How Serena Williams Loses to Navratilova on Every Surface

Serena and Venus still a Rung Below Navratilova and Evert.

A Jon Wertheim article in Sports Illustrated followed up by other blog postings argue that Serena Williams was the greatest tennis player of all time.  Numerous tennis writers from tennis magazines , other periodicals and websites posed their own opinions throwing out numbers like TARP money for union jobs.  Though we think the story of women’s tennis since the advent of the Williams’ sisters is about Richard Williams Zen-like coaching methods, we weigh in with our opinion on Serena.  We approach the greatest woman’s player by using the same methodology as our evaluation of men’s all time great players, the SHOTS framework.

SHOTS is a simple and easy framework where we establish that there is no reason to hypothesize about who would win a match when there is actual data that shows who won those matches.  We look at winning percentage among all time greats, rank each grand slam tournament victory for toughness, provide a score and add them up.  We also normalize data throwing out bad data, only looking at players during their peak years, providing extra points for longevity.  Though Wertheim concludes that Williams would beat anyone on hard courts (her best surface), we look at every slam tournament on all surfaces and come up with a cumulative all-surface rating as well as a more granular rating per surface.

Our qualitative argument for Martina Navratilova vs. Serena is that there is nothing Serena could show Navratilova on a serve, volley or ground stroke basis that Navratilova had not seen before.  Wertheim argues that Serena’s first serve, which averaged 105 mph at the 2010 Wimbledon final, is something unlike anything seen in women’s tennis.  Yet Navratilova, the greatest mixed doubles player of all time, defeated big serve and volleyers Todd Woodbridge,  John Fitzgerald and Paul Annacone at fast surface grand slam mixed doubles championship.  (Special Note:  Paul Annacone was the serve and volley coach to Pete Sampras during his kamikaze run to 3 consecutive US Open finals in Sampras’ late 20s and early 30’s.  Annacone has now been hired by Roger Federer to provide a similar boost. )  And Navratilova looks like a pixie against those fellas too.  Likewise, Navratilova had an 85% winning record against master groundstroker, Evert, post 1980 on all fast surfaces.

Wertheim argues further that Serena has been thwarted in title matches by Venus, yet Venus and Serena have only played each other 23 times vs. 80 matches played between Navratilova and Evert.

Again, we don’t argue who would win one match between all time greats as it isn’t determinant but instead who would win more than 5 out of 10 matches between the players or who would win a tournament of all time greats.  With Navratilova we know we would get the fittest and fiercest competitor ever to play the game, hardened by losses to Tracy Austin at the US Open and her father’s suicide as a child.  She was able to beat then number 1 Graf, at the age of 35 to reach the US Open final before losing to Seles the next day.  In her late 40’s, within the last 7 years, she won two mixed doubles grand slams.  With Serena, we wouldn’t know who would show up, the great serving and fit player from the 2010 Wimbledon championships or the substantially overweight player from the Australian Open a year or so before.

Our own quality ratings of Navratilova’s cumulative wins against the field show at least a 2X advantage on every surface and at Wimbledon a 3X advantage vs. Serena.

SHOTS ANALYSIS

Martina Navratilova and her long time rival, Chris Evert, come out head and shoulders above the rest of women’s tennis via SHOTS analysis.  Serena Williams finishes in the top 6 using our SATERICCON metric.  Using SITDON analysis, Serena Williams finishes behind Martina Navratilova, and Steffi Graf, but ahead of Venus and Chris Evert though only playing 60% as many matches as Evert.

We give a special mention to pre-stabbing Monica Seles, who we consider one of the greatest champions of all time but was cut down in her prime by a crazed German fan.  Seles had won  8 out of 10 slams and her 9 out of 10 finals streak is unparalleled in women’s tennis where she had clearly ended the Steffi Graf era.   Seles missed 3 years of her prime tennis playing career where a continuance of her 80% winning approach could have resulted in 10 more slams.  Steffi Graf went on to win 11 more Grand Slams 7 of them while Seles was rehabilitating.   Graf never lost to Seles again upon her return despite being 2-3 against her in the years prior to Seles stabbing.

Women’s tennis lends itself well to SHOTS analysis as 84% of women’s open era grand slams have been won by 16 women, our Pantheonists.  We measure these top women’s play versus each other and hypothesize who would win a theoretical tourney based on real world match outcomes.  Slightly less than 3,000 matches have been played between grand slam titlists in the Open era.  More than 1,700 of these matches have been played between Pantheon players  and we looked at all of them.  The top 16 players are:

Steffi Graf	Justine Henin-Hardenne
Martina Navratilova	Evonne Goolagong
Chris Evert	Martina Hingis
Serena Williams	Arantxa Sanchez-Vicario
Margaret Court	Hana Mandlikova
Monica Seles	Maria Sharapova
Billy Jean King	Lindsay Davenport
Venus Williams	Jennifer Capriati

We normalize data only considering open era champions.  Though this does not give enough due to Margaret Court, we think the impact of prize money served to make the sport more competitive.  Though there is no “dream team” effect in women’s tennis as happened in men’s tennis .  (When admitted into  the grand slams in 1968 Rod Laver won a Grand Slam and an old Ken Rosewall challenged for Slams late into his 30s.)  We consider the lack of money prevented great female players from competing in the Australian Open and for a time in the French Open.   We also throw out Australian Open data for women’s play since it was regularly skipped on the circuit by top players such as Navratilova and Evert as well as Graf.

Our approach is supported by the fact that our metric, the Slam Triple, which has been accomplished 31 times in women’s open era was accomplished for the first time in Australia in 2001.  Instead, we substitute a Slam Yield Metric (SYM) for  the top 4 women which normalizes their data to consider how many slams they would have won if they had played every slam instead of taking breaks from slams as frequently has happened for everyone of the Pantheonists.    SYM is analagous to the NBA statistic for rebounds per 48 minutes played which looks at the rebounds of a player for the amount of time they play and then normalizes the data to a 48 minute framework.

SITDON Analysis and Serena Williams

Serena Williams finishes third using SITDON analysis.  Careful review of the Pantheon matches shows that it is hard for top players to beat each other consistently.  Serena’s 60% normalized winning percent is barely better than Hingis’ and Evert’s 57% and a bit better than Venus’ 55%.  Though Serena has a winning record over the overwhelming majority of Pantheon players she has played, she does not have an overwhelming advantage over her rivals as does every player before her.  Players like Navratilova who one year lost only 1 match and in a 3 year time frame lost only 6 matches clearly outperformed her vs. the field.  Evert who had clay winning streaks of 125 and 75 matches finishes far ahead of her in winning percent.

We have to leave it to conjecture as to what sort of winning percents Evert and Navratilova would have had if they did not have each other as rivals but it is possible they would have surpassed Graf even with Monica Seles sidelined.

Serena’s inability to get substantial winning percentages over Justine Henin who she lost to on Clay and Hard courts and her sister Venus on grass reduce her SITDON score.  Likewise an early rivalry with Martina Hingis was inconclusive on head to head matchups.

Interestingly, much of Steffi Graf’s winning percent for the SITDON analysis is generated by overwhelming winning records against Hana Mandlikova, Aranxta Sanchez-Vicario, Martina Hingis and Jennifer Capriati while she was only able to play the 13 year older Navratilova and Evert to a standstill breaking even against them in more than 30 matches long after their prime.   Both Evert and Navratilova would have had significantly better records (near 80%) in their era if one or the other had not played as they nearly split their matches with Navratilova winning 43 of their 80 matches played against each other, most in finals of tournaments.   Indeed, Evert and Navratilova, playing late into their 30’s played twice as many slam winners as Serena in their careers and had better winning percents even including play in their 30’s.

Yield Analysis and Serena Williams

Though the top 6 women’s players have won many Grand Slams they have frequently skipped slams.  Evert and Navratilova skipped the French Open 3 times each and the Australian Open several times in the period between their first slam title and their last slam title.  Steffi Graf, almost recognizing the lack of competition due to Seles absence skipped a whopping 11 slams, Serena has skipped 9 in the 11 years spanning her first and last slam victories.  We normalize the data by assuming that players would continue winning at the same rate they won their other slams.  From this we get a theoretical yield.

In the case of Graf, we calculate based on what would have happened if Seles had not been stabbed voiding her 11 slams but recalculating for a full 48 slams played.  Though Seles is clearly the second greatest clay court champion of all time based on just 3 years of data and had a 2-0 hard court record vs. Graf in Slams, we calculate Graf’s yield on a conservative 50% basis post Seles injury.  We post the results we think would happen below.

One number that stands out in the Yield Analysis is the stunning number of finals the all time great women reached while playing.  Each all time great player reached the finals of at least 69% of the slams they played with the exception of Serena who has only reached 46% of the Slams she has played.

Projected Grand Slam Yield Between First to Last Slam Titles
	Slam Chance Conversions	Slam Finals Reached	Slams Missed	Slams Played	Normalized Slam Yield Assuming Full Schedule
Monica Seles*	80.0%	90.00%	0
Steffi Graf*	59.46%	83.78%	11	37	24
Martina Navratilova	46.15%	69.23%	9	39	22
Chris Evert	45.00%	72.50%	8	40	21
Serena Williams	37.14%	45.71%	9	35	16
*  Only Consider Seles 10 Grand Slams after 1st win before stabbing
**  Normalize Graf’s slams for those missed and if Seles had not been stabbed

Serena Williams SATERICCON Analysis

Using SATERICCON analysis, we were able to rank each and every grand slam event won by a Pantheon player since 1968.   When we add up the scores we get the following rankings.

Much of Serena’s place in the rankings is based on the overall quality of the competition she played in the majors she won which is around the middle of the pack vs the other Pantheonists.  Both Venus and Billie Jean King won more difficult Wimbledons and US Opens.  Monica Seles won 3 of the top 5 most difficult French Opens.  Of the top 10 players, Serena has played the fewest Pantheonists, Navratilova playing almost 80 more such matches on a normalized basis.  Venus and Martina Hingis played more difficult slates as did Navratilova and Evert who faced each other 80 times.

Serena benefits from the passage of one of the greatest eras in women’s tennis which seemed to end around 2005-2007 with the retirements and semi-retirements of Hingis, Capriati, Davenport, Henin and Clijsters.  Serena has won 5 slams since 2007 and has benefitted from the sidelining of her competition.   By our measures, Venus and Serena have close to the same winning percentage and cumulative grand slam quality rankings.   Again our argument isn’t that Navratilova has a 2.6 times greater chance of winning a grand slam than Serena (though we think it is indicative of some sort of advantage) but that in a tournament 2.6 times more difficult than Serena’s average, she would be far more likely to win than Serena.

Measuring Eras with Slam Triples

In support of our thesis that Serena has benefitted from the retirement of one of the great generations of tennis, we look at the number of slam triples.  There were a few slam triples in the 1970’s and the 1980’s saw very few opportunities as Evert won 7 French Opens and Navratilova won 9 Wimbledons.  Evert and Navratilova won nearly 50% of the slams they played in the 12 years separating their first and last slam championships.  Even when they did not win they reached the finals roughly 70% of the time eliminating almost all chances for others to win slams in their era.

Slam triples increased dramatically as Evert and Navratilova’s retirement, Seles attack and later injuries to Graf opened the gateways for 8 different French Open Champions, 8 different Australian Open Champions and 8 different US Open Champions over an 11 year period.  No Slam Triple has been won in the last 3 years.  Though the Williams sisters win more than most, parity has been the rule of the post Graf era.  From 1975 to 1996 when Graf began to suffer injuries, only 7 women held the number 1 computer ranking.  Between 1997 and  2009, 13 women have held the ranking.  Serena has been number 1 less than half the time of Navratilova, Evert or Graf.

The most disturbing trend as exemplified by the slam triple chart is the shortened lifespan of the average top women’s tennis player.  Hingis, Henin and Clijsters seemed to have retired prematurely and Capriati, a child prodigy, was out of tennis more than in it during her prime.  Maria Sharapova and other attractive tennis players may be distracted by their sponsorships and seem to be playing in between commercial spots.

Richard Williams genius and why he has two daughters in the top 6 of all time players comes from his Phil Jackson-like coaching approach.  With his constant remarks that his daughters didn’t need the sport, by not burning them out on the children’s tennis tour or in tennis camps, and by deflecting criticism from them to him as the best coaches do, he has created a throw back group of players with longevity almost equal to Graf, Evert, Navratilova and the other all time greats.



Ranking Federer by Surface All Time

ADDING THEM UP – MEASURING TENNIS GREATNESS

We’ve covered how Roger Federer stacks up vs the other all time greats on an all surface basis.  Our methodology, SHOTS , argues that for tennis greatness it is important to establish a consistent framework.   SHOTS relies on 2 metrics we created, SITDON, which looks at career winning percent between all time greats, Pantheonists, vs. each other and SATERICCON, a multi-dimensional snapshot of the competitiveness of open era slams ranking each one of them.  When we aggregate the results of those slams for each winner, it gives us a portrait of the all time most competitive slam champions, those players who were greatest when greatness was required.   So rather than hypothesize, we look at player records, value the toughest tournaments and add them up.  In the absence of an alternative methodology, we provide a robust framework to answer the question, who really is the Greatest Tennis Champion.

Breaking out Federer’s match record by surface, he places 7^th at Wimbledon in SATERICCON score vs. Sampras who comes in first with a 2.63 ranking.   Sampras may not have a 2.63 times greater chance of winning a Wimbledon championship than Federer (though we think it is somewhat indicative).  But he is more likely to win such a championship where players have won 2.63 times more slams than in Federer’s era.  (Remember that we have adjusted the SATERICCON rating to almost double Federer’s chances of winning a slam vs all time greats based on cumulative rather than average score. )  Some consider Sampras’ victory in 1993 to be the greatest Wimbledon.  Quarterfinalists included 6 slam winners (Agassi, Edberg, Becker, Stich, Courier and Sampras) and 2 multi-slam finalists (Todd Martin and Cedric Pioline).  However, Becker’s 1989 victory had 4 quarterfinalists that had won as many grand slams as the 1993 quarterfinalists combined (Edberg, McEnroe, Lendl, Wilander). Both are great feats and show the confluence of great all time players and styles in that era.

Again, SATERICCON Analysis shows the quality of a player by the field they defeat.  Not surprisingly the 4 toughest Wimbledon’s occurred within a 6 year time frame, between 1988 and 1993.

A similar story plays out at the US Open where Federer places 6^th to John McEnroe’s 2.93  cumulative rating.  Though some have argued that Sampras’ first US Open victory was the hardest with 5 Pantheon quarterfinalists (McEnroe, Lendl, Agassi and Becker), on a SATERICCON basis, Edberg’s 1991 victory had at least the same difficulty and his 1992 victory was superior with 5 Pantheon quarterfinalists (Sampras, Agassi, Courier and Lendl) plus Michael Chang, a slam winner. McEnroe had a similarly challenging 1980 victory with 4 Pantheonists (Borg, Connors, Lendl) and multi-slam winner, Johan Kriek.

Again, SATERICCON Analysis shows the quality of a player by the field they defeat.  Not surprisingly the toughest US Open’s are concentrated in two eras when great new talents emerged to challenge the established talents still in their prime.   2 of the top 5 occurred when McEnroe and Lendl came to challenge Borg and Connors and the remaining 3 occurred when Agassi, Sampras, Courier and Chang came to challenge Becker, Edberg and Lendl.

When taken on a weighted average across fast surfaces Federer ranks a cumulative 7^th on the SATERICCON scale.   Though Borg never won the US Open his cumulative Wimbledon score puts him ahead of Federer.  Likewise, Becker and Edberg who won less than half the titles of Federer, rank ahead of him when you consider that they played at  the nexus of all time great play.

When reviewing the fast surface data, we do a reality check of Federer’s play vs. comparable players who emulated the champions listed above him.  With a 10 year age difference there is insufficient match experience between Federer and Sampras to make a judgement, though Sampras frequently won tournaments at least twice as difficult as Federer and served 5-10 mph on average faster.  Federer lost to Nadal on fast courts many times.  Would he do any better against a left handed serve and volleying McEnroe.  Using Federer’s record against Agassi prior to Agassi turning 33, would Federer do well against a left handed, tough returning player like Connors?  Finally, Patrick Rafter dominated Federer before he retired, would Federer do better against Stefan Edberg who stylistically is a similar yet vastly superior player to Rafter?

When we started this discussion of all time greatness, it was pre-Wimbledon before Rafael Nadal won his 2^nd title in 4 consecutive finals.  Nadal fits the prototypical model of the Pantheon players as a teenage winner of a slam or a person who wins a slam within a year or two of turning pro such as Connors in 1974.  All of the top 5 Pantheon  players established themselves this way but not Federer who finishes 10th all time / all surface via SATERRICON .  Nadal is likely to move to #2 all time with his next slam win based on his SATERRICON rating while still maintaining the highest winning percent among Pantheon players via SITDON analysis.

We now look at Nadal’s French Open record.  Nadal’s French Open record is marked by his victories over Federer in 3 finals and 1 semifinal and Federer’s participation in each quarterfinal during Nadal’s era.  Borg’s era is marked by the absence of great clay court players and the retirement of the great Aussie generation early on and Lendl’s is hindered by his quest for a Wimbledon title where he skipped multiple French Opens.   In Lendl’s absence, Wilander and Andres Gomez (who Lendl handily beat 4 times at the French Open) won the title.

Again, SATERICCON analysis shows the quality of the players by the fields they played against.  Not surprisingly the toughest French Open’s involve the Lendl – Wilander rivalry which saw many great multi-surface matches and by Jim Courier’s two year dominance of the surface over Agassi.

To improve his legacy, Federer  will need to do more as the path of his career continues to mirror Sampras with fewer victories and even less victories against other grand slam winners.  A US Open, Wimbledon or Australian Open win against Nadal (the French seems out of the question) and tournament wins over the next generation of greats will be significant and could improve his SATERICCON ratings so he passes Becker and Edberg.  Sampras, Borg, Nadal and McEnroe seem unobtainable and surpassing Connors will depend on his success against a younger generation of future slam winners perhaps including Juan Martin Del Potro (should he return successfully from wrist surgery) and maybe Sam Querrey as both Nadal and Djokovic are on the down side of their career grand slam trajectories and are unlikely to add significantly to their totals.  We don’t see a new generation of players on the horizon like Sampras and Agassi or Lendl and McEnroe that will catalyze the game via a great rivalry with Federer and Nadal.

If Federer Isn’t the Best of His Era, How Can He Be the Best of All Time?

Simulation Metrics for All Time Greats

Tennis Grand Slam winners have played more than 3,000 matches between each other.  Pantheonists, the 16 greatest tennis players of all time, have played more than 1,400 matches against each other while winning 70% of the available grand slam titles of the Open era.  We crunched the numbers and created two metrics to measure the overall greatness during a career for Pantheonists.  In reviewing the data, we normalized it by considering only matches where players were younger than 31 (there are only a few slam winners over this age) or if there was less than 5 years of age between the players, such as with Sampras and Agassi, we included that data into their later years.

Results are below which show that on a winning percentage vs. other all time greats, Roger Federer ranks #14 on the SITDON scale.  Rafael Nadal’s #1 ranking is built entirely on his dominance over Federer and we consider there to be insufficient data to rank him #1.  We believe that we will need to see Nadal’s performance over the next few years vs. other rising players to clearly see where he ranks among the all time greats, but he has a fantastic start and it is clear that we are now in the midst of the “Nadal” era overshadowing Federer’s era with his defeat of Federer at Grand Slam finals on every surface the last time they played.

On a cumulative quality of slams all surface ranking, Federer ranks #10 all time.  Federer’s all surface slam does little to influence his overall ranking as there was only one other single slam winner in the quarterfinals of his French Open victory.  Assuming Federer does not change stylistically i.e. adapt a kamikaze net rush style at all costs approach like McEnroe, Rafter, Edberg and Sampras in their later years, we don’t anticipate he will win another slam without injuries or upsets to his central competitors.  Nadal is on course to surpass both Borg and McEnroe from a cumulative quality point of view on his next slam victory and may surpass Sampras with 2 or 3 more slams.  We consider this a difficult task since Nadal is the same age as Wilander at the time he won his last slam and one year younger than Borg when he retired.

Our two metrics are SITDON, the Secada Index of Tennis Dominance with Overt Normalization, and SATERICCON, the Secada Absolute Tennis Era Relative Influence and Championship Competitiveness Over Normalization.  SITDON looks at career winning percentage before the player turns 31 to determine how they did vs. other all time greats.  We consider this to be an excellent substitute for weeks at number 1 ranking and number of slams won.  From our point of view, SITDON is the equivalent of baseball ERA which tells you week in, week out, what was the consistency of that player.  However, SITDON is far more granular as it only looks at the statistics in matches between all time greats, like pitching against Reggie Jackson or Barry Bonds and does not include the equivalent of baseball’s bum of the month.

SATERICCON, measures individual greatness at any point in time.  Though SITDON measures overall career performance, SATERICCON answers the question, in a tournament of the greatest players, who would win those tournaments?   Historically, who was the greatest at the instant when it mattered.   It is a complementary statistic to our Slam Triple metric yet it considers the cumulative value of winning a number of slams which may be less competitive vs. winning a few ultra-competitive slams.  So it answers the question, if you won a slam in the ultra competitive 1987-1993 period, how would that translate into playing in slams in the far less competitive, A32 era and vice versa.

Normalization is the process of looking at data, in this case, 1,400 plus match results and selecting the good data while throwing out the bad.  Though not a perfect process we erred on the side of conservatism in determining when a player was at or near their peak.   In this case we included all data for players from the time they began playing pro tournaments to their 31^st birthday.  Long-playing champions such as Connors, Lendl, Sampras and Agassi are rewarded by both metrics for their longevity.  They are more than just champions for tennis, they are part of the fabric of the sport, tennis DNA.

SITDON has 4 advantages over other measures.   (1) it eliminates factors that others say make era comparison indeterminate such as equipment,  fitness or seedings.  All that matters is the results between top players, (2) it makes it easier to evaluate how age, mileage and style impact the outcome of a match and (3) it refines overall win record and overall match record to only those matches between the greatest players of eras at their peak.   (4) It eliminates computer ranking which is frequently subverted for business to incentify players to play more with higher risk of injury.

SITDON is an absolute measure of competitiveness between Pantheonists in the same era.  Federer’s total match record vs. Pantheonists ranks him 15^th all time in number of matches and with normalization, he ranks 14^th out of 16 in winning percentage vs. Pantheonists  as well.  John Newcombe  takes the last spot as all of his wins over Laver and Rosewall are eliminated via normalization i.e. they were all over 30 when he played them.  Even without normalization i.e. elimination of matches vs. Pantheonists far from their peak, Federer finishes near last in the Pantheon.

Note:  though cumulative career statistics matter on an absolute basis, there is a danger in quoting mid-career statistics average or percentile statistics for tennis players as they are surely to decline in the second half of a career.  Nadal and Federer’s percentages and averages are surely to decline as has every Pantheonist before them as they play longer and deeper into the latter half of their careers.

SITDON measures what would happen if Pantheonists were to play one singles match against each other, SATERICCON measures what would happen if Pantheonists were to play a succession of matches against each other.   SATERICCON’s basis is to determine who was greatest when the greatest all played each other assuming a winner of an all time great tourney would be indicated by past performance.

To create this measure, we use analytic methodology and then we consider ancient and present competitive folklore.  Larry Holmes was undefeated in his first 44 bouts beating an old Muhammad Ali in 15 rounds.  Ali defeated 6 heavyweight champs in or near their prime.  Experts consider Ali a greater champion.  Michael Jordan’s Bulls became champs after beating Isiah Thomas’ Detroit Pistons (at their prime) who had beaten the Celtics and Lakers before them.    Olajuwon’s Rockets won the championship when Jordan semi-retired and the Bulls and Pistons were long past their glory.  They disappeared when Jordan returned.  Jordan’s Bulls are considered greater than the Rockets.

In the Trojan war, Achilles retired briefly over compensation issues, in the interim period Hector laid waste the Greeks slaying far more than Achilles that year and nearly destroying their navy, almost altering history.  But when Achilles came out of retirement, everyone well knew who would win, it was destiny.  Our methodology borrows heavily from this philosophy and the Highlander series.  When the Highlander defeats another Highlander he gains the power of that Highlander and all their previous victims.   Likewise in the Volsunga Saga of Nordic and Germanic literature, if you defeat a dragon and eat it’s heart you gain it’s power.  When measuring greatness, history has always looked at the quality of your victories over the quantity. In SATERICCON, when you defeat another slam champ or the person who defeated them in the slam, you gain their power rating as a cumulative score.

We measure the overall difficulty of winning a grand slam championship by the quality of the field at the quarterfinal stage of the tournament.  We consider the number of grand slams won by the other quarterfinalists, excluding the winner and score the slam as having the value of difficulty assigned by all quarterfinalists.   So for example, Pete Sampras’ first Wimbledon championship had Jim Courier, Boris Becker, Mats Wilander, Stefan Edberg, Goran Ivanisevic and Andre Agassi in the quarterfinals.  The cumulative score of that win is a 28 which is the number of career slams by those players.  In Federer’s first US Open victory, Andy Roddick, Lleyton Hewitt and Andre Agassi were quarterfinalists .  Those players won 11 grand slams between them and Federer’s score is an 11.  However, to come up with a true “normalized” measure of greatness, we only measure players who were at or near their peak in skills and athleticism, so Federer’s score was reduced to 3 by eliminating Agassi who was already 33 at the time of this tournament.  Every player was impacted by this measure as almost each player had an all time great long past their prime in the quarterfinals of one or more of their championships.

The impact of normalization on Federer’s record is far greater since he has played  so few Pantheonists and is dominated by Nadal, the only other Pantheon player in his era.  Without normalization, based on his wins over a 33 year old Agassi, Federer would move past Becker and Wilander on the all time list but that would only get him to #8 all time on an all surface basis.  It was important to use the cumulative score of grandslam wins vs the average score as it balances out the dearth of grand slam champs Federer defeated per tournament but gives him extra points for his cumulative slam wins.  (On an average Slam victory Q-Rating score basis, Federer would rank near last.)  This study does not address what would have happened if Lendl had given up his Quixotic quest for a Wimbledon title and won 2 more French Opens, what would have happened if McEnroe had not taken a break during his career or what would have happened should Jimmy Connors have been allowed to play the French Open in 1974.  All these players ranked ahead of Federer in SITDON and SATERICCON rating.

To create the metric for competitiveness of grand slam victory, we used Federer as the baseline for all other players since so many journalists and talking heads rank Federer as #1 we gave Federer’s cumulative score a 1 and then graded the other players on a scale relative to Federer.  Though several other players have won a Slam Triple, Federer has never been able to do it which questions his ability to win a tournament of all time greats.  Though our SATERICCON rating does not say Pete Sampras is 2.37 more likely to win an average slam than Federer (though we think it is positively indicative), it does say he is far more likely to win a slam 2.37 times more difficult than Federer.

We also eliminated the Australian Open from consideration because it had little relevance in tennis until it became the first slam on the calendar in 1987.  Players like Orantes never played the Australian Open and Borg and Nastase played in it once.  Next, The toughest tournaments of all time on each surface and the winners.

SHOTS, the Hierarchy of Tennis Supremacy

There are around 1 million articles or comments on the internet about Sampras vs. Federer.  Few articles ask, in a tournament of all time greats on any surface, exactly who would Federer (the first seed by A32 rules) beat and how?  If you had to bet your last dime on who would win a tournament of all time greats, would it go towards Federer or someone else?  To understand this concept we created SHOTS, Secada’s Hierarchy Of Tennis Supremacy.  We discuss SHOTS at the midway point of this article while delving into more of Federer’s career statistics comparisons in the next few paragraphs.

Successor Champions

Successor champions occur in tennis when the prior number 1 player is on the decline or has retired and there is a succession fight for number 1.  Martina Hingis, became number 1 without beating Steffi Graf; Roy Emerson remained amateur as other Aussies turned pro.  Sampras is the only open era player to win a slam in his teens, 20’s and 30’s.  Federer never won a Sampras era slam.  He became number 1 after Sampras retired and as Agassi became too long in the tooth to compete with him as displayed in the  “rope a dope” 2005 US Open final.   With Sampras and Rafter’s retirement, tennis saw the same absence of high quality serve and volleyers experienced in the 1974 – 1980 era when the Australian greats retired.

Federer’s career winning percentage of 80.66% trails Borg, Lendl and Connors.  In the diluted A32 era, he won 16 slams, a career grand slam and reached more semifinals than others (as the A32 rules enabled).  Federer won the French Open when there was only a one time slam winner in the quarterfinals.   He lost his last 3 slam finals to Nadal on every surface.  But Federer’s career slam is exaggerated and, SATERICCON analysis shows, happened with weaker fields.  Connors was undefeated in slams  in 1974 dominating Borg on clay.  Would he have won the French Open, and completed a one year slam if he had not been banned from the tourney?

What happens when we adjust for all time greats i.e. Pantheonists who have won slams on all surfaces.  In that case both Connors and Nadal enter the discussion and Federer’s all surface slam Q-rating is last using our SATERICCON methodology.  Nadal’s supremacy over Federer is dispositive  since no other top Pantheon player has had a significant losing record on every surface against another during their period of dominance.  If Federer wasn’t his era’s best, how could he be the greatest ever?

Federer’s Career All Surface Slam Quality (Q) Rank
via SATTERICON Analysis
1		Connors
2		Nadal
3		Wilander
4		Agassi
5/Last		Federer

With SATERICCON on an all surface slam record we modify it to take only the best results on that surface during a slam victory.  Each player ranked ahead of Federer beat a field more than twice as competitive as Federer’s in their respective all surface slams.   On that basis, Connors wins over Borg on hard courts and clay to win the US Open and his grass win over McEnroe at Wimbledon are dispositive with Nadal ranking a slight second.

Nadal is a classic all time great emerging as a teen like McEnroe, Borg, Sampras, Becker, Wilander and Agassi and he won a grand slam early in his career.  Federer was unable to show an extra gear vs. Nadal on any surface, unlike a Boris Becker on grass vs. Edberg or Lendl on hard courts vs. Wilander.   We saw the limit of his game.

Federer’s 82% and declining, winning percent against non-slam winners matches Sampras’ first 874 matches at the same point in his career.   Federer’s record against non Pantheon slam winners was built on a gaudy 40-5 record against 1 time slam winners and baseliners (for the most part) such as Gaston Gaudio, Thomas Johansson, Juan Carlos Ferrero and Andy Roddick  (not a natural serve and volleyer).  Sampras’ record against the power serve and volleyers Krajicek and Stich was no better than 8-10.  Outside those players, Sampras overall record is superior to Federer’s.  With respect to matches between Pantheonists, Sampras ranks first for players with more than 35 of these matches, Federer last.

Creating a Framework for Tennis Greatness

So how do you control for rule changes and the many other variables in different eras of tennis when tennis corporatists inflate statistics and smooth the way to championships?  In a tournament of all time greats, who would win?  SHOTS is a 4 step pyramid where to get to the highest level of realization, you must first complete the prior levels.  Level one is experience as a Grand Slam winner.  Level 2 is experience as a top 16 Grand Slam winner (Pantheon level player).  Level 3 is won-lost percentage vs. other Pantheonists as reflected in SITDON analysis.  Level 4 is the difficulty of slam championships won using SATERICCON analysis or Slam Quality (Q) Rating.  So although one can argue that today’s players may be taller, stronger and use better equipment or that the fields have been diluted due to rule and surface changes, they can’t argue the number of slam winners at any one tournament or their head to head record.  It is known data.

.

Our Hiearchy of Tennis Supremacy is dominated by SITDON – Normalized head to head record and SATERICCON – Normalized difficulty of winning a slam.

GOVERNING TENNIS PREDICTIVE MODELS

Tennis is governed by a few 80-20 rules of match play.  After 3-5 matches it becomes settled science as to who will win 80% of the time if one player shows dominance over the other, moreso in a slam with Pantheonists as seen in this year’s Wimbledon final between Nadal and Berdych.  The lone caveat is a “breakthrough” event when a player reaches another level of tennis such as Pete Sampras after his loss to Stefan Edberg at the 1992 US Open or Ivan Lendl after his French Open victory over John McEnroe.   They both went to a next level of greatness, dominating most opposition and fighting the remainder to a draw at worst.  In contrast, each time Federer lost to Nadal, he came back and lost worse the next year.  Watching Nadal’s career progression shows that Nadal has an extra gear that Federer doesn’t.

Absent match competitive data, style of play matters, certain players have a style that beat other players.  Kick serve and volleyer Rafter dominated Federer, similarly styled Edberg may have the same result.  Left handed Nadal dominated Federer at his prime, then left handed, kick serving McEnroe, a clutch player, may have a significant chance against Federer.   Or if Agassi who hits off the bounce early, dominated Federer, then Connors a similar lefty may have a chance against Federer and his backhand.

And finally, youth triumphs over experience when there is a significant age difference and mileage.   For example, Jimmy Connors overall career record vs. Pantheonists is less than 40%.  When isolating for when he was at his peak i.e. younger than 31 and eliminating players over 31, his normalized performance was 57% about the same as Borg.  And of course a young Federer beats a 35 year old Agassi.  Next, SITDON analysis of the Federer record – crunching the numbers.

Secadametrics Introduction

Welcome, we’ve been blogging on the finance industry for more than 2 years on another blog.  This is more of a blog site to discuss social and cultural topics using mostly quantitative approaches and some times more qualitative.   Our background is in mathematics, statistics, computer science and finance.

First posts will cover tennis and Roger Federer’s relative record as we wrap up Wimbledon and move towards the US Open.