A statistical comparison of some of the 'great' matches in recent years

[Moose Malloy]

here are the +/-(winners to errors) of some of the most famous matches of the last 10 years(going just by the 'official' stats from tournament websites/ap reports/tv, etc)

2000 AO SF Agassi d Sampras

Agassi was +33, Sampras was +30


2001 W Final Ivanisevic d Rafter

Ivanisevic was +12, Rafter +38


2001 USO QF Sampras d Agassi

Sampras was +40, Agassi +36


2004 W Final Federer d Roddick

Fed was +27, Roddick was +19


2005 AO SF Safin d Federer:

Safin was +5, Fed was +13


2007 W Final Federer d Nadal

Federer was +31, Nadal +26


2008 W Final Nadal d Federer

Fed was +37, Nadal was +33

 

[diggler]

Very interesting. Thanks for posting.

I always thought that 2001 Sampras Agassi had the best quality although it didn't have the drama of service breaks.

 

[alienhamster]

The only thing I'd say here is that your measurement has an obvious bias against clay, since a good clay match will likely still have a larger number of errors than a match on any other surface.

What are your proposed candidates for recent "great" French Open matches?

 

[Moose Malloy]

The only thing I'd say here is that your measurement has an obvious bias against clay, since a good clay match will likely still have a larger number of errors than a match on any other surface.

The finals of Rome '06 was a great match. I'll try to find stats on it, but I doubt Fed was more than +10 in that one.

What are your proposed candidates for recent "great" French Open matches?

Not sure if there's been one since Agassi-Medvedev(and that was great more for the drama than quality of play)

 

[Benhur]

here are the +/-(winners to errors) of some of the most famous matches of the last 10 years(going just by the 'official' stats from tournament websites/ap reports/tv, etc)

2000 AO SF Agassi d Sampras

Agassi was +33, Sampras was +30


2001 W Final Ivanisevic d Rafter

Ivanisevic was +12, Rafter +38


2001 USO QF Sampras d Agassi

Sampras was +40, Agassi +36


2004 W Final Federer d Roddick

Fed was +27, Roddick was +19


2005 AO SF Safin d Federer:

Safin was +5, Fed was +13


2007 W Final Federer d Nadal

Federer was +31, Nadal +26


2008 W Final Nadal d Federer

Fed was +37, Nadal was +33

Thanks for those stats. They seem to confirm the wide impression that it was a pretty high-quality match in spite of the tension and the conditions.

I imagine the significantly higher proportion of bad bounces on grass (even with the new grass) in comparison to hard courts should increase the error count. The fact that it was pretty damp and windy during most of this final should is also be a significant factor in the error count.

What's amazing is that, in spite of that, the winner-to-error balance of this final (from the above list) is second only to the Sampras-Agassi 2001 USO QF - and not by much. I find that pretty amazing, and it confirms that those who describe it as a sloppy low-quality match seem to be highly confused.

(Also it must be kept in mind that the significance of that USO match was not comparable to this Wimbledon final -- not by any stretch of the imagination. It was a quarterfinal, and it did NOT even involve the number 1 player: Sampras was ranked 10, and Agassi was ranked 2).

A statistic I find interesting in this final is net approaches. They were by no means insignificant: 41% of the total points played in this match ended with one player at net. That is pretty high for two players who are considered baseliners. Federer is responsible for about two thirds of net approaches, not surprisingly since he is supposed to be the better volleyer. But he was much less succesful than Nadal at net, winning 56% of his approaches vs 71% for Nadal.

It would be interesting to see the actual winner/error count for those matches to also calculate the ratio, not just the balance.

 

[5263]

these stats show just how useless winner/UE ratio is. You can't even tell who won the match with that stat or anything much about quality. Many of the winners are from serving anyway, so the point play past the serve was probably negative in most of those.

Forced Errors are the key stat in a match and I can't figure out why they don't figure them in.

A better way using these stats you cite would be to throw out the winners and just give the UE numbers.

for example.
121 winners and 99 UEs is an ugly match! It shows no defense and sloppy misses.

33 winners and 11 UEs is a much better match and probably has about 50 forced errors as well.

I know the commentators have told us this about winner/UEs but it is the only measure they have readily available for some crazy reason.

 

[krosero]

What's amazing is that, in spite of that, the winner-to-error balance of this final (from the above list) is second only to the Sampras-Agassi 2001 USO QF - and not by much.

A good point.

A statistic I find interesting in this final is net approaches. They were by no means insignificant: 41% of the total points played in this match ended with one player at net.

Don't know how you calculated it, but you'd have to know the number of times that both were at net. True, there were not a lot of times that both were standing toe-to-toe hitting volleys, but net stats would also include times that both were running forward and hitting something (not necessarily volleys) in the net area.

Federer is responsible for about two thirds of net approaches, not surprisingly since he is supposed to be the better volleyer. But he was much less succesful than Nadal at net, winning 56% of his approaches vs 71% for Nadal.

But that's what you would expect. The more approaches, the less success rate. With approaches down to about 20, it's much easier to have a 70% success rate.

 

[Moose Malloy]

I imagine the significantly higher proportion of bad bounces on grass (even with the new grass) in comparison to hard courts should increase the error count. The fact that it was pretty damp and windy during most of this final should is also be a significant factor in the error count.

What's amazing is that, in spite of that, the winner-to-error balance of this final (from the above list) is second only to the Sampras-Agassi 2001 USO QF - and not by much.

Well, statisticians vary on the definition of what an 'unforced error' is. And the do take bad bounces into consideration(I doubt some of those whiffs went down as UE's on Sunday)

Brad Gilbert says that the Wimbledon statisticians are historically a little stingy when it comes to unforced errors(Ted Robinson mentioned this during the '07 final as well)

Still kudos to Rafa & Fed for playing so well on Sunday under those conditions.

I remember commentators blaming the wind on the fact that Agassi-Sampras didn't produce a classic in the '95 USO final.

 

[jeffreyneave]

intersting list, although the fact that rafter lost his match despite his huge advantage in the statistic suggests they don't tell the full story off a match. I suspect that since his best shot is his volley and not his serve, that in he racked up lots of volley winners. In contrast Invanisevic's best shot is his serve and that rafter's numerous failed attempts at returning the serve are counting as forced errors rather than poor play(i feel failure to return a second serve should count as an unforced error).


It certainly shows that this year's match was great match , altough in quality not much different than last years' which had fewer games.

I 'm suprised at how high the sampras/ agassi us open is. I did not rate it that great when I saw it. It probably comes out well because agassi and sampras play nore consevatively from the back of the court (agassi being renowed for his very low unforced error count) and possibly failure to return the serve,which sampras often did against the ordinary agassi serve, not counting in the stats.I remember that sampras won the match by raising his game in the tiebreaks to play 2 or 3 quality points on the Agassi serve when he got his superb forehand into play; the set play often seem to consist of many easier holds on serve. I regarded their aussie sf as more entertaining with more great shots and breaks of serve.


jeffrey

 

[NoBadMojo]

Does anyone know a really good source for match statistics? I found one before and cant seem to find the bookmark for it. One of the stats they publish is '% of service returns in play' which i think in additon to the obvious indication of how someone is returning, is a good indication of how someone was serving. They also broke down UE's by FH and BH..anyone know?

 

[iamke55]

intersting list, although the fact that rafter lost his match despite his huge advantage in the statistic suggests they don't tell the full story off a match. I suspect that since his best shot is his volley and not his serve, that in he racked up lots of volley winners. In contrast Invanisevic's best shot is his serve and that rafter's numerous failed attempts at returning the serve are counting as forced errors rather than poor play(i feel failure to return a second serve should count as an unforced error).

jeffrey

Since Ivanisevic had such a worse differential, was he playing as a baseliner in that match?

 

[crazytennis]

Since Ivanisevic had such a worse differential, was he playing as a baseliner in that match?

He was playing Rafter i.e. another volleyer and Rafter went in on everything.

 

[akv89]

here are the +/-(winners to errors) of some of the most famous matches of the last 10 years(going just by the 'official' stats from tournament websites/ap reports/tv, etc)

2000 AO SF Agassi d Sampras

Agassi was +33, Sampras was +30


2001 W Final Ivanisevic d Rafter

Ivanisevic was +12, Rafter +38


2001 USO QF Sampras d Agassi

Sampras was +40, Agassi +36


2004 W Final Federer d Roddick

Fed was +27, Roddick was +19


2005 AO SF Safin d Federer:

Safin was +5, Fed was +13


2007 W Final Federer d Nadal

Federer was +31, Nadal +26


2008 W Final Nadal d Federer

Fed was +37, Nadal was +33

Thanks for the stats, but wouldn't the winners-UE ratio be more indicative of quality of play than the differential. Based on his game, I would think that a baseliner like Nadal would have fewer winners, but also fewer unforced errors than an aggressive player like Federer. Because of the smaller numbers, the differential would tend to be smaller. I would appreciate it if you could post the winners-UE ratio as well.
Thanks

 

[superman1]

Sampras/Agassi was better.

 

[Benhur]

Don't know how you calculated it, but you'd have to know the number of times that both were at net. True, there were not a lot of times that both were standing toe-to-toe hitting volleys, but net stats would also include times that both were running forward and hitting something (not necessarily volleys) in the net area.

I calculated it from the match statistics at the Wimbledon site. From the total points won by each player you see there was a total of 413 points played in the match. Net approaches by Federer add up to 117. Nadal's net approaches add up to 53. So (117 + 53)/413 = 41%. I was surprised to see how high this seems for two players who are in essence baseliners. Somehow it does not look that high when you just look at the numbers at first sight, or when you just remember the match. But in fact the ratio of points ended at net is not that far from being half the total.

 

[Moose Malloy]

I calculated it from the match statistics at the Wimbledon site. From the total points won by each player you see there was a total of 413 points played in the match. Net approaches by Federer add up to 117. Nadal's net approaches add up to 53. So (117 + 53)/413 = 41%. I was surprised to see how high this seems for two players who are in essence baseliners. Somehow it does not look that high when you just look at the numbers at first sight, or when you just remember the match. But in fact the ratio of points ended at net is not that far from being half the total.

Federer only came to net 75 times. He was 42 of 75 there(you added 42 & 75)

Nadal came to net 31 times.

So the total is 75 + 31. That's 106 points or 26% of all the points.

 

[Benhur]

Well, statisticians vary on the definition of what an 'unforced error' is. And the do take bad bounces into consideration(I doubt some of those whiffs went down as UE's on Sunday)

Well obviously the whiffs shouldn't go down as errors, as they are usually due to a very bad bounce. But there is a range of "badness" in bad bounces, and many of them are not obvious to the spectator, yet they are significant enough to draw an error from the player. I imagine these are much more frequent on grass.

But in general, yes, the "unforced error" count is by far the most subjective. It all depends who is counting and how they interpret the word "unforced". For example, a player pulling the trigger or going for a very aggressive shot, and missing, is an unforced error, but it should belong in a different category as someone trying to do nothing special and still missing.

Still I suspect that grass should normally produce a higher UE count than hard courts, unless those doing the counting on grass are extremely stingy as you say. There are many minor bad bounces you cannot really detect unless you are the one hitting the ball.

 

[Benhur]

Federer only came to net 75 times. He was 42 of 75 there(you added 42 & 75)

Nadal came to net 31 times.

So the total is 75 + 31. That's 106 points or 26% of all the points.

Oh! You are right! That was a big mess-up on my part. I somehow must have interpreted the 42 and 75 figures as points loss/won at net, and just added them up.

Yes, 26% sounds a lot more like it. I kept thinking about this, saying to myself: "it sure didn't feel like they were going to net that often" and attributing it to a kind of misperception (trusting my numbers). I had written my erroneous net approach numbers on a piece of paper, 117 for Federer and 53 for Nadal, and it never occurred to me to re-check how I had arrived at those numbers.

Thanks for catching this embarrassing boo-boo.

 

[Moose Malloy]

2001 W Final Ivanisevic d Rafter

Ivanisevic was +12, Rafter +38

I'm pretty sure the stats I listed for this match are incorrect. Got it from this article:

http://www.usatoday.com/sports/tennis/01wim/stories/2001-07-09-game.htm#more

The article has Goran at 42 winners, 30 unforced errors, Rafter at 49 winners, 11 unforced errors. I don't believe aces were counted in the winner count. Goran had 27, Rafter had 13. If aces were counted in that winner total, that means that Goran only had 15 winners outside of aces in the entire match, which seems unbelievable.

When I add the aces, Goran now has a differential of +39, Rafter at +51, which makes more sense for a match of this length & a match that Goran won(and both players were S&V on 1st & 2nd serve, that type of tennis usually lead to high winners/low errors)

 

[Kelcher]

I think a key difference between this match and many of the others you listed is that both Nadal and Federer are such great defenders. Shots that would go for winners against Roddick, Sampras or Ivanisovich were getting returned in this match, reducing the amount of winners, and forcing each player to go for more, increasing the amount of errors.

 

[5263]

I'm pretty sure the stats I listed for this match are incorrect. Got it from this article:

http://www.usatoday.com/sports/tennis/01wim/stories/2001-07-09-game.htm#more

The article has Goran at 42 winners, 30 unforced errors, Rafter at 49 winners, 11 unforced errors. I don't believe aces were counted in the winner count. Goran had 27, Rafter had 13. If aces were counted in that winner total, that means that Goran only had 15 winners outside of aces in the entire match, which seems unbelievable.

When I add the aces, Goran now has a differential of +39, Rafter at +51, which makes more sense for a match of this length & a match that Goran won(and both players were S&V on 1st & 2nd serve, that type of tennis usually lead to high winners/low errors)

I think Aces are normally included.

 

[Cindysphinx]

Thanks for the stats, but wouldn't the winners-UE ratio be more indicative of quality of play than the differential.

It might depend on what kind of tennis you like, but I think the best stat to measure match quality would be average number of strokes per rally.

When they put together those features of the "10 best points in a tournament" there is never an ace included. That's because aces are kind of dull. The stuff that makes the viewer gasp is the long rally with lots of crazy defense.

Give me a match with lots of long rallies over a match with lots of "winners" (including aces) any day.

 

[BeHappy]

yeah, these numbers are meaningless to anyone who has any understanding of tennis.

They:
1)Never take into acount forced errors
2)Are compiled by amateurs
3)Are highly subjective
4)Aren't consistant, eg: s&v
5)don't take into acount conditions like: wind, surface speed etc
6)Don't take into account the standard of defense etc

They are almost meaningless and should be taken with a large pinch of salt.

 

[krosero]

I'm pretty sure the stats I listed for this match are incorrect. Got it from this article:

http://www.usatoday.com/sports/tennis/01wim/stories/2001-07-09-game.htm#more

The article has Goran at 42 winners, 30 unforced errors, Rafter at 49 winners, 11 unforced errors. I don't believe aces were counted in the winner count. Goran had 27, Rafter had 13. If aces were counted in that winner total, that means that Goran only had 15 winners outside of aces in the entire match, which seems unbelievable.

I think you're right, 42 non-service winners is far more realistic for that match. And he had 40 non-service winners in the '98 final, also over 5 sets.

So then this boxscore is like the one in the W. Post for the '00 final, which said that the winners included service when they actually didn't.

However, this makes me think that maybe the line in which the unforced errors is reported does not include the double-faults.

Obviously, a double-fault is an unforced error, but what I mean is that if the aces were not included in the row for Winners, then how do we know that the doubles were included in the row for Unforced Errors?

And that sort of uncertainty, I think, might be one of the risks in comparing differentials across the years and from many different sources.

(Goran had 16 doubles, and if they're included in the ue's, it means he had only 14 ue's on other strokes for the whole match, which again seems too low).


I have these stats in my notes on this match:

Cox News Service, Palm Beach Post:

Ivanisevic ripped 27 aces and had 80 winners against 33 unforced errors in the course of the three-hour match. Rafter, incredibly, had 66 winners against 11 unforced errors.

The Sports Network:

Rafter played an extremely clean match, committing only 11 unforced errors and striking 49 winners.

The former of these two seems to be including service while the latter does not.