Wednesday, April 1, 2015

Using Slam Scores: Overrated and Underrated

I've expended quite a bit of virtual ink in this space discussing Grand Slam Scores, which evaluate historical performance in Slams as a way of comparing tennis players. But any player evaluation method worth its salt should be able to do more than just supply an answer to "which guy is better?" questions. Are Grand Slam Scores up to that task?

Let's find out. We'll start by applying them to the question of which tennis players are the most likely to be overrated and underrated in conventional opinion.

Any attempt to determine whether players are over- or underrated must first answer the obvious question of how those players are rated in the first place. My basic assumption is that conventional opinion will generally evaluate players (historically speaking) based solely on Grand Slams won. Therefore, players who accomplish things other than winning Slams are likely to be underrated, while players who win more Slams than would be expected from their other feats will tend to be overrated.

Using Slam Scores to address this question has at least one significant limitation, which is that it prevents us from examining performance in other events. This ends up being less damaging than might be expected, however, because players who make repeated deep runs in Slams also tend to win quite a few matches outside of them. So Slam Scores should at least be capable of functioning as a starting place for this analysis.

Here are the highest-ranked players when considering Slam Scores accumulated only in Slams that the player did not win:

1. Ivan Lendl 9.24
2. Jimmy Connors 9.12
3. Roger Federer 8.34

Federer's early appearance reveals an obvious problem with using just the raw total, which is that it ignores the fact that Federer has won 17 Slams, and therefore it would be virtually impossible for him to be the third-most underrated player ever; it's hard to be underrated when you're the plurality choice as the greatest of all time.

What if, instead of total score in Slams not won, we use that quantity as a fraction of total score? That would adjust Federer downward...

1. David Ferrer, Tomas Berdych, Todd Martin, and everyone else who has never won a Slam, 1.0

... but presents an even more pressing problem.

The formulation I settled on is a combination of those two options - specifically, the combination achieved by taking their product. It works out to ((Slam Score - Slams Won)^2)/(Total Slam Score).

Here are the top 20, in reverse order:

20. David Nalbandian, 2.84
19. Jo-Wilfried Tsonga, 2.89
18. Tim Henman, 2.98

It's probably worth pointing out that for a player who hasn't won a Slam (like all three of those listed above), this value will be identical to total Slam Score. If you remember the "best players who haven't won a Slam" post, you know we're not done with this group yet.

17. Lleyton Hewitt 2.99 (Total 6.36, 2 Slams won)
16. Stefan Edberg 3.12 (12.16, 6)

I can attest to Edberg being underrated, at least, because I spent much of my life as a pretty casual tennis fan, and had never heard of this two-time year-end #1 player until my attention to the sport became more serious.

15. Goran Ivanisevic 3.17 (4.97, 1)
14. Tom Okker 3.22 (no Slams won)
13. Arthur Ashe 3.28 (8.18, 3)
12. Novak Djokovic 3.30 (15.05, 8)

Novak would have been four spots higher on the list before winning the Australian Open this year. If he keeps winning Slams, of course, his conventional wisdom place in history will improve, and his position on the list of underrated players will correspondingly decline (a consequence that I suspect he would be willing to accept.)

11. Michael Chang 3.38 (5.19, 1)
10. Tomas Berdych 3.48 (no Slams won)
9. Ken Rosewall 3.49 (15.31, 8)

I've written about Rosewall a bit before, and may eventually do so at more length. So much of his career was spent not playing Slams at all that a method that uses Slams exclusively to evaluate him is going to have severe inherent limitations. So for now, I'll say that I think he's probably quite a bit more underrated than this list captures, at least in comparison to his more famous contemporary compatriot, Rod Laver.

8. Todd Martin 3.52 (no Slams won)
7. David Ferrer 3.96 (no Slams won)
6. Andre Agassi 4.11 (16.15, 8)
5. Andy Roddick 4.28 (6.12, 1)

Roddick somehow spent his career being considered a disappointment because he wasn't Pete Sampras, and kept losing Grand Slam finals to Roger Federer. I have a hard time describing how ridiculous I find that to be, and as a result, I am very comfortable with listing him as one of the most underrated tennis players ever.

4. Andy Murray 4.65 (8.16, 2)
3. Jimmy Connors 4.86 (17.12, 8)
2. Ivan Lendl 4.95 (17.24, 8)
1. Tony Roche 4.98 (6.84, 1)

Tony Roche may not actually be the most underrated tennis player of all time. There are factors working against him, most notably the fact that his career began in the '60s, and therefore compiled many of his accolades when several of the world's best players weren't eligible for the Slams.

On the other hand, Roche is probably the fifth-best Australian tennis player who peaked in that decade (trailing Laver, Rosewall, Emerson, and Newcombe). So you can kind of see how he might have suffered from a publicity deficit.

Lendl and Connors are placed very similarly here (and just about everywhere else Slam Scores go). Both of them played forever, spent about five years' worth of weeks at #1, and are often left out of discussions of the very greatest players of all time, because they "only" won eight Slams apiece.

And that leaves us one more player to discuss: Andy Murray, who has some of the worst timing imaginable for being a merely great tennis player.

Murray has played 23 matches in the semis or finals of a Grand Slam, and won 10 of them. This is a very respectable record in historic terms. It becomes moreso when you remember that 16 of those 23 matches have been against three of the greatest players ever. Murray is 4-12 in semis or finals against Federer, Nadal, and Djokovic; he is 6-1 at those levels against everyone else. He has spent his entire career fighting to emerge from three prodigiously long shadows, and has only occasionally succeeded - but his other contemporaries have all fought the same battle, and none have fared nearly so well as Murray.

Murray sits in fourth on the list of underrated players right now, but as he is still active, his score is still in flux. And if, say, he makes two semifinals in the last three Slams of the year but doesn't win any of them, he'll become the player with the most-likely-to-be-underrated Grand Slam career of all time.

Which is what happens when you have the unfortunate distinction of being clearly the fourth-best member of a group of four phenomenal players.

Now, for the flip side: the players likely to be overrated. The lowest "underratedness score" in the database belongs to Juan Aguilera, whose Slam Score of 0.20 comes with no titles (obviously). Many of the other players in this area of the list are in the same boat, and none of them could fairly be described as overrated - not many people are placing Juan Aguilera particularly high on any historical player rankings. But if we filter the list to include only Slam winners, we get a more reasonable collection.

1. Gaston Gaudio 0.24 (Total 1.63, one Slam won)
2. Gustavo Kuerten 0.31 (4.13, 3)
3. Brian Teacher 0.41 (1.88, 1)
4. Sergi Bruguera 0.57 (3.38, 2)
5. Albert Costa 0.67 (2.22, 1)
6. Andres Gomez 0.75 (2.32, 1)
7. Mark Edmondson 0.76 (2.33, 1)
8. Thomas Johansson 0.76 (2.33, 1)
9, Marin Cilic 0.81 (2.39, 1)
10. Bjorn Borg 0.86 (14.55, 11)

We've seen most of these guys on related lists - worst total scores to win a Slam, highest percentage of total score earned at one Slam, and the like. We've also seen Borg before and looked at the points in his favor and against him; a method predicated on looking at performance in Slams other than those the player wins is not going to smile on him, because he stopped playing Slams immediately after he stopped winning them. If you accept the initial premise of the metric, then this is a reasonable list of players likely to be overrated when considering Slams only.

Side note, which has no effect on this post at all: I recently updated my Slam Scores database to include an additional subset of players. It previously contained everyone who has appeared in the top 10 of the ATP rankings, as well as everyone who has made a Grand Slam final in the Open Era (plus Roy Emerson as a bonus entry). Now, it also includes all players who have at least 300 match wins as listed by the ATP.

There are 151 players to date who have accumulated 300 victories, and 28 of them had neither reached the top 10 nor made an Open Era Slam final. One of those 28 was Emerson. Notable among the other 27 are Cliff Drysdale (both because he has the highest total Slam Score of the group at 2.71, and because he has long been one of the leading TV announcers for tennis), Fabrice Santoro (the only player of the 201 now listed to have played in more Slams than Roger Federer - 70 in total, making the round of 16 in only four), and Tomas Smid (the only one of the 27 new arrivals to have won over 500 matches in his career). There's also a trio of active players in the group - Feliciano Lopez, Philipp Kohlschreiber, and Jarkko Nieminen (whose recent reaching of the 400-win milestone inspired me to undertake this particular augmentation). I also considered selecting players based on total titles won, but it turns out to be redundant; adding all of the 300-win players to the existing list also covers everyone with at least 10 titles.

Up next in Slam Scores will be an alternative look at players likely to be over- or underrated, this one based on the difficulty of opposition faced in finals. It may have to wait a while, though, because baseball season is coming up (which will bring with it the resumption of the Game of the Day posts), and also because this happened recently, and I tend to become easily distracted when handed mountains of data to climb. (Seriously, it took me three days of not-exactly-strenuous work to run Melog ratings for the entire 2006 season, where it used to take months of data entry to get a full year's worth of numbers ready. This expands my research possibilities incomparably.)

No comments:

Post a Comment