Stats for Sampras and Agassi (9 Slams)

So now we have threads for all 9 meetings in Slams between Sampras and Agassi.

1990 USO - http://tt.tennis-warehouse.com/showthread.php?t=196287

1992 RG - http://tt.tennis-warehouse.com/showthread.php?t=388910

1993 Wimbledon - http://tt.tennis-warehouse.com/showthread.php?t=229031

1995 AO - http://tt.tennis-warehouse.com/showthread.php?t=206448

1995 USO - http://tt.tennis-warehouse.com/showthread.php?t=230227

1999 Wimbledon - http://tt.tennis-warehouse.com/showthread.php?t=185518

2000 AO - http://tt.tennis-warehouse.com/showthread.php?p=2836627#post2836627

2001 USO - http://tt.tennis-warehouse.com/showthread.php?p=2845813#post2845813

2002 USO - http://tt.tennis-warehouse.com/showthread.php?t=224267


This thread is to compile what I thought were the most interesting stats in the 9 matches (some just for fun).


SERVICE

Sampras had his highest service percentage at 2000 AO, with 63.46%. His lowest was 41% at 1992 RG.

Agassi had his highest service percentage at 1990 USO, with 77%. His lowest was 43% at 1999 W.

The loser usually served more points than the winner, with two exceptions. At 1995 USO, Sampras served 6 more points than Agassi (because he served one more game). And at 2002 USO, Sampras served 27 more points than Agassi, mostly in the last two sets when he was tiring.


Sampras’ success on first serve (we don’t have stats for 1992 RG or 1993 W):

92% at 1990 USO
88% at 1999 W
86% at 1995 USO
80% at 2000 AO and 2002 USO
79% at 2001 USO
76% at 1995 AO


Sampras’ success on second serve (again minus 1992 RG and 1993 W):

58.8% at 1990 USO
58.7% at 2001 USO
54% at 1995 USO
52% at 2002 USO
49% at 1999 W and 2000 AO
44% at 1995 AO


WINNERS/ACES

Agassi’s FH was usually the stroke that produced the most clean winners, except at 1990 USO and 2002 USO (at the latter, the leading stroke was Pete's forehand volley).

However Sampras, except at 1992 RG, usually led all strokes with his aces.

Sampras served a career high of 37 aces at 2000 AO. This amounts to aces on 24% of all the points he served in the match, his highest rate in the nine meetings. And he had just 5 double faults (his second-lowest in the nine meetings), giving him a service "differential" of 32, possibly a career high too.

His low was 8 aces at 1992 RG.

Agassi served 18 aces at 2001 USO (I wonder if that was a career high for him), and none at 1990 USO.


The highest rate of winners belongs to Sampras at 2002 USO: he struck a clean winner or ace on 27% of all the points played. Next highest was his 24.7% at 2000 AO, and his 24.1% at 1999 W.

Agassi’s highest rate was 19% at 1995 AO.


ERRORS

Agassi usually made fewer total errors (forced and unforced) than Sampras. He made more than Pete only in the two Wimbledon matches.

Sampras served a high of 13 double faults at 2002 USO. Agassi served 7 at 1993 W and 6 at 1999 W -- the only matches in which he served more df's than Sampras.

Each man served just one double-fault at 1990 USO (and Agassi served one at 1992 RG).

Sampras usually led Agassi in drawing return errors, except at 1999 W and 2000 AO.


PUBLISHED WINNERS AND ERRORS

The highest “official” tally of winners belonged to Sampras at 2000 AO: 86 winners. He had 84 at 2002 USO and 80 at 2001 USO.

Agassi’s highest “official” tally was 55 winners at 2001 USO.

The highest published figure for unforced errors probably belonged to Sampras at 2000 AO: 56. I don’t have his figure for 1992 RG but it was probably lower.

Agassi’s highest published count was probably at 1993 W but I don’t know what it was.


MISCELLANEOUS

Their longest match in terms of points (338), games (52) and elapsed time (3:32) was their four-setter at 2001 USO. Their five-setter at 2000 AO tied that match in games but not in points (304) or in elapsed time (2:55).

They split 4-4 in tiebreaks, and 1-1 in five-setters.

The winner of the first set took every match except at 1995 AO and 2001 USO.

Sampras always came to net more than Agassi, except at 1992 RG (in fact Sampras had more winners there from ground strokes than from volleys and overheads, which also happened at 1995 AO).

Agassi always got more winners from ground strokes and rarely ventured to net. His 110 approaches to net in the 1988 RG semifinal against Wilander remain, I think, unique in all the stats we've done for him -- unless Moose knows of something else.

Aggressive Margin

Finally I've calculated their Aggressive Margins where I have enough info to do so.

There's a clear trend or indication that both Sampras and Agassi got better as they grew older.


2002 USO Sampras - 30.32%
2002 USO Agassi - 21.30%

2001 USO Sampras - 34.62%
2001 USO Agassi - 30.47%

2000 AO Sampras - 24.34%
2000 AO Agassi - 26.32%

1995 USO Sampras - 21.61%
1995 USO Agassi - 16.53%

1995 AO Sampras - 18.29%
1995 AO Agassi - 23.35%

1990 USO Sampras - 21.51%
1990 USO Agassi - 6.40%


Thanks to Moose for pointing me to this method of measuring performances. I'm sure others here have heard about it already but I'll explain it for those who don't know.

A player's Aggressive Margin is a measure of how many points he is able to win with winners or forcing shots, compared to how many unforced errors he made. It's a measure of how efficiently aggressive he was.

All you need to know to calculate it is the Total Points Won and the unforced errors for each player.

If you know, for example, how many points Sampras won in a match, and you subtract from that number Agassi's unforced errors, the result is the number of points that Sampras won by hitting winners/aces or by getting forced errors from Agassi. Those are Sampras' Aggressive Points. Subtract from that number Pete's own unforced errors, and you have his Aggressive Margin. It's expressed as a percentage of the total points played in the match.

This method seems to accurately predict who won a given match. The method comes from Bill Jacobson and he has an article about it in Tennis Magazine from late 1994.

A detailed online article by Alberto Brignacca is here, and in HTML.

I don't have the Unforced Errors for the Sampras-Agassi meetings at RG and W, which in a way is a good because the list above is restricted to their meetings on hard court.

Jacobson writes that Aggressive Margins (AM's) "vary sharply by surface. On clay, longer rallies result in fewer winners and more unforced errors, and Aggressive Margins are 10-20% lower than on grass."

He says that Sampras' AM in winning Wimbledon in 1994 was 32.3%, but it's unclear to me whether he means the whole tournament or just the final.

Anyway what we've got on hard courts indicates that they were improving over time.

It also strikes me that the two men's performances at the 1995 AO are higher quality than what they produced at the USO -- even though you would expect the AM's to be lower on Rebound Ace, the slower court.

It gets a bit tricky here because this is the one match where I'm working without a precise number. The New York Times has Sampras making 50 ue's in the AO final, and Agassi committing "half as many mistakes", so I've given him 25.

If, for example, he really made 27, his AM would go down to 22.57%.

Just shows that there's a little bit of play there.

And that's really always the case with unforced errors, since they're not an exact science.

A final thing to keep in mind is that some of these percentages would change if published figures for Unforced Errors do not include the double-faults. I have several concrete indications that they do include them. For example, using the Unforced Errors at USrg for the 2005 USO final, I get the same AM's as Brignacca does in his article. And Leo Levin, when discussing statistics, has said that a df is by definition an unforced error. I just can't say for sure that all sources at all times in the past have included df's when reporting Unforced Errors; I wonder if even a professional statistician can know that.

Having said all that, the '95 numbers remain intriguing -- particularly because you would expect the AM's to be lower on Rebound Ace than on the faster DecoTurf at the USO, but the numbers suggest that the AO final was a higher quality match.

Perhaps it just hasn't gone down that way because the USO match had all the buildup and drama.

In calculating the Aggressive Margins, I found a discrepancy for the 1990 USO and I found that I'd made a simple error in calculating Agassi's total errors there (nothing to do with unforced errors). He made 64, not 74. I've edited that thread; and no discrepancies came up for the other matches.

It doesn't affect the AM's above.

One more small correction. I streamlined everything in Excel to get the program to calculate as much as possible, with minimum input from me (now I just type in the ue's and Total Points Won). Excel added up the total points that each player won in the 1990 USO correctly; I had done that manually before (in my head) and I was slightly off.

The AM's are slightly affected: Sampras goes up 1.4%, Agassi almost half a percent. I've edited the post.

Regarding your stat-correction krosero -- I saw that glaring mistake before you posted your explanation. How could you even make such an unforgivable mistake in the first place? I almost spilled my coffee when I saw it didn't add up...

I mean if you can't count correctly in your head in a Las Vegas-casino -- you don't stand a chance of getting buried alive in the desert by Joe Pesci!

Borgforever said:

I mean if you can't count correctly in your head in a Las Vegas-casino -- you don't stand a chance of getting buried alive in the desert by Joe Pesci!

Click to expand...

I would get buried alive in the casino! I'd much rather be on a tennis court.

Agassi had his highest service percentage at 1990 USO, with 77%. His lowest was 43% at 1999 W.

Click to expand...

Aww man. No wonder he lost that friggin match. C'mon Andre it's Wimbledon for christ sakes, get the serve in the box.

Hey krosero -- I wonder if you would like to describe your thoughts a little more in depth about the Sampras-Agassi-rivalry in GS-tourneys?

I would like to hear your analysis on it's development and what you consider were the major factors and the whys concerning how it played out.

I mean you seem to have a very good lock on this particular subject (among several others I believe)...

Borgforever said:

Hey krosero -- I wonder if you would like to describe your thoughts a little more in depth about the Sampras-Agassi-rivalry in GS-tourneys?

Click to expand...

Sure, I can do that. Right now I'm getting stats on some Borg matches and I'm going to concentrate on that; I do too much when I split my attention over two big topics. But when that's done I'd like to say more about the Aggressive Margins, and to get more of them if I can.

If anyone can find Agassi's unforced errors in '99W, and Becker's in '95W, we can get Pete's AMs for those two matches.

Cool. More Borg-stats? Hmmm... I can't wait...

Looks like Sampras ownage.

Borgforever said:

Hey krosero -- I wonder if you would like to describe your thoughts a little more in depth about the Sampras-Agassi-rivalry in GS-tourneys?

I would like to hear your analysis on it's development and what you consider were the major factors and the whys concerning how it played out.

Click to expand...

Well Borgforever, I don't know that I have anything original to say about a rivalry that's been analyzed so much. And I don't think that our stats turned up anything really surprising, because a lot of their matches already had published stats, and our counts just added more detail.

But one thing did surprise me, not in their matches, but in Agassi's net approaches against Wilander at the 1988 French Open: 110 per NBC, and app. 90 in our own counts. He had 33 winners from volleys and overheads, 32 from ground strokes; I can't think of another match where a majority of his winners came at the net. Usually it's not even close.

Against Sampras, I counted 31 approaches at the '92 FO, and that may have been his highest number against Pete in a Slam. It was on clay and only a few years after the Wilander match, so I don't know, maybe he still had some of the adventurousness he showed in '88. Maybe he came in 31 times simply because he felt safer approaching with his ground strokes on clay. Or maybe it was because Sampras only approached 19 times himself.

Whatever the case, at the faster Slams he rarely did much at the net, and in that last USO final in '02 he approached just 13 times. So in a way you could say that in this rivalry, Agassi was coming in less as the years went by, while Sampras (due to Annacone) was coming in more.

The reason I think it's important is that Agassi never really found a way to take Pete out of his comfort zone. In his book, Sampras says that Agassi was good enough to force him to do more (like coming to net constantly on his second serve); but he also says that he wasn't taken out of his comfort zone.

So the rivalry was basically cast as the best server against the best returner. That was the big tag line in the buildup before their '95 USO final, and I remember thinking that by itself, a great serve against a great return wasn't going to make a great rivalry; on a fast surface the server will usually win, all else being equal (I think Andre had the edge on Rebound Ace, but not by much).

When I took the Wilander-Agassi stats, it was Moose who made the contrast against the later Agassi; I hadn't seen the Sampras-Agassi matches in years. Now I've seen them a lot, so when I went back recently to watch a set of the Wilander match, I was more surprised than ever, watching Andre run up to net and handle low and high volleys, stretch volleys, overheads, like they're no problem at all. His net game is not really mature; it does feel a little like a fearless experiment he's trying; but the talent is evident. And it does make you wonder what he could have done if he had mixed it up more.

Who knows what difference it would have made. But I'm coming at this from the perspective of having seen it make a difference for Wilander himself: he broke through in '88 by approaching the net more than he ever had: 137 times in the USO final against Lendl. At the start of '88, I did not think that Wilander was really going to find a way to get more aggressive. I'd seen him trying, with discomfort, to get up to net in the '87 USO final, so the last thing I would have expected was that he'd nearly double his net approaches in a rematch.

I had to see it to believe it was possible -- and that it could work.

Same thing with Borg, who was famous for camping out well behind the baseline to receive serve: I would not have believed that he'd ever return McEnroe's serves from inside the baseline, unless I saw it myself (at AKAI); nor would I have believed that it would work.

You get so used to seeing certain players in a certain part of the court, the least I can say is that it's interesting to see them break their own habits. Quite apart from the rivalry with Sampras, I would have liked to see Agassi mixing it up at the net more than he did.

i know this thread is about former players...but it would be interesting to compute aggressive margins of current players and compare them to say guys like sampras and agassi

Hello Krosero. Funny, I hadn't heard "aggressive margin" for quite a while. I was about to post about Bill Jacobson, founder of Computennis, when I saw you are aware of him.

One thing I noticed in skimming this is you mentioned:

"clear trend or indication that both Sampras and Agassi got better"

This is an interesting statement, which of course could be interpreted many ways.

A few points:

It's not clear to me that Agassi's margin was increasing. Particularly if you eliminate his abysmal 1990 final, where he hit one decisive winner in the entire match, which was over before he even shook off the nerves.

Sampras' numbers do appear to increase. Perhaps a result of his switch to SV play....or another very plausible explanation is to remember that they were both plausibly slower in the late years.....one also wonders if another reason might be increased familiarity with each others games and big occasions.

World Beater said:

i know this thread is about former players...but it would be interesting to compute aggressive margins of current players and compare them to say guys like sampras and agassi

Click to expand...

I'd hazzard a guess that it would be higher for current players. It's a result of the style being played. A relatively one-dimensional baseline slugfest, for which the players are exquisitely trained. In essence, like long-time practice partners, they set each other up to hit their best shots. In the past, chip and charge, serve and volley, crush and rush, off-pace baseliners, flat hitters, tended to result in more errors.

All good comments, although I have not been able to confirm that the figures I used for unforced errors included double-faults. True, double-faults are only going to change the AM's by a few percentage points; but still that needs to be settled before you could really analyze, for example, whether these two players were improving, how they compare to current players, etc.

bet said:

I'd hazzard a guess that it would be higher for current players. It's a result of the style being played. A relatively one-dimensional baseline slugfest, for which the players are exquisitely trained. In essence, like long-time practice partners, they set each other up to hit their best shots. In the past, chip and charge, serve and volley, crush and rush, off-pace baseliners, flat hitters, tended to result in more errors.

Click to expand...

you may be right but for the wrong reasons.

its quite difficult to hit a winner these days because players move and defend so well.

World Beater said:

you may be right but for the wrong reasons.

its quite difficult to hit a winner these days because players move and defend so well.

Click to expand...

Not sure what you're saying, if it's that hard to hit winners today (I don't think it is), then the margin would be lower today. Particularly since play is overwhelmingly done at the baseline.

In any case, I think it's largely a myth that the players today are faster. They were ALWAYS fast. Very fast. That has always defined pro tennis. The fastest players from the 70's are ever bit as fast as the fastest players today. The average may have gone up but the amount would be absolutely miniscule if measured, it'd be amazing if the top 100 is 2% faster than the top 100 from 30 years ago. Even that figure is doubtful. Steroids and doping aside, human speed capabilities had maxed out long, long ago.

The idea that players today are so much faster, is an interesting sociological/psychological characteristic, it's basis in reality is.....rather weak.

bet said:

Not sure what you're saying, if it's that hard to hit winners today (I don't think it is), then the margin would be lower today. Particularly since play is overwhelmingly done at the baseline.

In any case, I think it's largely a myth that the players today are faster. They were ALWAYS fast. Very fast. That has always defined pro tennis. The fastest players from the 70's are ever bit as fast as the fastest players today. The average may have gone up but the amount would be absolutely miniscule if measured, it'd be amazing if the top 100 is 2% faster than the top 100 from 30 years ago. Even that figure is doubtful. Steroids and doping aside, human speed capabilities had maxed out long, long ago.

The idea that players today are so much faster, is an interesting sociological/psychological characteristic, it's basis in reality is.....rather weak.

Click to expand...

well i compare this gen of players to prev gen for example.

federer, nadal, djokovic and murray are superior movers and defenders than sampras, agassi, becker, kuerten etc.

only hewitt, chang can compare but they are not as strong athletically to hit passing shots from several feet behind the baseline.

what im saying is that the aggressive margin might be higher but for reasons due to tennis level increasing. federer runs through most of his opponents like a hot knife through butter...destroying guys in his prime whereas sampras used to win sets 6-4 and 7-6 even against journeymen. the reason why pete did is not important but the result is his agg.margin is lower.

nadal is similar in that he is hyper consistent and almost always forces errors out of the opponent while making almost no errors himself. so his agg margin should be high as well.

the fact that the top players especially defend so well and are able to hit winners and force errors from seemingly defensive positions makes their agg.margins higher.

so there is a group of players who are becoming so consistent in performance and results that they are totally outclassing the rest of this competition. this group is federer, nadal, djokovic and now murray with davydenko being a honorable mention.

World Beater said:

well i compare this gen of players to prev gen for example.

federer, nadal, djokovic and murray are superior movers and defenders than sampras, agassi, becker, kuerten etc.

only hewitt, chang can compare but they are not as strong athletically to hit passing shots from several feet behind the baseline.

what im saying is that the aggressive margin might be higher but for reasons due to tennis level increasing. federer runs through most of his opponents like a hot knife through butter...destroying guys in his prime whereas sampras used to win sets 6-4 and 7-6 even against journeymen. the reason why pete did is not important but the result is his agg.margin is lower.

nadal is similar in that he is hyper consistent and almost always forces errors out of the opponent while making almost no errors himself. so his agg margin should be high as well.

the fact that the top players especially defend so well and are able to hit winners and force errors from seemingly defensive positions makes their agg.margins higher.

so there is a group of players who are becoming so consistent in performance and results that they are totally outclassing the rest of this competition. this group is federer, nadal, djokovic and now murray with davydenko being a honorable mention.

Click to expand...

Well I completely disagree with your assertions about movement and I find this idea of an elite group rather strange. Even if their margins were higher, the tour in general wouldn't be doing very well with their margins against them.

In any case, Certainly Fed, Nadal, Djoko, Davy are not better than than Sampras, Agassi, Becker, Safin, etc. so, this would indicate a general lowering of the depth of the rest of the tour. I don't necessarily think this is the case, but if you wish to argue that those elite players have higher margins because they are so much better than their opponents, ok.

Again, I tend to think that changes in margin would more likely be due to the homogenization of game style.

This also in large part is why Nadal and Federer tend to win so many matches. Everybody they play is limited to playing their own baseline game which is not as good. As a result, they may have good rallies but invariably they tend to come up short. If Nadal and Federer had to play a Richard Krajicek, Stich, Mecir, Leconte....whoever, even though they may win more matches against these guys than lose, because they are better, they will also suffer more losses than they will playing inferior baseliners.

A SV player ranked 30 has a much better chance of taking advantage of a bad day from a top player than a baseliner ranked 30. The baseliner has to hit many balls and the top player has lots and lots of time to find his form and pull himself together. Especially when the baseliner plays the same topspin game every other baseliner does. He also knows he is a worse baseliner and fears that when the other player finds his form, it's all over. The SV'er may know he is not as good as the top player, but he also knows he has knockout punches, can potentially end it quickly and does not need to beat the top player at his own game.

bet said:

Well I completely disagree with your assertions about movement and I find this idea of an elite group rather strange. Even if their margins were higher, the tour in general wouldn't be doing very well with their margins against them.

In any case, Certainly Fed, Nadal, Djoko, Davy are not better than than Sampras, Agassi, Becker, Safin, etc. so, this would indicate a general lowering of the depth of the rest of the tour. I don't necessarily think this is the case, but if you wish to argue that those elite players have higher margins because they are so much better than their opponents, ok.
.

Click to expand...

the tour may not be doing better in terms of agg.margins. but i want to compare the great players in each gen.

well i didnt argue that they are better. i said they are better movers. there is a difference.

bet said:

Again, I tend to think that changes in margin would more likely be due to the homogenization of game style.

This also in large part is why Nadal and Federer tend to win so many matches. Everybody they play is limited to playing their own baseline game which is not as good. As a result, they may have good rallies but invariably they tend to come up short. If Nadal and Federer had to play a Richard Krajicek, Stich, Mecir, Leconte....whoever, even though they may win more matches against these guys than lose, because they are better, they will also suffer more losses than they will playing inferior baseliners.
.

Click to expand...

well. this is your hypothesis. whatever the reason, the result will still be that the agg.margins are higher.

bet said:

A SV player ranked 30 has a much better chance of taking advantage of a bad day from a top player than a baseliner ranked 30. The baseliner has to hit many balls and the top player has lots and lots of time to find his form and pull himself together. Especially when the baseliner plays the same topspin game every other baseliner does. He also knows he is a worse baseliner and fears that when the other player finds his form, it's all over. The SV'er may know he is not as good as the top player, but he also knows he has knockout punches, can potentially end it quickly and does not need to beat the top player at his own game.

Click to expand...

this does not contradict the theory that agg.margins are higher today, though, which is what i want to investigate.

World Beater said:

the tour may not be doing better in terms of agg.margins. but i want to compare the great players in each gen.

well i didnt argue that they are better. i said they are better movers. there is a difference. .

Click to expand...

Yes, and I said they aren't. This is one of the areas that has changed the least.

World Beater said:

this does not contradict the theory that agg.margins are higher today, though, which is what i want to investigate.

Click to expand...

Um...yes, that is why I hypothesized they might be higher today. However, what you should keep in mind is that, they won't provide you with any evidence of actual play level, regardless of whether they are higher, lower or the same. It is completely meaningless in that regard, there are dozens of confounding variables.

In fact, even comparing between matches today, it is not clear if aggressive margins say anything about the "quality" of the match, which is probably why the statistic never caught on.

bet said:

In fact, even comparing between matches today, it is not clear if aggressive margins say anything about the "quality" of the match, which is probably why the statistic never caught on.

Click to expand...

Interesting, and I've wondered why we don't see Aggressive Margins being talked about more. I do know that it's a complex stat to calculate -- not terribly complex, but certainly more trouble for a broadcaster to explain than simply throwing up winners and ue's, which are self-explanatory (though it turns out that even those stats have variable definitions that are rarely explained).

I think I know what you mean about AM's not necessarily reflecting the quality of a match, but couldn't that be said about any of the other stats we use like winners and ue's? AM's in the end just use those same stats. What do you mean specifically about the quality?

I calculated the Aggressive Margin's on some recent matches(using info at the official tournament websites):

2009 AO Final
Federer 19.88%
Nadal 19.59%

2009 AO SF
Federer 37.89%
Roddick 27.37%

2009 AO SF
Verdasco 23.63%
Nadal 23.89%

2009 AO 3R
Gasquet 21.46%
Gonzalez 21.46%

2008 USO Final
Federer 17.36%
Murray 4.31%

2008 USO SF
Murray 20%
Nadal 11.63%

2008 USO SF
Federer 24.09%
Djokovic 15.65%

2008 USO QF
Djokovic 24.55%
Roddick 17.41%

2008 W Final
Federer 30.25%
Nadal 31.47%

2005 AO SF
Federer 20.75%
Safin 18.99%

Thanks, Moose, for those calculations.

I've been trying to gather up all the evidence, one way or another, for double-faults being included in official counts of ue's. What I have so far suggests that df's have been included in recent years but perhaps not in previous decades.

Evidence that "Unforced Errors" include double-faults:

- The ATP site had this at the 2008 USO: "The 28-year-old Muscovite, who defeated Pete Sampras to lift the US Open trophy in 2000, won just five points more than Spadea in the match that featured 10 breaks of serve and a total of 85 unforced errors, including 16 double faults."

- During the 2008 Wimbledon final, NBC credited Federer with one “service unforced error,” presumably his one double-fault up to that point.

- For the 2005 USO final (Federer-Agassi), when I use the official UE count at USrg, I get the same Aggressive Margins that Alberto Brignacca calculated in an online article.

- In the 2003 USO semi between Henin and Capriati, USA broke down the total Unforced Errors into the categories of Service, Forehand, Backhand, Net. (They did the same with Winners).

- In the 1995 Wimbledon final, Sampras had 7 doubles but was credited with only 7 unforced errors. Mary Carillo later said his only unforced errors were double-faults.



Evidence that "Unforced Errors" do not include double-faults:

- Wimbledon.org does not seem to be including df's in its UE's for the 1999 final (see the link in the OP), or the fourth-round match between Sampras and Federer in 2001. Agence France Presse reported the numbers for that latter match as Wimbledon.org had them.

- During the 1987 U.S. Open final, CBS kept a running count of Wilander’s total unforced errors, and the count remain unchanged even a few games after he threw in a double-fault.


Evidence both ways:

- A Reuters report did not include df's in its UE's for the 2006 Wimbledon final, but Wimbledon.org and the NY Times did include them. Reuters had Federer leading Nadal 31-23 in "Unforced Errors." Wimbledon and the Times had 32-26. The difference is the df's (Federer 1, Nadal 3).

I know some posters are already looking at this; anything else you can think of, please post it here.

In the 1993 AO Final, Fred Stolle referred to a double fault as adding to Courier's unforced error total.

krosero said:

Thanks, Moose, for those calculations.

I've been trying to gather up all the evidence, one way or another, for double-faults being included in official counts of ue's. What I have so far suggests that df's have been included in recent years but perhaps not in previous decades.

Evidence that "Unforced Errors" include double-faults:

- The ATP site had this at the 2008 USO: "The 28-year-old Muscovite, who defeated Pete Sampras to lift the US Open trophy in 2000, won just five points more than Spadea in the match that featured 10 breaks of serve and a total of 85 unforced errors, including 16 double faults."

- During the 2008 Wimbledon final, NBC credited Federer with one “service unforced error,” presumably his one double-fault up to that point.

- For the 2005 USO final (Federer-Agassi), when I use the official UE count at USrg, I get the same Aggressive Margins that Alberto Brignacca calculated in an online article.

- In the 2003 USO semi between Henin and Capriati, USA broke down the total Unforced Errors into the categories of Service, Forehand, Backhand, Net. (They did the same with Winners).

- In the 1995 Wimbledon final, Sampras had 7 doubles but was credited with only 7 unforced errors. Mary Carillo later said his only unforced errors were double-faults.



Evidence that "Unforced Errors" do not include double-faults:

- Wimbledon.org does not seem to be including df's in its UE's for the 1999 final (see the link in the OP), or the fourth-round match between Sampras and Federer in 2001. Agence France Presse reported the numbers for that latter match as Wimbledon.org had them.

- During the 1987 U.S. Open final, CBS kept a running count of Wilander’s total unforced errors, and the count remain unchanged even a few games after he threw in a double-fault.


Evidence both ways:

- A Reuters report did not include df's in its UE's for the 2006 Wimbledon final, but Wimbledon.org and the NY Times did include them. Reuters had Federer leading Nadal 31-23 in "Unforced Errors." Wimbledon and the Times had 32-26. The difference is the df's (Federer 1, Nadal 3).

I know some posters are already looking at this; anything else you can think of, please post it here.

Click to expand...

I think it depends on who's scoring it and how they score it. For intance the 1990 US Open final Sampras was credited with 27 "placement winners", ut his final stats (even in an article posted yourself) he finshed with 39 winners. And article I read credited Pete with 13 aces and 12 service winners....
http://espn.go.com/sportscentury/features/00016454.html

Which would give him the aforementioned 39 winners (which the boxscore credited him with). If the aces were counted he'd have 52 total winners.

Same with 1997 Aus Open final. A boxscore you posted had Sampras with 38 winners. Another article I read credited him with "12 aces and 14 service winners". His winner breakdown from yet another article was serve 14, ground 13, volley 11.

But as you've noted there is evidence that suggests aces do get counted in winner counts at times.

380pistol said:

For intance the 1990 US Open final Sampras was credited with 27 "placement winners", ut his final stats (even in an article posted yourself) he finshed with 39 winners. And article I read credited Pete with 13 aces and 12 service winners....
http://espn.go.com/sportscentury/features/00016454.html

Which would give him the aforementioned 39 winners (which the boxscore credited him with). If the aces were counted he'd have 52 total winners.

Click to expand...

Actually I've never seen a boxscore or article that added up his total winners and reported them explicitly as 39 or 52. The explicit totals are 27 placement winners, 13 aces, 12 service winners. Those are consistent in every source I've seen. No source, that I know of, adds them all up into something like we see today, eg, "WINNERS (INCLUDING SERVICE)." Have you seen totals given as 39 or 52?

380pistol said:

Same with 1997 Aus Open final. A boxscore you posted had Sampras with 38 winners. Another article I read credited him with "12 aces and 14 service winners". His winner breakdown from yet another article was serve 14, ground 13, volley 11.

Click to expand...

The first article, with the aces and sw's, is from the Daily Telegraph (Sydney). Do you have a link to the article with the winner breakdown (or if not, a quote from it?)

Yes, the boxscore in The Age (Melbourne) gives Sampras 38 total winners, and Moya 13. The LA Times says how many were volleys:

Sampras obliged with 38 winners--11 on volleys--while the baseline-hugging Moya managed only 13 winners and no volleys.

Click to expand...

krosero said:

Actually I've never seen a boxscore or article that added up his total winners and reported them explicitly as 39 or 52. The explicit totals are 27 placement winners, 13 aces, 12 service winners. Those are consistent in every source I've seen. No source, that I know of, adds them all up into something like we see today, eg, "WINNERS (INCLUDING SERVICE)." Have you seen totals given as 39 or 52?

Click to expand...

"52" is just the final # he would have had if aces were included. Buy I recall 39 winners being his total count cuz it went along with 34 unforced errors. So I'm assuming the 39 is (27 placement and 12 service winners), excluding the aces.

krosero said:

The first article, with the aces and sw's, is from the Daily Telegraph (Sydney). Do you have a link to the article with the winner breakdown (or if not, a quote from it?)

Yes, the boxscore in The Age (Melbourne) gives Sampras 38 total winners, and Moya 13. The LA Times says how many were volleys:

Click to expand...

There is this link......
http://www.geocities.com/SouthBeach/Strand/2181/pete/article1997.html#mentor

...and try....
http://216.194.87.192/ozopen_1997.htm.

Not a complete breakdown but a start. I'' try and found the one where they mentioned groundstrokes. It was something about Sampras actually hitting to Moya's forehand to open up the court.

380pistol said:

"52" is just the final # he would have had if aces were included. Buy I recall 39 winners being his total count cuz it went along with 34 unforced errors. So I'm assuming the 39 is (27 placement and 12 service winners), excluding the aces.

Click to expand...

That would not be a bad assumption, if you could find a source stating explicitly that he had 39 winners -- but I don't believe I've ever come across a source that excluded aces while including service winners, in the Total Winners.

The convention is to take the aces, add them up to volleys and ground strokes, and throw in a few serves that are judged as service winners. The idea is to include every ball -- whether a volley, ground stroke or serve -- that either gets past the opponent cleanly (eg, aces) or wins the point outright because there is essentially "no play" on the ball (eg, service winners).

I call it a convention because it's a general rule with exceptions. I've seen matches where all the return errors were automatically called service winners and thrown into the Total Winners (along with the aces). Obviously there, you're giving the label of "winner" to quite a few balls on which there was still a play.

So that's one exception to the rule. I also found a match (Mac-Wilander at 85 USO), where it looks like all the forced errors were counted as winners.

But I haven't seen any set of numbers that I interpreted as service winners included, aces excluded. It's not that I think it can't happen; I've seen countless arrangements of stats that surprise me; I just need it documented. I'd like to find out if some stats have been counted that way, so if you can document it I'd be grateful.

380pistol said:

Not a complete breakdown but a start. I'' try and found the one where they mentioned groundstrokes. It was something about Sampras actually hitting to Moya's forehand to open up the court.

Click to expand...

Thanks -- it's the one with groundstrokes that may be the key. What I'm looking for is whether the source actually provided a breakdown the way I understand that term. I think of a breakdown as something explicit, eg, "He hit 38 winners, of which X were on serve and Y on other strokes." A number of the sources do that with the 38 winners, they tell you how many were volleys and how many were not. But the Telegraph, while it mentions 14 service winners, does not say that they were part of the total winners; in fact it doesn't even mention the total.

So I'm not expecting the 38 total winners to include all 14 service winners and to exclude the aces -- but I'm ready as usual to be surprised.

380pistol said:

For intance the 1990 US Open final Sampras was credited with 27 "placement winners", ut his final stats (even in an article posted yourself) he finshed with 39 winners. And article I read credited Pete with 13 aces and 12 service winners....
http://espn.go.com/sportscentury/features/00016454.html

Which would give him the aforementioned 39 winners (which the boxscore credited him with). If the aces were counted he'd have 52 total winners.

Click to expand...

As I said, I never saw a published total for that match, but if a total were done today you wouldn't get something as high as 52.

Back in the 80s there were a number of matches with inflated placement winners (CBS gave Mac 96 winners against Mats even though he hit only 53 clean ones). Sampras-Agassi is not nearly so bad but Pete was given 5 placement winners that were judgment calls, on top of his 22 clean winners. Today I think he would probably get just two extra, because there's a rule I use (basically following what Leo Levin told me) that seems to track well with official stats for winners today: if someone returns the ball into the stands, or he can't lift the ball and it bounces before it reaches the net, it's a winner for the other guy.

That method was really helpful when I counted the winners in the 2005 USO final and compared them against CBS's running counts.

Anyway, by that method I'd give Pete 4 service winners. He got 12 in the boxscore, but some old matches feature high numbers of service winners, usually as high as the aces (same thing seems to be going on with the Moya match, actually). Altogether I counted 17 serves from Pete that AA touched but did not put back in play; so in this match most of those were counted service winners. Today most would remain forced errors and only around 4, going by Levin's method, would be thrown into the Total Winners as service winners.

So I'd say he'd get 24 placement winners, 13 aces, 4 service winners -- a total of 41.

380pistol said:

But as you've noted there is evidence that suggests aces do get counted in winner counts at times.

Click to expand...

We have evidence from statisticians and the press, but I also found definitive proof in
Federer-Agassi. At one point CBS had Federer at 25 total winners. They moved him up to 28 after he’d hit two clean aces and a FH winner; in that time period he hit no other unreturned serves of any kind.

krosero

If you look but the USA Today stats(you posted with stats) from the 1990 US Open final, it says....

SAMPRAS
Service Winners - 12
Placement Winners - 27

... I naturally would assume they are differentiated or they just would have said "winners" or "total winners". That's why I got 39, but I will try and locate the link/article as well.

Another thing that doesn't make sense to me about statisticians and articles is how the prase. We have both read articles where Pete was credited as having "13 aces and 12 service winners" from that match. But to me that's like saying Krosero had "14 forehand winners and 9 winners".

You're count gave Sampras 22 "clean" winners. If we just add the aces (13) that gives Pete 35, yet we both have read where he was credited with 27.

If aces are included as/with service winners how could he have more, unless they were counted seperately. Another one was the 2005 US Open final that we talked about before. At one point during the match (3rd set I believe) CBS posted a count of service winners. Federer had betwen 20-23 and Agassi about 11-14. I couldn't really keep track off how they (CBS) were tabulating this cuz they kept showing Federer's backhand winners/unforced errors. I know Roger finished with 19 aces that day, and other than his backhand errors, CBS didn't give any further insight to "service winners" and how they were being accounted for. I wish you (or Moose) would do stats for that match so I can get further insight.

I wish the ATP would post these stats as well. usopen.org really didn't give me anything viable, when I asked.

380pistol said:

If you look but the USA Today stats(you posted with stats) from the 1990 US Open final, it says....

SAMPRAS
Service Winners - 12
Placement Winners - 27

... I naturally would assume they are differentiated or they just would have said "winners" or "total winners". That's why I got 39

Click to expand...

They are certainly differentiated, in the sense that they're not the same thing. Placement winners are groundies and volleys. Service winners are serves.

You got 39 by adding them together, but we have no evidence that the statistician(s) actually added them together to get the total winners. Until we can find a source that reports the total winners, we can't know for sure how any total was put together.

However I don't expect that the total was 39. To get that number you need to add his 27 ground stroke and volley winners to 12 serves that drew errors from AA -- and leave out his 13 clean aces. That's an arrangement that I'm telling you I've never seen before. Nor have I heard a statistician or reporter say that it's done that way.

The total winners of a match are made up of clean winners from volleys, ground strokes and serves (aces), plus any shots that the opponent touches with the racquet but that the statistician judges to be as good as a winner. Those judgment calls, when they're serves, are what I usually mean, along with Moose and, I think, most tennis fans, when we say "service winners". Not every source uses the term that way, but this boxscore in 1990 certainly did. The service winners are the judgment calls, not the clean aces. Pete had 13 clean aces; 12 additional serves were called service winners. Andre had no aces; but 9 of his serves were judged service winners.

So that's why I say that it would surprise me if they reported Pete's total winners as 39. I would expect them to add his 27 winners from volleys and ground strokes to his 13 aces, and that would already put them at 40. On top of that, if they added his 12 service winners, they'd get 52.

That would make more sense than 39 -- but I still see a problem. Levin told me that in the early 80s, winners started to include judgment calls. But he didn't say that every source all at once started to do it. It looks like a more gradual process, and you can't be sure that any single stat you find for "service winners" in those years was meant to be understood as part of the total winners. The service winners may just have been an independent stat that the source wanted to provide, like first serve percentage or unforced errors.

Sometimes in fact, sources disagree over the number of service winners in the same match. When Becker beat Lendl at the 86W, the NY Times had this:

The West German had 15 aces and 13 clear service winners. In addition, there were numerous points that he won when Lendl's return of serve sailed wide or sank into the net.

Click to expand...

The Atlanta Journal-Constitution agreed on the aces but not on the judgment calls:

Becker laid the strong hand of his service game on Lendl, 15 aces and 23 other service winners.

Click to expand...

When they played at W in '89, you can gather up their numbers from various sources, but they don't fit together; there's not enough room for both aces and service winners in the total winners.

In the St. Petersburg Times (Florida):

He finished with 17 aces and 100 winners, compared to nine and 86 for Lendl.

Click to expand...

In the St. Louis Dispatch:

Becker had 18 aces and 21 service winners, while Lendl had nine aces and 20 service winners.

Click to expand...

Well if you add Lendl's 9 aces to the 62 clean winners that I counted for him on volleys and groundies, you're up to 71. If you added his 20 service winners, you'd be up to 91, well over his reported total. And I didn't even count the likely judgment calls on volleys and ground strokes.

Same thing happens with Becker -- I got 64 clean winners apart from serves. Adding his aces and service winners would put him over his reported total of 100; and there would be still more judgment calls apart from service.

So when you get numbers for Sampras-Moya from different sources, there's no guarantee at all that they're meant to fit together -- no guarantee that the "service winners" reported in one source are part of the total winners reported in the other. Even the source reporting the service winners might not have included them as part of whatever total they drew up (if any).

Today, both methods and numbers seem more unified -- the same numbers seem to be reported on TV, the press, and tournament websites; and the numbers seem to be added up the same way (with minor exceptions).

And for Sampras' 12 service winners in the 1990 match, there's no indication, yet, that they would have been included in the total. If any total was drawn up, I'd expect the aces included instead. Aces are clean winners, and if "Winners" are going to be reported, I see no good reason why clean winners would be left out.

Yet the history of tennis stats has been so messy that I'm not ruling out your suggestion.

If you could find a source stating that Pete had 39 winners against AA, that would suggest what you're saying, that total winners can include some serves but not the aces.

(And Andre's own winners would be reported as 19, in that case).

380pistol said:

You're count gave Sampras 22 "clean" winners. If we just add the aces (13) that gives Pete 35, yet we both have read where he was credited with 27.

Click to expand...

Yes, I counted 22 clean winners from volleys and groundies (22 placement winners). He got 27 in the press because AA reached 5 of his shots but they were judged to be as good as winners.

What I was explaining above is that those judgment calls were overly generous, in those days, compared to what's done today. Today I'd give him only 24 placement winners, and 4 service winners (on top of his 13 aces).

380pistol said:

Another one was the 2005 US Open final that we talked about before. At one point during the match (3rd set I believe) CBS posted a count of service winners. Federer had betwen 20-23 and Agassi about 11-14. I couldn't really keep track off how they (CBS) were tabulating this cuz they kept showing Federer's backhand winners/unforced errors. I know Roger finished with 19 aces that day, and other than his backhand errors, CBS didn't give any further insight to "service winners" and how they were being accounted for. I wish you (or Moose) would do stats for that match so I can get further insight.

Click to expand...

Okay, I'll finish all my stats for that match and post them. I did figure out the service winners, though it had me baffled for a while. I said above that "service winners" usually mean the judgment calls, and that's true -- but the term is still problematic because it has been used in different ways, usually without explanation.

In that match, the "service winners" displayed in the third set were the clean aces plus the judgment calls. It's actually the first time I've seen the term used that way (I had already seen three different definitions of the term). I'll put in the details when I post the full stats.

I'm waiting for the stats on Federer/Agassi. But something still has me baffled.

In the 1990 US Open Agassi was credited with 10 "placement" winners (you had him with 9). He had zero aces but 9 service winners.

So if he finshed with 9 "service" winners, and was noted by USA Today (and other sources) has having 10 winners for the match, that would leave him with just winner apart from serve. How is that possible??

And why would USA Today have two seperate categories ("service" winners and "placement" winners)?? The way it's written gives the impression that they didn't include service winners in the total winner count.

380pistol said:

I'm waiting for the stats on Federer/Agassi. But something still has me baffled.

In the 1990 US Open Agassi was credited with 10 "placement" winners (you had him with 9). He had zero aces but 9 service winners.

So if he finshed with 9 "service" winners, and was noted by USA Today (and other sources) has having 10 winners for the match, that would leave him with just winner apart from serve. How is that possible??

And why would USA Today have two seperate categories ("service" winners and "placement" winners)?? The way it's written gives the impression that they didn't include service winners in the total winner count.

Click to expand...

There is no total winner count in the boxscore. Again, the "placement winners" are not the total winners; that's a term for ground strokes and volleys. Service winners are, of course, serves. So there's no common ground between them.

Yes, in some sources, Pete's 27 placement winners and Andre's 10 look like the total winners. But it's just imprecise writing. For example, ESPN:

Sampras hits 27 winners while Agassi has just 10.

Click to expand...

But the boxscore makes it clear that 27 and 10 are just the placement winners; nothing to do with any serves.

And other sources are more clear that 27 and 10 don't refer to serves. The Charlotte Observer:

Agassi, who lives on his groundstroke winners, fired just 10 past Sampras all match.

Click to expand...

The Philadelphia Inquirer:

In fact, it was Agassi who was supposed to be winging ground-stroke winners. But the final statistics showed that Sampras had 27 winners to Agassi's 10.

Click to expand...

One problem I have with these last sources is that they're still imprecise enough to suggest the possibility that Pete had 27 winners from ground strokes alone, not even including the 12 clean volley winners that I counted for him. But the boxscore is clear that the number 27 refers to all placement winners, that is, ground strokes and volleys.

In the end it just looks like all the articles are being imprecise to some degree. That's not a surprise when journalists are on deadline (and not all of them may fully understand the stats that are handed to them after the match). The only really precise source is the boxscore, because it was printed without interpretation or comment. It's the boxscore that makes everything clear -- particularly compared against my counts of the placement winners, which are close to the boxscore numbers.

I got 9 winners for Agassi because in my own counts I always stick to clean winners. Undoubtedly he got 10 in the boxscore because one more of his shots (a ground stroke or a volley) was not a clean winner; Pete must have barely got to it but it was judged as a winner. I can't tell you which one it was, but that's rarely possible; you just know that if the clean winners fall short of the official winners, then some judgment calls are inflating the total.

That's why Sampras was credited officially with 27 winners from volleys/groundies when in fact only 22 of those were clean winners.

Agassi's 9 service winners are hard to justify, because while he got 16 return errors from Sampras, on none of those did I judge that he had a service winner. That's what I mean about service winner judgments being overly generous in many old matches. Sampras got a return error from Agassi 17 times, and of those they judged 12 serves as being service winners; I only judged 4 that way.

krosero said:

There is no total winner count in the boxscore. Again, the "placement winners" are not the total winners; that's a term for ground strokes and volleys. Service winners are, of course, serves. So there's no common ground between them.

Yes, in some sources, Pete's 27 placement winners and Andre's 10 look like the total winners. But it's just imprecise writing. For example, ESPN:

But the boxscore makes it clear that 27 and 10 are just the placement winners; nothing to do with any serves.

And other sources are more clear that 27 and 10 don't refer to serves. The Charlotte Observer:

The Philadelphia Inquirer:



One problem I have with these last sources is that they're still imprecise enough to suggest the possibility that Pete had 27 winners from ground strokes alone, not even including the 12 clean volley winners that I counted for him. But the boxscore is clear that the number 27 refers to all placement winners, that is, ground strokes and volleys.

In the end it just looks like all the articles are being imprecise to some degree. That's not a surprise when journalists are on deadline (and not all of them may fully understand the stats that are handed to them after the match). The only really precise source is the boxscore, because it was printed without interpretation or comment. It's the boxscore that makes everything clear -- particularly compared against my counts of the placement winners, which are close to the boxscore numbers.

I got 9 winners for Agassi because in my own counts I always stick to clean winners. Undoubtedly he got 10 in the boxscore because one more of his shots (a ground stroke or a volley) was not a clean winner; Pete must have barely got to it but it was judged as a winner. I can't tell you which one it was, but that's rarely possible; you just know that if the clean winners fall short of the official winners, then some judgment calls are inflating the total.

That's why Sampras was credited officially with 27 winners from volleys/groundies when in fact only 22 of those were clean winners.

Agassi's 9 service winners are hard to justify, because while he got 16 return errors from Sampras, on none of those did I judge that he had a service winner. That's what I mean about service winner judgments being overly generous in many old matches. Sampras got a return error from Agassi 17 times, and of those they judged 12 serves as being service winners; I only judged 4 that way.

Click to expand...

See this is what I'm getting at. Agassi was credited with 10 "placement" winners (apart) from serve and 9 "service" winners. That to me, denotes 19 "total winners". The reason I come to this conclusion is 10 + 9. If a player has 8 FH winers and 6 BH winners, we take that as 14 groundstroke winners. If the "placement" winners (hypothetically) included service winners, that would leave Agassi wit one winner outside of the serve, which you counted 9 (clean), and was credited with 10.

So my assumption is that Agassi had 19 "total" winners (10 placement, 9 serve), and these exclude aces. Likewise if Sampras had had 13 aces and 12 service winners that would give him 25 winners from the serve alone. If he finished with 27 winners, that would mean he only had 2 winners outside of the serve, leading me to my assumption that the 27 "placement" winners and 12 "service" winners give Pete a total of 39, excludin aces.

I mean think... "13 aces and 12 service winners". If service winner include aces, that would equate negative one ace (all 12 service winners were aces, and one ace remaining). If they total 25 (winners from serve), then the 27 placement winners along with the 12 from the serve give him 39 (excluding aces).

If there was a stat.....
Forehand Winners - 14
Winners - 21

..... and you count 9 BH winners and 8 volley winners, you would assume the "21" doesn't include the forehand winners, or it wouldn't add up mathematically.

380pistol said:

See this is what I'm getting at. Agassi was credited with 10 "placement" winners (apart) from serve and 9 "service" winners. That to me, denotes 19 "total winners". The reason I come to this conclusion is 10 + 9. If a player has 8 FH winers and 6 BH winners, we take that as 14 groundstroke winners. If the "placement" winners (hypothetically) included service winners, that would leave Agassi wit one winner outside of the serve, which you counted 9 (clean), and was credited with 10.

Click to expand...

There's no question, of course, that placement winners are separate from service winners. You're deducing that with mathematical arguments, and that's fine. But it was already known: "placement winners" are known already to be separate from serves. It's a term that I've seen in boxscores as old as the 1920s, and it stands for ground strokes and volleys.

There's no issue with your claim that placement winners and service winners are distinct items; I'm doing that too. The issue is with the bolded term above: the Total Winners. It's your argument that the statisticians of the time regarded the Total Winners as excluding the aces, while including other serves. That's the issue -- and it's the only issue I have with your arguments. If you think there's something else, let me make it clear -- I'm only disagreeing with you about what the statisticians of the time regarded as the total winners.

There is no evidence that total Winners were ever added up the way you're saying; you need a source stating that Pete's winners were 39 (or that Andre's were 19). Mathematical arguments are no good here -- at least, not at this point. What we need at this point is some documentary evidence.

Secondly, and this is something you need to consider very seriously, there is not even any evidence that the statisticians for the match ever added up the placement winners with any serves in order to get what we would know today as the Total Winners. Such a term may not have been in use then -- at least not universally. If you asked them what the winners in the match were, they may just have responded with the placement winners -- that is, with the ground strokes and volleys. No serves of any kind.

That's exactly what ESPN did. They mention the aces and service winners in one paragraph. But in another paragraph, when they speak in general terms about how many "winners" Sampras and Agassi had, they just give the figures for the ground strokes and volleys: 27 winners for Pete, 10 for Andre.

Today it's not done that way: when articles mention the winners in the match -- just the terms "winners", without any detail or breakdown -- it means aces and service winners included.

(I can't recall any recent use of the term that was meant to be understood as only ground strokes and volleys, though I'm not ruling it out since that was the meaning of the term for many years, and some people might still use it that way).

I started watching tennis in the mid 80s, and for me a "Winner" was precisely that -- a ground stroke or a volley. If an article in SI or ESPN told me that 40 winners were hit in a match, I would have presumed 40 shots apart from service.

But I'm not using my memories as primary evidence. I've been looking through all the boxscores I have from that time period, from 1976 through the 90s. The earliest boxscore I can find with our modern term, "Winners (Including Service)", is actually the 1995 USO final between Sampras and Agassi. In that boxscore it's clear that serves are included. Maybe Moose knows of another. But in all the earlier boxscores I can find, they either say nothing about winners, or they give figures for "placements" or "placement winners." Sometimes they go into detail like FH or BH winners. But they're all sticking to shots apart from service. By using terms like "placements" or "forehands," they're telling the reader that service is not included. I haven't seen any boxscore yet, earlier than 1995, that says it includes serves.

It makes sense that once boxscores started to include aces in the Winner total, that they would take the trouble to specify "Winners (Including Service)." Otherwise the reader might read the aces on one line, the Winners on another, and think that they were entirely separate when they were not.

Before the inclusion of service, there would have been no confusion; aces can't be confused with Placements or Placement Winners.

So it looks likely to me that the source for the 1990 boxscore -- if they wanted to come up with a figure for total winners -- never added serves to the ground strokes and volleys.

That doesn't mean that no sources in 1990 were including serves. Moose and I have seen evidence of the various TV networks, at different times, starting to throw aces and service winners into their "Winner" counts. 1990 was a funny time, a kind of transitional time as far as tennis stats are concerned; and in this thread about Sampras/Agassi it makes some sense that we're talking so much about the 1990 match, because that boxscore stands out uniquely from the boxscores for their later meetings at the USO, W and AO.

But I want to be clear where the issue is. I don't have any arguments with your mathematical deductions about what categories like "placement winners" and "service winners" mean. (At least, I think we're in agreement). The argument is about whether the statisticians who produced the 1990 boxscore ever produced a figure for Total Winners in the way we understand that term (to include serves), and if they did, whether they did it the way you contend they did, including some serves but not the aces. That's a contention still without evidence. Total winners, from all the evidence I have, always include aces if the category of service is included.

The way to show that certain figures were added up into a total is not to add them up the way you think they were added up. You need to start with the total first, then work backward and see if you can make sense of it by working with the numbers that might have gone into it. If you had a source stating that Pete had 39 winners, then it's reasonable to induct that the categories of service winners (12) and placement winners (27) were added up to get 39. If you don't have that source, you're not analyzing a number, you're just throwing one out there (39) and stating, without evidence, that the statisticians of the time regarded it as the winner total.

In your last post you rightly call your winner total an "assumption". That's all it is, until you can find your number in an actual document.

A couple of corrections to points I made in my last post.

krosero said:

If you asked them what the winners in the match were, they may just have responded with the placement winners -- that is, with the ground strokes and volleys. No serves of any kind.

That's exactly what ESPN did.

Click to expand...

Well, ESPN did this, but we're talking about ESPN.com, which of course didn't exist in 1990. How and when they put together their current page on the 1990 match is anyone's guess. The only certain thing I can say about it is that in their article, the term "Winners" does not include service.

In any case, the articles that were actually published in 1990 do use the term "Winners" as excluding serves. They're not 100% clear about it, but we know it because their numbers show up in the boxscore as placement winners (that is, groundstrokes and volleys).

krosero said:

But in all the earlier boxscores I can find, they either say nothing about winners, or they give figures for "placements" or "placement winners." Sometimes they go into detail like FH or BH winners. But they're all sticking to shots apart from service. By using terms like "placements" or "forehands," they're telling the reader that service is not included. I haven't seen any boxscore yet, earlier than 1995, that says it includes serves.

Click to expand...

Doing a more thorough search of all my material, I still don't see a boxscore earlier than 1995 which says that service is included in the "Winners." But I do find a unique case, the 1991 USO final between Edberg and Courier -- unique because it's unclear whether service is being included. It does not say that its winners are placements, or forehands, or anything. It just uses the term "Winners" by itself.

Moose counted the winners here: http://tt.tennis-warehouse.com/showthread.php?t=199742. There is not enough room, in the boxscore's "Winners," to include both their aces and Moose's placement winners. The numbers are close, so it's not a certainty, but it looks like the "Winners" in this boxscore do not include the aces.

There is enough room to include the boxscore's one service winner, while excluding the aces. That's the arrangement 380Pistol is proposing: service winners, but not aces, in the total winners.

But I think to demonstrate such a particular arrangement, we'd need something more exact than merely having enough room for the service winners. And finding room for 1 solitary service winner is not hard to do.

(I'd prefer to see something exact, like 39 - 27 = 12).

All I think we can say specifically about the Edberg-Courier boxscore is that it (probably) does not include aces.

And if aces are not included as winners, I, for one, have a hard time seeing why other serves would be counted instead.

But more opinions are welcome here. I'd like to hear how people read the statistical terms used in tennis.

And I've seen other arrangements of stats that were far less logical (like the debacle with the Winners at the RG site).

krosero said:

A couple of corrections to points I made in my last post.

Well, ESPN did this, but we're talking about ESPN.com, which of course didn't exist in 1990. How and when they put together their current page on the 1990 match is anyone's guess. The only certain thing I can say about it is that in their article, the term "Winners" does not include service.

In any case, the articles that were actually published in 1990 do use the term "Winners" as excluding serves. They're not 100% clear about it, but we know it because their numbers show up in the boxscore as placement winners (that is, groundstrokes and volleys).

Doing a more thorough search of all my material, I still don't see a boxscore earlier than 1995 which says that service is included in the "Winners." But I do find a unique case, the 1991 USO final between Edberg and Courier -- unique because it's unclear whether service is being included. It does not say that its winners are placements, or forehands, or anything. It just uses the term "Winners" by itself.

Moose counted the winners here: http://tt.tennis-warehouse.com/showthread.php?t=199742. There is not enough room, in the boxscore's "Winners," to include both their aces and Moose's placement winners. The numbers are close, so it's not a certainty, but it looks like the "Winners" in this boxscore do not include the aces.

There is enough room to include the boxscore's one service winner, while excluding the aces. That's the arrangement 380Pistol is proposing: service winners, but not aces, in the total winners.

But I think to demonstrate such a particular arrangement, we'd need something more exact than merely having enough room for the service winners. And finding room for 1 solitary service winner is not hard to do.

(I'd prefer to see something exact, like 39 - 27 = 12).

All I think we can say specifically about the Edberg-Courier boxscore is that it (probably) does not include aces.

And if aces are not included as winners, I, for one, have a hard time seeing why other serves would be counted instead.

But more opinions are welcome here. I'd like to hear how people read the statistical terms used in tennis.

And I've seen other arrangements of stats that were far less logical (like the debacle with the Winners at the RG site).

Click to expand...

I see where you're coming from. You make good points about how stats are counted at different periods in time, and tahat may throw me off.

I persnally believe that there are some instances where aces are counted as winners, and others they aren't. Figuring who;s counting what as which, I have no clue.

When I'm at a loss, I go with what I know. I know mathematics is infinite. 2 +2 = 4, it was, it is and alway will be. That's where some of the stats from you have and they differ from what some publications say. If you have X # of points won, I will add up what I can and see what I get.

With Sampras 1990 US Open (it's one we've been using), you have 99 pts won by Pete. The breakdown would give me.....
13 - aces
12 - service winners
27 - placement winners
28 - unforced errors by Dre

That gives (or leaves) me 18 forced errors by Dre which seems feasable.

But when I get 27 winners (from publications), you had 22-25 on your own (strokes and volleys), that leaves me to believe those 27 did not include anything from aces.

Now we have a problem with "English" how it's written in boxscores and publications. Like I said before when I read a boxscore.....

Service winners - 12
Placement winners - 27

.... that leads me to believe the player had 39 "total" winners (excl. aces). As I've read other boxscores wherer it will say "Winners" or "Winners (incl. service)".

Or one we spoke about a journalist wrote "12 aces and 14 service winners". The use f the the word "and" leads me to believe they are separate and not a total of 14 servive winner (which includes the 12 aces) as proper grammer would be......

"he had 12 aces included in his 14 service winners...."
Of his 14 service winners, 12 were aces....

Cuz grammatically, it leads to believe they are different, meaning 26 points fromm the serve 12 via aces and 14 via service winners.

What gives crdence to that is back to the 1990 US Open final where Sampras was credited with "13 aces and 12 service winners". If aces (in this match) were counted as service winners, that statement becomes impossible. If the 12 service winners where clean aces, what would be the explanation for the 13th???

Anyway it comes back to who scored the match and how they did. It one of your counts, you said, what NBC (I believe) counted as service winners, were waht you had counted as return errors after subtracting clean winners, aces, doubles and forced/unforced errors(from rallies), from the # of total points. Then the time frame and how the scoring has changed.

As you posted earlier there's evidence to supprt that aces are counted as winners, and others where they aren't. But can you figure out who's doing what. I tried it for one set (1994 Sampras/Chang Wimbledon) counting service winners (or what I deemed them to be. I had Pete at 16 winners, 7 unforced in the 1st set, while NBC had him at 19 winners and 9 unforced.

380pistol said:

With Sampras 1990 US Open (it's one we've been using), you have 99 pts won by Pete. The breakdown would give me.....
13 - aces
12 - service winners
27 - placement winners
28 - unforced errors by Dre

That gives (or leaves) me 18 forced errors by Dre which seems feasable.

Click to expand...

Actually that leaves 19 forced errors.

And the way I would prefer to break down the 99 points is with my own counts, since I don't know exactly how they defined their numbers. For instance, do their unforced errors included the double faults? And their service winners are judgment calls, so how many of those would I regard as forced errors?

If I'm counting only clean winners, I give Pete 13 aces and 22 placement winners. That leaves 64 points that he won on errors by AA, either forced or unforced (Andre officially made 28 or 29 unforced, depending on whether they included his df). In this arrangement, I'm applying the term "error" to any ball that Andre's racquet touched, no matter how barely he scraped it.

If I include judgment calls, then I'd give Pete 13 aces, 4 service winners, 24 placement winners. That leaves 58 points won on errors, either forced or unforced.

380pistol said:

Now we have a problem with "English" how it's written in boxscores and publications. Like I said before when I read a boxscore.....

Or one we spoke about a journalist wrote "12 aces and 14 service winners". The use f the the word "and" leads me to believe they are separate and not a total of 14 servive winner (which includes the 12 aces) as proper grammer would be......

"he had 12 aces included in his 14 service winners...."
Of his 14 service winners, 12 were aces....

Cuz grammatically, it leads to believe they are different, meaning 26 points fromm the serve 12 via aces and 14 via service winners.

What gives crdence to that is back to the 1990 US Open final where Sampras was credited with "13 aces and 12 service winners". If aces (in this match) were counted as service winners, that statement becomes impossible. If the 12 service winners where clean aces, what would be the explanation for the 13th???

Click to expand...

Yes, the grammar is a problem. Often, unfortunately, there's no way to know what a given stat means until you count everything yourself.

You're saying that in proper grammar, 12 aces "and" 14 service winners means that the 12 and the 14 are entirely separate. That's true for the sources we've been looking at, but have a look at these two:

Bloomberg.com: “Williams avenged that defeat today with a display of powerful forehands and serves, walloping 28 winners and seven aces.”

ABC.net: “dominating with an impressive 28 winners and seven aces.”

When I read that, I didn't know whether Serena's total winners, including aces, were really 28 or 35. (At the time I couldn't find the boxscore to make it clear). When I counted the winners myself I saw that she had 20 clean winners apart from service, and 7 aces. That puts her at 27. Her 28th winner was presumably a judgment call on a serve (a service winner), volley or groundstroke.

That's just the way tennis is now: unclear grammar is the rule. The journalists write as if they didn't know that the total winners included the aces, because you'd expect them to say, "she hit 28 winners, including 7 aces." It's strange because the AO site makes it clear that her total is "Including Service." But that's how winners are referred to now, with the journalists simply repeating the numbers they're given -- the total winners and the aces -- without bothering to clarify that the total winners include the aces. Maybe they're not sure themselves; the fans often aren't.

I would not rely on grammar from articles in the popular media to clarify for you what the stats are. Most of that material is written on deadline and sometimes dumbed down (stripped of detail), often by writers with limited knowledge of tennis. Some are clear and some just aren't. It's far more reliable to do your own count and compare.

380pistol said:

I tried it for one set (1994 Sampras/Chang Wimbledon) counting service winners (or what I deemed them to be. I had Pete at 16 winners, 7 unforced in the 1st set, while NBC had him at 19 winners and 9 unforced.

Click to expand...

Your numbers came close, which is all you can expect when judgment calls are involved.

But I don't know if you're including aces. If you're not, what numbers would you get, putting the aces back in? And what's the breakdown?


By the way, late in the 1990 USO final, Tony Trabert said that Andre had only 7 winners in the match. By my count he'd hit 6 clean groundstroke winners. This tells us a couple of things.

1) Either Trabert could use the term "winners" to mean just groundstrokes and volleys, without any serves, or

2) Trabert was referring to CBS's running count of all winners, including service. In that case, they had given Andre one service winner at most, and clearly they were not using the boxscore stats in USA Today, where Andre ended up with 9 service winners. That isn't possible, because after Trabert's comment Andre drew only two more return errors from Sampras.

So all of this shows that when we see all those "service winners" mentioned in old articles and boxscores, we can't assume that they're meant to be part of Total Winner counts. Merely being designated as "service winners" does not mean that anyone regarded them as equal to clean winners and aces.

krosero said:

Yes, the grammar is a problem. Often, unfortunately, there's no way to know what a given stat means until you count everything yourself.

You're saying that in proper grammar, 12 aces "and" 14 service winners means that the 12 and the 14 are entirely separate. That's true for the sources we've been looking at, but have a look at these two:

Bloomberg.com: “Williams avenged that defeat today with a display of powerful forehands and serves, walloping 28 winners and seven aces.”

ABC.net: “dominating with an impressive 28 winners and seven aces.”

When I read that, I didn't know whether Serena's total winners, including aces, were really 28 or 35. (At the time I couldn't find the boxscore to make it clear). When I counted the winners myself I saw that she had 20 clean winners apart from service, and 7 aces. That puts her at 27. Her 28th winner was presumably a judgment call on a serve (a service winner), volley or groundstroke.

That's just the way tennis is now: unclear grammar is the rule. The journalists write as if they didn't know that the total winners included the aces, because you'd expect them to say, "she hit 28 winners, including 7 aces." It's strange because the AO site makes it clear that her total is "Including Service." But that's how winners are referred to now, with the journalists simply repeating the numbers they're given -- the total winners and the aces -- without bothering to clarify that the total winners include the aces. Maybe they're not sure themselves; the fans often aren't.

I would not rely on grammar from articles in the popular media to clarify for you what the stats are. Most of that material is written on deadline and sometimes dumbed down (stripped of detail), often by writers with limited knowledge of tennis. Some are clear and some just aren't. It's far more reliable to do your own count and compare.

Click to expand...

I agree. When I see.....

"28 winners and 7 aces" the "and" makes me think the total is 35. If it didn't like you said it would (or should) be grammatically written as "28 winners including 7 aces".

But then again you hit the nail on the head when you say these journalists/writers are not the most knowledgeable where tennis is concerned, are probably are trying beat deadlines, and are going off facts given to them and using that without analyzing them.

I think that's a problem I have incurred and you seem to yourself. I assumed the Aus Open 2009 stats (from website) included aces cuz in their boxscores their statline said.....
"Winners (Incl. service)".

But even so does that always mean aces or just service winners., as we look back to the 1990 US Open F boxscore.....
Placement winners - 27
Service winners - 12

I think it would be cleare if the was a certain distinguishable way we know how stats are being tabulated, or they could make more clearer to how each tournament, scorer etc. is doing it. Personally I would like to get my hand on as many boxscores as possible so I can do some king of comparison.

krosero said:

Your numbers came close, which is all you can expect when judgment calls are involved.

But I don't know if you're including aces. If you're not, what numbers would you get, putting the aces back in? And what's the breakdown?


By the way, late in the 1990 USO final, Tony Trabert said that Andre had only 7 winners in the match. By my count he'd hit 6 clean groundstroke winners. This tells us a couple of things.

1) Either Trabert could use the term "winners" to mean just groundstrokes and volleys, without any serves, or

2) Trabert was referring to CBS's running count of all winners, including service. In that case, they had given Andre one service winner at most, and clearly they were not using the boxscore stats in USA Today, where Andre ended up with 9 service winners. That isn't possible, because after Trabert's comment Andre drew only two more return errors from Sampras.

So all of this shows that when we see all those "service winners" mentioned in old articles and boxscores, we can't assume that they're meant to be part of Total Winner counts. Merely being designated as "service winners" does not mean that anyone regarded them as equal to clean winners and aces.

Click to expand...

My count of that on set did not include aces and doubles. But Pete had 5 aces and 1 double in the opening set, which would bring my tally to 21 winners and 8 unforced. I only counted what you call "clean" winners (from grounies and vollies), and with discretion regarding service. I don't recall giving Pete a service winner for a 2nd serve, and the ones for 1st serves, were if Chang was strecthing, reaching, lunging,a little late (couldn't keep up) or could not make clean or firm contact with the serve.

Keep the the good work. You're insight and information is helpful.

380pistol said:

I assumed the Aus Open 2009 stats (from website) included aces cuz in their boxscores their statline said.....
"Winners (Incl. service)".

But even so does that always mean aces or just service winners., as we look back to the 1990 US Open F boxscore.....
Placement winners - 27
Service winners - 12

Click to expand...

Yes, winner totals that include service can be expected to include aces and not just service winners. (You might find this thread interesting: http://tt.tennis-warehouse.com/showthread.php?t=176760). It would be very interesting if you could show me a winner total that seemed to include service winners while excluding aces, but the 1990 USO final just doesn't have that. We've beaten this nearly to death -- it's getting boring -- so at this point I'll only say this, which is really what it boils down to: we need a winner total that includes service, if we're even going to ask which serves were included. The 1990 boxscore does not provide such a total. Until you can show that there ever was such a total -- by citing it -- there's really not much left to say.

380pistol said:

I think it would be cleare if the was a certain distinguishable way we know how stats are being tabulated, or they could make more clearer to how each tournament, scorer etc. is doing it.

Click to expand...

I wish there was one clear way of figuring out tennis stats. There are some general rules about what the terms mean, but I'm amazed at how many different ways some terms can be defined. At the Olympics last year, the boxscores listed "Winners" that were comprised of only ground strokes (http://results.beijing2008.cn/WRM/ENG/INF/TE/C73A/TEM001101.shtml#TEM001101). Last year at the Roland Garros site, "Winners (Including Service)" meant the total number of points won on serve.

That's how weird tennis stats are.

And these are recent examples, so it's not as if the only problems were in the past and everyone is now coming to accept one standard way. I wish that there were universal agreement, so that, you know, we could read a number and actually have some confidence that we know what it's supposed to count.

Really if you want to analyze stats the best way is to do your own counts. Then find the published stats, analyze them, debate them, make whatever inquiries you can -- and then make your best judgments. It's a lot of different methods, in other words, not just one.

380pistol said:

Personally I would like to get my hand on as many boxscores as possible so I can do some king of comparison.

Click to expand...

There are a lot here: http://tt.tennis-warehouse.com/showthread.php?t=207411.

380pistol said:

My count of that on set did not include aces and doubles. But Pete had 5 aces and 1 double in the opening set, which would bring my tally to 21 winners and 8 unforced. I only counted what you call "clean" winners (from grounies and vollies), and with discretion regarding service. I don't recall giving Pete a service winner for a 2nd serve, and the ones for 1st serves, were if Chang was strecthing, reaching, lunging,a little late (couldn't keep up) or could not make clean or firm contact with the serve.

Click to expand...

At 21 winners you're starting to go over NBC's count -- and you might go up even further if you included judgment calls on groundstrokes and volleys. By 1994 judgment calls had already been in use for some years.

True, I only count clean winners for my own stats but I want to be clear that Moose and I do that so we have a unified, objective method by which to get winner totals for many matches. For comparing against counts in the media, you need to keep some track of the judgment calls they might have made.

So if you do your own counts again, see what you get if you count clean winners and aces, along with judgment calls on serves, volleys and groundstrokes.

The rule I mentioned above was, if a player knocks a ball off into the stands, or he hits it but it bounces on his side of the net, then I credit the other guy with a winner.

That rule is basically what Levin described to me and it often leaves my winner count somewhere close to the available count (esp. in Federer-Agassi). The only caveat is that, you still need to be flexible with it and use common sense. Federer has a tendency to shank balls into the seats, and often it's his own unforced error, not a winner for the other guy. There were a few of those against Agassi that CBS did not seem to throw into the running counts.

Let me see if I can finish those stats.

Stats for 2005 USO final (Federer-Agassi)

Federer d. Agassi 6-3, 2-6, 7-6 (1), 6-1

The USrg boxscore:



I got the same figures for total points won, and so did the ATP. So I'm just going to talk about the winners.


By my count:

Federer had 19 aces, Agassi 6.

RF by set: 4, 4, 5, 6
AA by set: 4, 0, 1, 1

Federer got 25 return errors from Agassi, of which I judged 9 as service winners.

Agassi got 20 return errors from Federer, of which I judged 3 as service winners.


Federer made 36 clean winners apart from serves: 25 FH, 9 BH, 1 FHV, 1 BHV.

Agassi made 22 clean winners apart from serves: 13 FH, 6 BH, 2 FHV, 1 BHV.

RF by set: 8, 8, 11, 9
AA by set: 4, 9, 6, 3


CBS/USrg stats:

This match had constant updates on the winners. The network provided so many winner totals and breakdowns that with a little effort it was possible to see the judgment calls they were making, wherever their numbers exceeded my count of the clean winners/aces.

CBS gave running counts of Federer’s FH, BH and total winners. As far as I can tell, it seems that apart from clean winners and aces they gave him: 6 service winners and 2 FH’s in the first set; 2 service winners in the third set before displaying their last overall winner count; and 4 more judged winners through the end of the match, which I put down as 1 FH, 2 BH’s, and a service winner. So I think that CBS gave Federer 41 placement winners, 19 aces and 9 (additional) service winners, producing the 69 total winners seen in the boxscore.

CBS gave running counts for Agassi’s FH and total winners (no BH’s). From those it seems that apart from clean winners and aces they gave him: 1 service winner in the first set and two more in the second; one FH (his 9th by their count) and probably a BH in the third; and probably a FH winner in the opening game of the fourth. So I think CBS gave Agassi 25 placement winners, 6 aces and 3 (additional) service winners, producing the 34 total winners from the boxscore.

My own rule for making judgment calls was to credit a winner if the ball was knocked into the stands or it bounced, before reaching the net, on the same side of the court as the player trying to return it.

With that method I kept pace with all of CBS's winner counts for Federer, in the individual strokes and the overall total, through the end of the match.

With Agassi I exceeded CBS's count at times, and later I saw the reason. Using my method too rigidly, I gave Agassi a few winners when Federer shanked a ball into the seats, but these were not strictly winners. In one case Agassi's shot skidded off a line -- maybe you could call it an error forced on Fed -- and in another Roger just made an unforced error. Which just means you have to use common sense before crediting a winner.

When I started my count I had not intended to go into such detail, but it was kind of cool that my own counts and judgments produced the same total of 69 winners that Federer got officially.

And the detail helped me with a strange CBS stat given during the tiebreak.

They put Federer at 21 "service winners", Agassi at 8.

At that point I had Federer at 13 aces and 19 other unreturned serves (the return errors), including 8 that I judged myself as service winners.

I had Agassi at 5 aces and 17 other unreturned serves, including 3 that I judged myself as service winners.

So the CBS numbers couldn't refer to the return errors alone; Federer had not even gotten 21 return errors from Agassi yet.

But add up the aces and the serves I judged as service winners, and you get the CBS numbers in that “service winners” graphic.

So the CBS graphic for "service winners" was essentially meant to display the number of serves that either got past the receiver cleanly (aces) or were judged to be as good as winners (probably in the sense that the receiver has not made an unforced error, or even a forced error, but has simply been beaten with a winner).

Not that you would know what CBS meant, without doing your own counts. I have to wonder what is the point of putting up a graphic that no one will understand. Forehands and backhands, fine. But "service winner" is just an extremely problematic term with a lot of confusion around it.

Definitions of service winners

This deserves a separate post.

These are the definitions of "service winner" that I've come across in official stats:

1) Serves that the receiver reaches with a racquet, but that are judged to be as good as winners (sometimes called "unreturnable"). This is an extremely common definition, and today you can expect these winners to be included in Total Winner counts.

2) All serves that the receiver gets a racquet on but does not put back in play (what I like to call the return errors). Example: 1987 Wimbledon final, Cash-Lendl, New York Times boxscore.

3) All unreturned serves, including aces. Example: 1998 Wimbledon final, Sampras-Ivanisevic, Sports Illustrated boxscore.

4) All the serve in category #1, combined with the aces. Example: 2005 USO final, Federer-Agassi, CBS.

Among fans I've also seen "service winners" taken to mean just the aces. I haven't seen that definition in a boxscore but it could be out there.

On top of this, the judgment calls made in category #1 vary widely. Old boxscores from the 80s and 90s sometimes seem to judge most of the return errors as service winners. In recent years (Fed-Agassi, for example), I don't see judgment calls made so liberally; but I've come to expect to see anything.

This is really where it's too bad that tennis scorers are not making more of an effort to explain what they do.

I didn't find any extra stats not already in the boxscore, but I did find one source that shows we shouldn't dog all writers for being unclear (as I was doing when writers loosely mention total winners and aces). Some can write precisely.

The Atlanta Journal-Constitution: “that translated into Federer churning out 69 winners to Agassi's 34, including 19 aces compared to Agassi's six.”