|Hints And Notes|
Thank you for visiting my self compiled ATP Tour player statistics! They are based on the results of the singles matches played in the main round at tournaments of the ATP Tour (not Challengers and Satellites). Thus following events are not involved: qualification matches, exhibitions, Davis Cup, World Team Cup, ATP Tour Final, Compaq Grand Slam Cup.
I am not affiliated with the ATP Tour. I'm just a fanatic. One of my hobbies is collecting the ATP Tour results. I have spent a lot of time on it, and I'm happy now to be able to publish the outcome.
Errors occur naturally, so I cannot guarantee that the statistics are error free. But I promise that the results being the base for the statistics were collected in the most careful way.
In case you don't know what to do with all those numbers, I have some explanations for you what my statistics could possibly imply. Of course it is better if you make your own interpretations on the statistics, but here are some suggestions:
1. What do the statistics say?
1.1. Number of tournaments played.
Here you can see who are the hardest-working players on the Tour. One
should know that only the best 14 tournament results are counted for the
ATP Tour world rankings. So the more tournaments you play, the more bad
results (e.g. first round defeats) you are able to delete.
And, of course, players who play more than 20 tournaments every year have to be very injury-resistant.
1.2. Number of matches played.
Compared to the number of tournaments played, here the success factor plays a role. Someone can play 30 tournaments and always lose in the first round, and someone can play 16 tournaments and always make it at least to the quarterfinals. In general you could take the number of matches played as the "work density" of a player.
1.3. Matches per tournament.
Quotient of matches played and tournaments played; indicates the success of a player at tournaments. The farther he advances in each tournament, the larger gets his match account. Someone having an average 4 matches per tournament reaches the 4th round of every tournament on an average (that would mean the semifinals in a 32 player tournament).
1.4. Tournament titles.
Unfortunately, the number of tournament titles does not contain any information about the quality of those tournaments. Nevertheless a player first has to have got what it takes to win a tournament. If he does you can see in this number.
1.5. Match record.
Number of won matches in compared to number of lost matches. Unfortunately
this does not tell us against which players those maches were played. But
from a certain number of total matches you can see all opponents as one
average opponent. Assuming that this average opponent is the same for all
players with a certain number of total matches this is a good way of comparing
the levels of performance.
You can distinguish between best of 3 and best of 5 as it is possible to have some players with outstanding concentration constancy and physical condition who can use their advantages only in best of 5 matches.
You can also distinguish between the different types of courts as there are players who feel especially good on one particular surface and thus have their best results on that.
1.6. Tiebreak record.
Number of won tiebreaks compared to number of lost tiebreaks. A successful tiebreak player maybe has psychological strength, good concentration ability, an excellent serve or a little bit of luck, too.
1.7. Sets per won match.
This average number (best of 3: 2.0-3.0; best of 5: 3.0-5.0) tell you about how often a player lost a set (or two in best of 5 matches) in the matches he won. You could guess that a player who lost a set in nearly every of his won best of 3 matches (e.g. 2.94 sets per match) has starting problems or concentration problems.
1.8. Sets per lost match.
Like1.7. Here you can see which players most frequently won a set in the matches they lost. Players with low average sets per match (e.g. 2.08) maybe lose these matches in their heads already after losing the first set. Players with high average sets per match (e.g. 2.92) in fact mostly win one set, but they don't make it to win the decisive set (mind that only lost matches count; in case those players do make it, you'll find them with a high number of average sets per won match).
1.9. Games per won set.
Like 1.7., but one level down. This number naturally has a minimum of
6.0. If a player wins all of his matches 6:3 6:4 his average will be 9.5.
Players who win their sets more clearly on an average will have a lower
A low average could indicate a player being both a good returner and server.
A high average could for example indicate a player with a good serve but bad baseline play, making most of his sets end in a tiebreak.
1.10. Games per lost set.
Just like 1.9. This number gives information about how clearly a player
loses his sets.
A low average could indicate a player who does not make much effort to get up after getting down.
1.11. Longest winning streak.
Highest number of matches won in a row. To win a standard tournament
a player has to win 5 or 6 matches in a row. Winning two tournaments in
a row a player would be undefeated for 10 to 12 matches.
This number only has historical meaning and exists only to be beaten. It only tells us how long one player totally dominated the scene.
1.12. Match record as unseeded player against seeded players.
If an unseeded player beats a seeded player it means that a lower ranked
player beats a higher ranked player. Unfortunately I don't have information
about who's the favorite in each single match, so all I can do is take
the seed numbers. Favorites are seeded - almost always by world ranking
- so they do not immediately meet in the first round. When an unseeded
player plays against a seeded player one can presume that the seeded player
was the favorite.
If a player has a good match record against higher ranked players one can conclude that he would have deserved a higher spot in the world rankings. Normally he will climb up in the rankings if he goes on defeating higher ranked players, and finally he himself will be seeded. But as long as he keeps knocking the favorites out of the tournaments he'll be a "seed killer".
1.13. Tight matches record.
To win a tight match a player must either have strong nerves, good luck
or some psychological tricks.
But if a player has a high percentage every year you can exclude the luck factor.
(see the definition of tight matches below)
1.14. First round defeats.
First round defeats in relation to the number of total tournaments played.
This could indicate the quality of tournament preparation.
1.15. Wins after 0:2 sets down.
Hardly happens; players who made it have a strong will and never give up.
1.16. Defeats after 2:0 set lead.
Opposite of 1.14. Someone who did not convert a 2:0 set lead into a victory is afraid to win or is not able to come back into a match after once having lost the concentration.
Number of matches that were lost because of retiring or withdrawal.
A high number could say that a player is very careful with his body or that his body is very susceptible to injuries under pressure.
Only to complete the picture. It could take some time until I'll have
a real disqualifications ranking here. At least within a single year there
are usually not more than one or two disqualifications.
Nevertheless this list contains all players who once were "bad".
2. General information
2.1. About the making of these statistics.
Just in time for the 1997 statistics (version 3) I completely restructured
my database. The matches are now stored in a special Delphi database programmed
by myself. The making of the statistics has been automated as far as possible,
so from 1998 on I will have more time for ideas.
The only thing I have to do manually is to eliminate unreliable data. What is unreliable data? For example, a player played exactly one match in a certain year and lost this match 6:7 6:7. His average games per set are 13.0. He'd be the first one in the list. But this statement would be uncertain because 13.0 is an average. An average out of 100 values is more reliable than an average out of 2 values. It's the same with percentages: 30 won matches out of 30 are 100% just like 1 won match out of 1, but which 100% are better? So I have to define a threshold and delete those players who have too few matches, sets, tournaments whatever. I put the threshold and the resulting number of players just below each ranking. So if you miss your favorite player in one of the statistics, please consider that he didn't play enough matches or sets or something.
2.2. Definition of tight matches.
A player wins a tight match when he either had match point(s) against
him or when he had set points against him in every set he won.
Unfortunately I don't have any information about denied set points or match points. Thus the results have to be examined for the possibility of such situations.
Match points can arise at the earliest when one player has 5 games.
If one player wins a set 7:5 it is possible that his opponent had set points
at 4:5. If a set is won in the tiebreak, it is even more possible because
the opponent could have had set points at 5:4, 6:5 and of course in the
Such a tight set (i.e. a set with at least 12 games) does not necessarily have to be the decisive set of a match to make this match a tight match. Player A can win the first set, Player B wins the second set after repulsing some match points of Player A, then Player B also wins the third set. Player A lost a tight match (like 6:2 6:7 3:6).
If a player wins all his 2 (resp. 3) sets tightly it is also a tight match (because his opponent could have had set points in every set; for example 7:6 7:5 or 7:5 7:5 3:6 7:6).
2.3. What's the thing about median and mean value?
Since version 2 of the ATP Tour Statistics you can find - wherever practicable
- median and mean value added to the data.
The median is the value which divides the data into two halves. 50% of the data are above the median, and 50% are below.
The mean value is the average of all values. It is preceded by the "Ø" symbol.
In case median and mean value are the same, we have a standard distribution. That means there are as many high values as low values in the data.
Example 1: Mean value and median of all "games per won set" values of 1991-1996 are both 9.68. If you take a look at the frequency of set results, you will understand why: 9.68 equals an average set result of 6:3.68, that's something between 6:3 and 6:4, a little more on the side of 6:4.
Example 2: The mean values and medians of the match record quotients mostly are a bit higher than 50%. That means that I dropped out more players with a low quotient than players with a high quotient.
I suggest you take a look at the stuff for yourself; maybe you'll get one or another idea of how to interpret the data.