that. I believe that it needs to be even more in depth, to be a more accurate indicator.

    I believe the calculation needs tweaking.
    Going by memory and subjective views of two players mentioned in this and the second post you linked I looked at a Roddick this year to Becker 1985 comparison.

    Roddick outpoints Becker in your system by a statistically significant margin, however I think looking at results, opponents faced, and trends during the course of the year I think you'll understand the critique.

    Roddick had a more "consistent" major campaign in '07 i.e.:
    AO: SF
    RG: 1R
    W: QF
    USO: QF

    going 13-4 overall. Add his 2-2, SF appearance at the TMC and he was 15-6 at the 5 biggest events of the year, roughly a .71 winning percentage.

    Roddick also won two events London/Queens and Washington, both requiring 5 rounds of play due to 1R byes but where the highest ranked player he defeated in those 10 wins was the #26.

    Becker in '85:

    AO: 1R
    RG: 2R
    W: Win
    USO: R16

    He also finalled in The Masters bringing his record for the year at the five biggest events, to 14-4 approx. a .78 winning percentage there.

    Becker also won London/Queens Club that year, winning six rounds and defeating two top 16 players en route. At Cincy, always a top tier event Becker won another 6 rounds again defeating two top 16 players.

    For 2007 Roddick was 3-8 vs. top 16 players.

    For 1985 Becker was 15-9 vs. top 16 players.

    I think those are important results which seem to be ommitted from the DOT.

  4. Wuornos

    Wuornos Professional

    Aug 2, 2007
    Thanks FiveO

    Plenty for me to take away and think about there. I love constructive criticism like this. It always helps to improve the ratings.



    Tim :)
  5. Wuornos

    Wuornos Professional

    Aug 2, 2007
    High FiveO

    I've had a look at the points you have raised and thought I'd let you know my initial findings.

    Regarding the significance of the difference between 1985 Becker and 2007 Roddick this equates to 24 points which means we are saying that Roddick would score around 56% to Beckers 44% against simlar quality of opposition.

    I agree that Becker's performances were better against top16 players in 1985 but I suspect his losses against lower players were worse too. However, I am not going to get into a discussion on the relative values of wins and losses against different qulaities of opposition as there is a far more basic reason for this differential in rating.

    Many pundits believe that a single year is not sufficient volume for a conclusive evaluation of playing standard as the sample size of the number of matches played is to small. Some think three years is better others five years. By and large I certainly agree that a single years is to small a sample for other than the crudest conclusions to be drawn and I am seeking to achieve a greater level of accuracy than that provides.

    The DOT ratings are therefore calculated using a diminishing weighting based upon how long ago an event took place compared with the period the rating is being calculated for. By and large this breaks down to approximately 50% of rating being based on the current year with the other 50% being based on results in older tournaments.

    Boris Becker was very much at the beginning of his career in 1985 and therefore his rating was not really a full rating. Every effort has been made within the DOT ratings for a player with a limited series of results to still be able to prove that he/she is the number 1 player off a single year but the resulst do need to be exceptional to do this. Becker's results just did not warrant this. To be honest I think he did well to achieve a top 5 place off this limited evidence.

    This is really no different to what we would see in the ATP rankings for a player who had played a part of a year and therefore has a deflated rank. The difference being the DOT ratings run on a result horizon of greater than one year and with a declining weighting rather than the flat rating used by the ATP. Also the DOT ratings ensure cross era comparison is possible by standardising the rating results and are some what more complex in the calculation of points received in relation to specific event.

    I will continue to look into the points raised but I suspectr the fact that Boris was at the start of his career deflated his rating due to insifficient evidence being available for him to achieve a full rating. This is supported by the fact that Becker had climbed to a rating of 2659 by the end of 1986, a level in excess of Roddick's peak and went on to peak himself at 2726 after the French Open of 1991.

    I'll let you know if I come up with anythink else in relation to your query.

    Take care and keep posting

    Last edited: Nov 21, 2007
  6. Ossric

    Ossric Semi-Pro

    Aug 31, 2007
    Ottawa, Canada
    Is it possible to have a lifetime rating system to rate a player for their entire career and then rank them statistically without bias?

    Either way I like what you're doing and would definately support your efforts by purchasing the book.
  7. Wuornos

    Wuornos Professional

    Aug 2, 2007
    Thanks again Ossric.

    Yes, I have tried the lifetime rating system but found it tends to penalise players with shorter careers. By taking a peak over a shorter period, but one which is statistically significant, you give the opportunity to players with shorter careers to compete on almost even terms with players with greater longevity. Of course the players with greater longevity have the equivalent of several chances to establish a peak rating compared to the short career players but sadly this doesn't always help them. Also it enables a level of 'greatness' to be established for players who are still active and comparatively earlier in their career. E.g Nadal.

    Thanks for the interest Ossric and will certainly think further about your point.


  8. FedForGOAT

    FedForGOAT Professional

    Oct 20, 2007
    So this settles the debate by comparing players only to their contemporaries. I agree, that's the right thing to do. besides, any other way, some people would still argue that some players had stronger opposition.

    Who are the top 10 on your list of all time?
  9. Wuornos

    Wuornos Professional

    Aug 2, 2007
    Well if you look at the DOT Peaks which is what I prefer the top 10 of the open era would be:

    1. Roger Federer
    2. Rod Laver
    3. Ivan Lendl
    4. Björn Borg
    5. Mats Wilander
    6. Pete Sampras
    7. John McEnroe
    8. Ken Rosewall
    9. Jim Courier
    10. Boris Becker

    However I have been working on Ossric's suggestion of rating the whole career from a statistically independent point of view and this currently gives the following output so far for the open era:

    1. Pete Sampras
    2. Roger Federer
    3. Björn Borg
    4. Ivan Lendl
    5. Jimmy Connors
    6. Andre Agassi
    7. Mats Wilander
    8. John McEnroe
    9. Stefan Edberg
    10. Boris Becker

    However much work would still need to be done on this as at the moment its treating all tournaments equally and assumimg all players have faced the same quality of opposition. Things which the DOT rating has already compensated for.

    This new rating system probably coincides better with popular opinion. However it does penalise those with shorter careers in the open era like Laver who ranks at 11th while boosting those with long careers like Agassi and Connors.

    I'm not sure I like it. The logic behind it seems flawed despite what looks like a good output.

    On this basis Nadal would currently rate at 16th best of the open era.

    Which list do you prefer?

    Last edited: Nov 22, 2007
  10. Klatu Verata Necktie

    Klatu Verata Necktie Hall of Fame

    Jul 4, 2007
    Miami, Fl
    I'll buy your book so long as it's a softcover ;)
  11. NadalandFedererfan

    NadalandFedererfan Banned

    Jul 16, 2007
    He is a truly amazing player. 2nd greatest clay courter in history at only 21, 2nd best grass courter in the game too, and one of the best hard courter players to boot. All this at only 21.
  12. akv89

    akv89 Hall of Fame

    Jun 23, 2007
    I find it very interesting that Borg is ranked 4th best at his peak but is 3rd best over for his entire career, considering that he retired at the age of 26.
    I understand your point about why it seems more appropriate to rank the players at their peak. But that implies that a player with one freakish year (eg Wilander, ranked 5th) can be ranked above someone who has had many more years of dominance of a lesser magnitude (eg Sampras, ranked 6th). Perhaps it is possible to make a compromise between looking at a player's peak and their consistency by looking at a larger portion of their career. For example, is it possible to devise a rating system that looks at the best 5 years of a tennis player's career; to me, this kind of system is more indicative of a tennis player's skill since it not only accounts for their peak performance or their consistency over their lifetime, but how well they were able to maintain their performance when at their peak.

Share This Page