

Thread Tools  Search this Thread  Display Modes 
01042013, 07:09 PM  #41  
Legend
Join Date: Dec 2006
Posts: 5,513

Problem with break points
 For FedererAgassi, 2003 Masters Cup final, each player has been incorrectly given his opponent's break points. Federer dominated the match but the ATP boxscore has Agassi facing no break points, and they've got Andre breaking serve twice. The numbers have been reversed.
All the other stats for that match are in the correct column, though. Quote:


05272013, 07:41 PM  #42 
Legend
Join Date: Dec 2006
Posts: 5,513

Correct ATP stats
Here are some more matches from the 1990s for which the ATP did not count aces and doublefaults twice.
 For SamprasAgassi, 1991 ATP Championships, roundrobin, Slice Serve Ace has Sampras leading 8377 in total points won, exactly in line with the ATP’s numbers.  For AgassiSampras, 1992 Atlanta final, Slice Serve Ace has Sampras serving on 66 points and Agassi on 60, right in line with the ATP stats.  For StichSampras, 1993 ATP Championships final, Slice Serve Ace has Stich serving 133 points, Sampras 141 points – right in line with the ATP numbers.  For SamprasBecker, 1994 ATP Championships final, Slice Serve Ace has Sampras serving 112 points, Becker 132 points – exactly in line with the ATP’s numbers.  For BeckerChang, 1995 ATP Championships final, the ATP has Becker serving on 96 points. That must be correct, because subtracting his aces and doublefaults once would leave him serving 66 points in the match and only 55 points in his 15 service games: an impossible average of 3.7 points per game.  For SamprasAgassi, 1996 ATP Championships, roundrobin, Slice Serve Ace has Sampras serving 38 points, in agreement with the ATP.  For SamprasIvanisevic, 1996 ATP Championships semifinal, Slice Serve Ace has Sampras serving 101 points, Ivanisevic 109, in agreement with the ATP.  For SamprasAgassi, 1998 Monte Carlo, R32, Slice Serve Ace has Sampras serving 63 points, in agreement with the ATP.  For SamprasAgassi, 1999 Cincinnati semifinal, Slice Serve Ace has Sampras serving on 59 points, Agassi on 71 – same as the ATP. 
05272013, 07:42 PM  #43 
Legend
Join Date: Dec 2006
Posts: 5,513

Incorrect ATP stats
Here are more matches for which the ATP has counted aces and doublefaults twice.
 For IvanisevicSampras, 1992 Wimbledon semifinal, the New York Times reported that Sampras never won more than two points in any of Ivanisevic’s 21 service games. So Ivanisevic cannot have served more than 138 points (counting 12 points he served in the tiebreaks)  the ATP has him serving 159. If Ivanisevic’s aces and doublefaults are subtracted once, then he served 118 points.  For SamprasBecker, 1992 ATP Championships, roundrobin, Nadal_Power and Slice Serve Ace both have Sampras serving 70 points and Becker 75. The ATP would have those same numbers if it subtracted aces and doublefaults once.  For BeckerIvanisevic, 1992 ATP Championships semifinal, Slice Serve Ace has Ivanisevic serving 101 points, exactly what the ATP would have if aces and doublefaults were subtracted once.  For IvanisevicBecker, 1994 Wimbledon semifinal, Slice Serve Ace has Ivanisevic serving 89 points, Becker 93. That’s exactly what the ATP would have if aces and doublefaults are subtracted once from their stats.  For SamprasPhilippoussis, 1995 USO, R32, Slice Serve has Philippoussis serving 147 points, which is what the ATP would have if it subtracted his aces and doublefaults once.  For KrajicekStich, 1996 Wimbledon R16, Moose has Krajicek leading in total points by 10597. The ATP has 10696 if aces and df’s are subtracted once.  For SamprasPhilippoussis, 1996 USO, R16, Slice Serve has Philippoussis serving 82 points, which is what the ATP would have if it subtracted his aces and doublefaults once.  For SamprasIvanisevic, 1996 USO semifinal, Slice Serve Ace has Sampras serving 134 points, Ivanisevic 116. That’s exactly what the ATP has if the aces and doublefaults are subtracted once from their numbers.  For SamprasPioline, 1997 Wimbledon final, Slice Serve Ace has Sampras serving 76 points, Pioline 78 points. The ATP would have the same numbers if it subtracted aces and doublefaults once.  For HaasSpadea, 1999 Australian Open quarterfinal, the ATP’s figures for success on first serve, for both players, are slightly higher than the figures from Australia’s Seven network, which suggests that the ATP counted aces (and doublefaults) twice. 
05272013, 07:43 PM  #44 
Legend
Join Date: Dec 2006
Posts: 5,513

Games or sets missing entirely
Here are more matches in which the ATP appears not to have counted entire games or sets.
 For IvanisevicBailey, 1993 Wimbledon, R16, each man served 31 times; the score was 57, 76 (73), 67 (37), 64, 97. But the ATP had only 46 service games played, as if the 16 games of the fifth set were dropped entirely. All the other stats may be undercounted as well (they’ve got Goran serving 22 aces, 10 doublefaults).  For HenmanCourier, 1999 Wimbledon, R16, the scoreline of 46, 75, 75, 67, 97 shows 62 service games altogether, plus one tiebreak. But the ATP – despite having the correct score – reports only 46 service games, showing that they have dropped the 16 games of the final set. 
05272013, 07:44 PM  #45 
Legend
Join Date: Dec 2006
Posts: 5,513

Data missing for entire rounds
The ATP has no data for the first and third rounds at Roland Garros in 1991 (click on the scores to get the stats):
http://www.atpworldtour.com/Share/Ev...x?e=520&y=1991. 
05282013, 08:04 AM  #46  
Rookie
Join Date: Dec 2009
Posts: 185

Quote:
hadn't notice that at all. so this doublecounting is not reduced to grand slam matches only. although, both these matches are from frankfurt 1992, maybe it's the only nongs tournament with wrong stats. i've been trying to calculate 2nd serve pts won for the 90s by substracting dfs from gs matches. this doesn't cover all faults from atp stats, but it'll get us lists which are much closer to reality. 

slice serve ace 
View Public Profile 
Find More Posts by slice serve ace 
05282013, 07:36 PM  #47  
Legend
Join Date: Dec 2006
Posts: 5,513

Quote:
Subtracting doublefaults from GS matches should definitely give us a more realistic picture (even if not necessarily an exact picture) and I'd be interested to see what you come up with for Sampras and other big players of the 90s. I think that the GS events in 1991, as well as the USO in 1999, did not have doublecounting. 

05282013, 07:38 PM  #48 
Legend
Join Date: Dec 2006
Posts: 5,513

Data missing in other Grand Slam rounds
The ATP has no data for the R32 at Wimbledon in 1996, except for one match (KrajicekSteven): http://www.atpworldtour.com/Share/Ev...x?e=540&y=1996.
They have no data for the R32 at the AO in 1997: http://www.atpworldtour.com/Share/Ev...x?e=580&y=1997. They also have no data for the R32 at the 1994 Paris Masters: http://www.atpworldtour.com/Share/Ev...x?e=352&y=1994. (Data is otherwise available for that tournament in 199199). 
07102013, 06:14 PM  #49 
Legend
Join Date: Dec 2006
Posts: 5,513

I've noticed that the KrajicekSteven data is wildly off: the ATP has each man winning 73% of the points in the match.
See http://www.atpworldtour.com/Players/...=K214&oId=S321. The points that each man won on his own 1st and 2nd serves are merely repeated in his opponent's column as receiving points won. Wimbledon.org has correct stats: Krajicek led in total points by 152134 (a total of 286 points played compared to 230 reported by the ATP). Krajicek served 151 points, Steven 135. See http://web.archive.org/web/199701301...RSTAT0099.html 
07102013, 06:51 PM  #50 
Legend
Join Date: Dec 2006
Posts: 5,513

Slice Serve Ace's thread correcting for the doublecounting: http://tt.tenniswarehouse.com/showthread.php?t=469808

08152013, 01:17 PM  #51 
Legend
Join Date: Dec 2006
Posts: 5,513

Here's another case where the ATP has dropped entire sets from a match.
 For KuertenSafin, 2000 Indianapolis final, the scoreline of 36, 76 (2), 76 (2) requires 33 service games, not including the tiebreaks. But the ATP boxscore reports only 9 service games, presumably those of the first set. And a case of games being slightly overcounted:  For HenmanKuerten, 2000 Cincinnati semifinal, the scoreline of 67(11), 63, 76 (2) requires 33 service games, not including the tiebreaks. But the ATP boxscore reports 34 service games. http://www.atpworldtour.com/Tennis/P...=2000&m=s&e=0# 
10302013, 02:08 PM  #52 
Legend
Join Date: Dec 2006
Posts: 5,513

Stats missing entirely for one player in a match
For BrugueraChang at 1997 French Open, R16, Chang’s numbers are all zero: http://www.atpworldtour.com/Tennis/P...997&m=s&e=520#

10302013, 03:58 PM  #53 
Legend
Join Date: Dec 2006
Posts: 5,513

Stats wrong in all seven rounds at 2003 French Open
The problem here is one we've seen already, in which two players combine to win more than 100% of the points in the match.
We saw this already with KrajicekSteven at 1996 Wimbledon. However this problem is appearing in all seven rounds at the 2003 French Open. I have not checked whether it appears in every single match, but each round does have matches with this problem. Just two examples. For FerreroCosta at 2003 French Open semifinal, each player is winning 64% of Ferrero’s service points, and each player is winning 56% of Costa’s service points: http://www.atpworldtour.com/Players/...=F316&oId=C378. Each player ends up winning 60% of the points in the match. Matchstat.com has the correct stats: http://matchstat.com/index.php?Contr...&Id_Player2=40. All of Ferrero’s activity at the 2003 French: http://www.atpworldtour.com/Tennis/P...003&m=s&e=520#. For SargsianRoddick at 2003 French Open, R128, each player is winning 57% of Roddick’s service points, and each player is winning 73% of Sargsian’s service points: http://www.atpworldtour.com/Players/...=S545&oId=R485. Each player ends up winning 64% of the points in the match. Entire 2003 RG draw: http://www.atpworldtour.com/Share/Ev...x?e=520&y=2003. (Click on the scores for the match stats.) Last edited by krosero; 10302013 at 04:03 PM. 
10302013, 05:43 PM  #54 
Professional
Join Date: Sep 2013
Posts: 872

The match that most interests me is Sampras v Ivanisevic at Wimbledon 1995, as I had long believed (based on ATP tats) that it was a rare case of the winner being massively out scored in points.
If I recall correctly, having not seen the match in 18 years, Ivanisevic was 200 on his serve in BOTH sets 2 and 4, which made his winning so many more points than Sampras believable. OTOH, Sampras led the game count 2724, which does indeed make a 20point gap in Ivanisevic's favor seem hard to believe. Either way, it was quite staggering idiocy on goran's part to serve five straight love games and then drop serve in the next game and to do so twice in the match. 
helterskelter 
View Public Profile 
Find More Posts by helterskelter 
11142013, 11:06 AM  #55 
Legend
Join Date: Dec 2006
Posts: 5,513

Doublecounting of aces and df's
 For SamprasCourier, 1996 French Open quarterfinal, the ATP has Sampras winning 138 of 186 points on serve. If his 28 aces and 2 doublefaults are subtracted once, he is left winning 110 of 156 service points, which agrees with a report in the New York Times that had Sampras’ serve winning 110 points.
 For MoyaRios, 1998 French Open quarterfinal, the official Roland Garros site has Moya leading 11091 in total points. The ATP would have Moya leading 11191 if aces and doublefaults are subtracted once.  For MoyaCorretja, 1998 French Open final, the official Roland Garros site has Moya leading 10688 in total points. The ATP would have Moya leading 10788 if aces and doublefaults are subtracted once.  For MoyaCorretja, 1998 U.S. Open, R16, the official USO site has Moya leading 142127 in total points. The ATP would have Moya leading 143127 if aces and doublefaults are subtracted once. (All ’98 USO players: http://web.archive.org/web/199812070...g/scores/bios/).  For KafelnikovEnqvist, 1999 Australian Open final, an AP boxscore has Kafelnikov leading 129113 in total points. The ATP would have him leading by 130113 if aces and doublefaults are subtracted once.  For RafterBecker, 1999 Wimbledon, R16, Rafter is leading 9866 in overall points in the boxscore at Wimbledon.org; the ATP has the same number if their aces and doubles faults are subtracted once.  For AgassiKuerten, 1999 Wimbledon quarterfinal, Agassi is leading 10077 in overall points in the boxscore at Wimbledon.org; the ATP has the same number if their aces and doubles faults are subtracted once. 
11142013, 11:11 AM  #56 
Legend
Join Date: Dec 2006
Posts: 5,513

 For AgassiKrickstein in R64 and AgassiStoltenberg in R32 at 1993 Miami, the ATP has Agassi, over the course of the two matches, winning 123 of 199 points. That is confirmed in a report in the Sun Sentinel (although it remains an open question to what degree such sources were independent or simply got their numbers from the ATP):
Andre Agassi has been overpowering in his first two matches: 
11162013, 08:56 PM  #57  
Legend
Join Date: Dec 2006
Posts: 5,513

Quote:
There are quite a number of famous matches from the 90s in which the ATP incorrectly has the loser outscoring the winner in total points: AgassiSampras at the '95 AO and AgassiIvanisevic at '92 Wimbledon are two prominent examples, being GS finals. Agassi, even in a straightset win over Krajicek at Wimbledon '93, is outscored in total points in the ATP boxscore (149151), though I strongly suspect that is another mistake. If the ATP doublecounted aces and df's in the Total Points Won, then the real margin in points would be in Agassi's favor, something like 133127. 


Thread Tools  Search this Thread 
Display Modes  

