Cumulative stats in AO16 - Fed vs. Nole (SF)

#1
During USO15 I bullishly predicted Federer would take Djokovic out, on the basis of stats. But he choked big time, and I made a thread about it.

Like USO15, in AO16 Federer has kicked Djokovic's ass on the paper battle. I hope these stats convince you that, pound-for-pound, Federer has been a better tennis player than Djokovic this past week and a half.

A large part of what remains is self-belief. Federer still needs to prove that he can mentally rise to the big occasion vs. peak Nole in a major. Until he does that, I will not predict a Federer victory based on these stats. Not least because, as @helterskelter points out, Nole's opponents may not have brought out the best in him; and he may have another gear! But I hope you enjoy the stats anyway.



1. Federer was much more dominant overall.
  • The one stat which captures a player's overall level is dominance ratio. i.e. % of pts lost on serve, divided by % of points lost by your opponents on your serve. For example, Federer won 45% of return points, and his opponents won 26% of points on his serve. So D/R = 45% / 26% = 1.70
  • Federer's ratio was much higher than Djokovic (1.70 vs. 1.46). This is what I mean that Federer was better pound-for-pound. (Given that their competition was comparable.)
  • Here is an illustration of how much better 1.70 is than 1.46. For example, 1.71 is the dominance ratio Federer achieved over 12 matches in the first three rounds in his 4 title years at the Australian open. Against players ranked 83 on average. 1.48 is the dominance ratio he achieved in the 8 semis and finals of those same tournaments. Against players ranked 16 on average*.

2. Federer was more dominant on serve.
  • Fed won more points on serve (74% vs. 70%). This was the result of a higher first serve winning percentage (84% vs. 75%.).
  • Djokovic's second serve was better (61% vs. 55%). In slow conditions when the second serve itself wins fewer points, this suggests Djokovic had a better ground game.
  • Federer was broken fewer times (5 vs. 8). This is because Federer faced break points in fewer service games (14% vs. 19%). Although they both lost exactly half of those games when they DID face BPs.

3. Federer was better on return of serve too
  • Fed was much better at return. Winning 39% of return games compared to 32% for Djokovic.
  • However, both players won roughly the same number of return points (45% vs. 44%), and had similiar success on break points (43% vs. 42%).
  • This implies Federer's return points were clustered better. Which means more return games lost at love and 15, but also more games with break points. It's hard to say whether this is chance, or good strategy at selective effort.
  • As a result of better clustering, Federer saw break points in more return games than Djokovic (53% vs. 47%). When he did earn break points, he earned more than Djokovic did (1.71 per game, vs. 1.62 per game - not shown in chart). As a result, Federer was able to win more of the games in which he had break points (74% vs. 68%).

4. Fed had a much cleaner game. Comparable number of winners but far fewer unforced errors, leading to a far better differential for Fed (+58 vs. -5). If we remove the Djokovic-Simon match, Federer still had a cleaner game (differential of +58 vs. +33).


5. Not much difference in net play. This was a shocker. Both went to the net about as often as the other (11% vs. 13%), and won a comparable number of points (71% vs. 76%)


6. Opponents were of comparable quality. Fed's opponents' average rank was a little higher (35 vs. 50), but they only had a slightly better record on hard courts in the last year (74% vs. 71%).


7. Federer should be much less fatigued. Far less time on court (9 vs. 12 hours), almost 200 fewer points played and two-thirds the distance run. Thanks to his aggressive style, Fed ran less per point (9.7 meters per point vs 12.6).



(blue means substantially better)




* this average of 16 excludes Safin's ranking of 86 in 2004
 
Last edited:
#2
Thanks for the stats!

I hate to say it, but we have definitely seen this script before. Fedr leading most of the statistics heading into a Slam clash with Djokovic, but the Serb is the one emerging victorious.

It will take an extraordinary serving performance for Fedr to buck the trend.
 

NatF

Bionic Poster
#3
A lot of work but I would be interested in the cumulative stats of their opponents as well ;)

Good job though. These stats don't mean much in the context of the match up though.
 
#7
Nice stats. However time spent on court doesn't really play a big part IMO since one player is still well in his prime while the other is in his mid 30s.
The more Federer leads in the stats, the more he should watch out for Djokovic. ;)
 
#9
I hate to say it, but we have definitely seen this script before. Fedr leading most of the statistics heading into a Slam clash with Djokovic, but the Serb is the one emerging victorious.
These stats don't mean much in the context of the match up though.
The more Federer leads in the stats, the more he should watch out for Djokovic. ;)
yes agree with all these comments. I was writing up the OP while you guys were making the same observations. Federer has it all to prove. There is no doubt that in this tournament he has played as good tennis, pound for pound, as Djokovic. He just needs to show up with self-belief to the SF and he can do this.

But will he? That's a different story! History suggests he won't....
 
#10
That's the frustrating part, because it doesn't seem to matter how well either played leading up to their Major matches, Djokovic always seems to wake up and play fantastic tennis vs. Federer (although lately even more fantastically vs. Nadal, partly due to Nadal being in poor form lately).

I'd be curious to see stats on how much Federer's game drops from its tournament average, and Djokovic's rises from its tournament average. Maybe it being a SF will help a bit, but the reality is that all of the pressure is on Federer and there is very little on Djokovic, unless he feels a lot of pressure to match #17, but even that won't matter much in the prime of his capabilities.

It was different in 2012, when Federer essentially went into Wimbledon with the mindset that no-one else was going to take it from him. The difference is that now every Major is, possibly, one of his last chances. That is a lot of pressure. And of course he's older.
 
#15
During USO15 I bullishly predicted Federer would take Djokovic out, on the basis of stats. But he choked big time, and I made a thread about it.

Like USO15, in AO16 Federer has kicked Djokovic's ass on the paper battle. I hope these stats convince you that, pound-for-pound, Federer has been a better tennis player than Djokovic this past week and a half.

All that remains is self-belief. Federer still needs to prove that he can mentally rise to the big occasion vs. peak Nole in a major. Until he does that, I will not predict a Federer victory based on these stats.

But I hope you enjoy the stats anyway.



Federer was more dominant on serve.
  • Fed won more points on serve (74% vs. 70%). This was the result of a higher first serve winning percentage (84% vs. 75%.).
  • Djokovic's second serve was better (61% vs. 55%). In slow conditions when the second serve itself wins fewer points, this suggests Djokovic had a better ground game.
  • Federer was broken fewer times (3 vs. 5). This is because Federer faced break points in fewer service games (14% vs. 19%). Although they both lost exactly half of those games when they DID face BPs.

Federer was better on return of serve too
  • Fed was much better at return. Winning 39% of return games compared to 32% for Djokovic.
  • However, both players won roughly the same number of return points (45% vs. 44%), and had similiar success on break points (43% vs. 42%).
  • This implies Federer's return points were clustered better. Which means more return games lost at love and 15, but also more games with break points. It's hard to say whether this is chance, or good strategy at selective effort.
  • Federer saw break points in more return games than Djokovic (53% vs. 47%). And when he did earn break points, he earned more than Djokovic did (1.71 per game, vs. 1.62 per game - not shown). As a result, Federer won more of the games in which he had break points (74% vs. 68%).

Fed had a much cleaner game. Comparable number of winners but far fewer unforced errors, leading to a far better differential for Fed (+58 vs. -5). If we remove the Djokovic-Simon match, Federer still had a cleaner game (differential of +58 vs. +33).


Not much difference in net play. This was a shocker. Both went to the net about as often as the other (11% vs. 13%), and won a comparable number of points (71% vs. 76%)


Opponents were of comparable quality. Fed's opponents' average rank was a little higher (35 vs. 50), but they only had a slightly better record on hard courts in the last year (74% vs. 71%).


Federer should be much less fatigued. Far less time on court (9 vs. 12 hours), almost 200 fewer points played and two-thirds the distance run. Thanks to his aggressive style, Fed ran less per point (9.7 meters per point vs 12.6).



(blue means substantially better)

Really solid stats. Love the analysis with these comparisons. Fed's been so strong... only a select couple of moments in his five matches (15-1 in sets!!) where I felt anxiety about him progressing. Have had to pinch myself many times that this is happening at 34 1/2... Djok has some vulnerabilities. Very optimistic about the upcoming meeting, particularly in SF.
 
#17
@falstaff78 Nice stats. I think it's clear that Federer has played better coming in. But I'm not sure that it's just his own self-belief that is important on Thursday. I think the other crucial thing is that we've seen Djokovic start coasting through early rounds without really trying that hard time and time again in the last year. He does just enough to win...and no more. Then when he faces Federer, Nadal, or Murray, he raises his game. Most recently in Doha, he beat L. Mayer 6-3 7-5 and Berdych 6-3 7-6, and then took out Nadal 6-1 6-2. At Wimbledon, also, he raised his game for the final significantly compared to the 4th round. Even at the US Open, where Federer being nervous was a crucial factor, Djokovic was, I think, slightly better than in the matches against Bautista Agut and Lopez.

So, Federer needs to be calm, but he also needs to be prepared to face a Djokovic whom he won't find on the videotapes if he watches the earlier rounds.
 
#18
@falstaff78 Nice stats. I think it's clear that Federer has played better coming in. But I'm not sure that it's just his own self-belief that is important on Thursday. I think the other crucial thing is that we've seen Djokovic start coasting through early rounds without really trying that hard time and time again in the last year. He does just enough to win...and no more. Then when he faces Federer, Nadal, or Murray, he raises his game. Most recently in Doha, he beat L. Mayer 6-3 7-5 and Berdych 6-3 7-6, and then took out Nadal 6-1 6-2. At Wimbledon, also, he raised his game for the final significantly compared to the 4th round. Even at the US Open, where Federer being nervous was a crucial factor, Djokovic was, I think, slightly better than in the matches against Bautista Agut and Lopez.

So, Federer needs to be calm, but he also needs to be prepared to face a Djokovic whom he won't find on the videotapes if he watches the earlier rounds.
First meeting btw. them with Ljubicic in Fed's camp, also. The game he's playing is markedly different. Djok's got an x-factor waiting for him here. He is not thrilled that Fed is still here, still firing on many cylinders (and utilizing previously dormant ones). :)
 
#19
Shows us again how underrated Fed's return is. That said however, I sadly have to agree with most posters here. This match-up seems to have become a real mental block for Federer lately, and sadly, I expect the monster to show its ugly head once again on Thursday. Regardless of what the stats show, I just know Djoker will Djoke us all and peak against Rog.

But really nice work OP. Awesome stuff. :)
 
#21
A lot of work but I would be interested in the cumulative stats of their opponents as well ;)
what would this tell us?

I included some measures of opponent quality. namely opponents' average rank, and opponents' hard court record in the last year against middling players. (ranked 20-100). so we can tell that their competitors were comparable. if not slightly better for Federer
 

NatF

Bionic Poster
#25
what would this tell us?

I included some measures of opponent quality. namely opponents' average rank, and opponents' hard court record in the last year against middling players. (ranked 20-100). so we can tell that their competitors were comparable. if not slightly better for Federer
A better indication of the form of their opponents. Ranking and record in the past year can only tell us so much. Their performance at the AO itself is going to be more revealing.
 
#26
Federer's first serve % in ALL matches against Djokovic:

2006 Monte Carlo = 53%
2006 Davis Cup - unknown
2007 Australian Open = 63%
2007 Dubai = 62%

2007 Montreal = 68%
2007 US Open = 60%
2008 Australian Open = 61%
2008 Monte Carlo = 52%
2008 US Open = 64%

2009 Miami = 60%
2009 Rome = 49%
2009 Cincinnati = 55%
2009 US Open = 58%

2009 Basel = 71%
2010 Toronto = 58%
2010 US Open = 53%
2010 Shanghai = 62%
2010 Basel = 58%
2010 WTF = 62%

2011 Australian Open = 60%
2011 Dubai = 60%
2011 Indian Wells = 56%
2011 French Open = 65%
2011 US Open = 61%
2012 Rome = 49%
2012 French Open = 56%
2012 Wimbledon = 64%
2012 Cincinnati = 59%

2012 WTF = 61%
2013 Paris = 66%
2013 WTF = 54%
2014 Dubai = 60%
2014 Indian Wells = 63%
2014 Monte Carlo = 70%
2014 Wimbledon = 69%
2014 Shanghai = 71%
2015 Dubai = 65%

2015 Indian Wells = 60%
2015 Rome = 64%
2015 Wimbledon = 67%
2015 Cincinnati = 59%
2015 US Open = 64%
2015 WTF RR = 69%
2015 WTF F = 67%

Fed won the matches in bold. This doesn't tell us much other than the fact he needs to rely more on his serve nowadays.
 
#28
A better indication of the form of their opponents. Ranking and record in the past year can only tell us so much. Their performance at the AO itself is going to be more revealing.
I see what you are saying. Yes I agree that ranking and 12 month form are not ideal. But at least they can give us some comfort that broadly speaking there wasn't any massive difference. Like, say, Nadal in 2010 or something.

Agree that a more serious analysis would be really painful. Not least because then you have to start thinking about the performance of opponents' opponents!!!
 
#30
Break point conversion is gonna be key for Fed. Don't have the hard numbers for the tournament handy, but I do know that he's been marginally better than some of the recent nightmares. Yes, there's been a lot of the same baffling botchery (e.g., wasting three break points at Love-40 in the blink of an eye), but it hasn't ended up being very costly so far. Anyone have the actual data on it? I want Fed to capitalize on these in a big way in the SF!
 
#32
Break point conversion is gonna be key for Fed. Don't have the hard numbers for the tournament handy, but I do know that he's been marginally better than some of the recent nightmares. Yes, there's been a lot of the same baffling botchery (e.g., wasting three break points at Love-40 in the blink of an eye), but it hasn't ended up being very costly so far. Anyone have the actual data on it? I want Fed to capitalize on these in a big way in the SF!
Yep. It's in the chart section 5 "return"

Fed converted 43%, which is 2% less than his win rate on other return points.
Nole converted 42%, which is 2% less than his win rate on other return points.

So they are comparable, and the level of "unclutchess" is comparable too!
 

NatF

Bionic Poster
#33
I see what you are saying. Yes I agree that ranking and 12 month form are not ideal. But at least they can give us some comfort that broadly speaking there wasn't any massive difference. Like, say, Nadal in 2010 or something.

Agree that a more serious analysis would be really painful. Not least because then you have to start thinking about the performance of opponents' opponents!!!
I actually started doing this for Grand Slam wins - I worked out the cumulative D/R of all Federer/Nadal/Djokovic's opponents in their slam victories (QF onwards). Then I started looking at their opponents opponents. It became a big of a drag but I hope to finish it at some point.

It was interesting though.

I hope you don't take my mentioning of even greater detail as me saying that I don't appreciate the work you put it. By itself this is certainly interesting. But you can never have too many statistics ;)
 
#34
I actually started doing this for Grand Slam wins - I worked out the cumulative D/R of all Federer/Nadal/Djokovic's opponents in their slam victories (QF onwards). Then I started looking at their opponents opponents. It became a big of a drag but I hope to finish it at some point.

It was interesting though.
haha yes I have an unfinished spreadsheet on that topic somewhere.

using tennis abstract I assume?
 
#35
Yep. It's in the chart section 5 "return"

Fed converted 43%, which is 2% less than his win rate on other return points.
Nole converted 42%, which is 2% less than his win rate on other return points.

So they are comparable, and the level of "unclutchess" is comparable too!
Okay... the percentage is helpful (missed that - thank you). 43% is still way low for me. Do you happen to have the actual numbers though? Like producing the break points in the first place is always great "scoreboard pressure" and probably contributes to wearing the opponent down (so if he's producing more than twice the number of his opponents, for instance...). I think the numbers themselves may be the most telling. 5/23 at USO was a "brown standard," and #18 went down the drain because of it.
 
#37
Okay... the percentage is helpful (missed that - thank you). 43% is still way low for me. Do you happen to have the actual numbers though? Like producing the break points in the first place is always great "scoreboard pressure" and probably contributes to wearing the opponent down (so if he's producing more than twice the number of his opponents, for instance...). I think the numbers themselves may be the most telling. 5/23 at USO was a "brown standard," and #18 went down the drain because of it.
Federer 28/65
Djokovic 25/60
 
#44
During USO15 I bullishly predicted Federer would take Djokovic out, on the basis of stats. But he choked big time, and I made a thread about it.

Like USO15, in AO16 Federer has kicked Djokovic's ass on the paper battle. I hope these stats convince you that, pound-for-pound, Federer has been a better tennis player than Djokovic this past week and a half.

A large part of what remains is self-belief. Federer still needs to prove that he can mentally rise to the big occasion vs. peak Nole in a major. Until he does that, I will not predict a Federer victory based on these stats. Not least because, as @helterskelter points out, Nole's opponents may not have brought out the best in him; and he may have another gear! But I hope you enjoy the stats anyway.



1. Federer was much more dominant overall.
  • The one stat which captures a player's overall level is dominance ratio. i.e. % of pts lost on serve, divided by % of points lost by your opponents on your serve. For example, Federer won 45% of return points, and his opponents won 26% of points on his serve. So D/R = 45% / 26% = 1.70
  • Federer's ratio was much higher than Djokovic (1.70 vs. 1.46). This is what I mean that Federer was better pound-for-pound. (Given that their competition was comparable.)
  • Here is an illustration of how much better 1.70 is than 1.46. For example, 1.71 is the dominance ratio Federer achieved over 12 matches in the first three rounds in his 4 title years at the Australian open. Against players ranked 83 on average. 1.48 is the dominance ratio he achieved in the 8 semis and finals of those same tournaments. Against players ranked 16 on average*.

2. Federer was more dominant on serve.
  • Fed won more points on serve (74% vs. 70%). This was the result of a higher first serve winning percentage (84% vs. 75%.).
  • Djokovic's second serve was better (61% vs. 55%). In slow conditions when the second serve itself wins fewer points, this suggests Djokovic had a better ground game.
  • Federer was broken fewer times (3 vs. 5). This is because Federer faced break points in fewer service games (14% vs. 19%). Although they both lost exactly half of those games when they DID face BPs.

3. Federer was better on return of serve too
  • Fed was much better at return. Winning 39% of return games compared to 32% for Djokovic.
  • However, both players won roughly the same number of return points (45% vs. 44%), and had similiar success on break points (43% vs. 42%).
  • This implies Federer's return points were clustered better. Which means more return games lost at love and 15, but also more games with break points. It's hard to say whether this is chance, or good strategy at selective effort.
  • As a result of better clustering, Federer saw break points in more return games than Djokovic (53% vs. 47%). When he did earn break points, he earned more than Djokovic did (1.71 per game, vs. 1.62 per game - not shown in chart). As a result, Federer was able to win more of the games in which he had break points (74% vs. 68%).

4. Fed had a much cleaner game. Comparable number of winners but far fewer unforced errors, leading to a far better differential for Fed (+58 vs. -5). If we remove the Djokovic-Simon match, Federer still had a cleaner game (differential of +58 vs. +33).


5. Not much difference in net play. This was a shocker. Both went to the net about as often as the other (11% vs. 13%), and won a comparable number of points (71% vs. 76%)


6. Opponents were of comparable quality. Fed's opponents' average rank was a little higher (35 vs. 50), but they only had a slightly better record on hard courts in the last year (74% vs. 71%).


7. Federer should be much less fatigued. Far less time on court (9 vs. 12 hours), almost 200 fewer points played and two-thirds the distance run. Thanks to his aggressive style, Fed ran less per point (9.7 meters per point vs 12.6).



(blue means substantially better)




* this average of 16 excludes Safin's ranking of 86 in 2004
Great work as always, thanks for sharing!!
 
#45
I think this is an excellent comment. I will update the OP to reflect this. You are absolutely spot on!
Thanks! I think it's a credit to Federer that he is not getting squashed by Djokovic in the way that Nadal and Murray are. Even when Djokovic raises his game, Federer usually wins a set, sometimes wins, and even when he loses in straight sets is respectable (e.g. 6-3 6-4 in the WTF final or 6-4 6-3 in the Rome final). He hasn't been beaten 6-1 6-2 and 6-2 6-2 like Nadal has recently, or 6-1 6-3 and 6-2 6-3 like Murray was last year.

Bearing in mind that Federer is 34, whereas Nadal is 29 and Murray 28, that says a lot about him.

On the other hand, he was a bit careless about the #2 ranking in the fall of last year. If he had put in more effort in Shanghai and Paris, he'd likely still be #2 now, and so wouldn't have to face Djokovic until the final. I think Federer could have come through anyone else. Of course, if he pulls off the win, it won't matter. But he's made life harder for himself than it needed to be.
 
#47
It's not like Feds level drops in big stages against Nole(and all that stats become crap), it's all about Nole don't afford Fed to play his game with him(like all the tour does). Hopefully Nole vs Murray final will be more unpredictable.
 
#50
I almost think it's advantageous that they are facing off at Semi stage, rather than Final. Re the ranking, I think Murray's Davis Cup points ended up being the difference maker. To paraphrase Tomas Berdych, "I like Fed's chances." ;)
I am inclined to think this is the case. Or at least I am inclined to HOPE this is the case!:)

Seriously though, the fact that it's a semi should relieve some of the pressure on Fred.

Fingers crossed
 
Top