Peak level of play (Federer, Nadal, Djokovic & Co.)

[krosero]

The longevity debate we’ve been having touches on the issue of peak level of play, but only indirectly. The reasons we give for Federer’s decline do imply something about his peak level of play: if even a GOAT level player has genuinely good reasons, in today’s game, why he would fall behind his main rivals at around the age of 30, then that supports the argument that his peak level was genuinely GOAT level. If, on the other hand, there is no good reason why a 30-year-old should be falling behind his rivals in today’s game, then that lends weight to the argument that Federer is losing now because he is encountering tough rivals for the first time and his peak level was never GOAT level to begin with.

But all that was an indirect way to talk about peak level. Match stats can directly illustrate peak level of play, and I want to list those that I have gathered for Federer, Nadal and Djokovic.

The stats still have to be interpreted, but this, at least, is a direct measurement of level of play.

The most common stats used today to illustrate level of play are winners and unforced errors. Subtracting the errors from the winners, we get winner/error differentials, which can be useful. But that method has one large drawback, in that it only counts the unforced errors. The forced errors – which are almost never reported – are nowhere to be seen, and often they tell a different story.

One method that does measure forced errors is the Aggressive Margin. You can read about it here: http://www.itftennis.com/shared/medi...1_original.PDF.

Basically, in tennis you want to be as aggressive as possible while making as few errors as possible. That's what the Aggressive Margin measures. It counts the points that you win aggressively -- either by striking clean winners or by forcing your opponent into errors -- and compares that with how many points you lost by making unforced errors.

To put it most simply, if you win 25% of the all the points played in a match through aggressive plays – either by striking clean winners or by forcing your opponent into errors – and you lose 10% of all the points played in the match through unforced errors of your own, then your Aggressive Margin is 15%.

To have a high Aggressive Margin does not mean that you have to be what we normally think of as "an aggressive player." A guy who makes relatively few winners and few errors, like Nadal, can have just as high an Aggressive Margin as a guy who makes a ton of winners and errors. What matters is whether you can win points but not pay too high a cost in errors. Whoever does better at that balancing act has the higher Aggressive Margin and is almost always the winner of the match.

Over the years I’ve collected official stats for many matches, from which I’ve calculated Aggressive Margins. I have stats for all of the GS finals played by Federer, Nadal, Djokovic and Murray, against each other or against any player. I have stats for dozens more of their matches, and stats for many other players as well.

NOTE: the highest Aggressive Margins tend to occur on grass. The slower the surface, the more difficult it is to hit winners or to force errors from your opponents.

The highest Aggressive Margin I’ve ever calculated belongs to John McEnroe in the 1984 Wimbledon final: 52.8%. That does not automatically mean that his performance was the best of all time. Stats can’t be used that literally; and it’s a matter of judgment whether 52.8% against an aging Jimmy Connors is as impressive as some other performances we could name.

Any additional stats, comments, questions, arguments and corrections are most welcome.

Lists for Federer and Nadal

Lists for Djokovic, Murray, Roddick, Hewitt, Safin, Del Potro, Tsonga

[krosero]

Federer’s Aggressive Margins against Nadal at Wimbledon:

30.5% (2006)
33.1% (2007)
30.3% (2008 )

Nadal’s Aggressive Margins against Federer at Wimbledon:

22.4% (2006)
31.0% (2007)
31.5% (2008 )

This run of 3 matches might be where the Aggressive Margin method is at its most impressive. All statistical methods have pros and cons, but whatever the drawbacks of the Aggressive Margin method, it tracks very well with the scorelines of these matches.

What I mean is that every once in a while you see a stat that does not reflect the result of the match, or reflects it poorly -- like when the loser of the match has a higher winner/error differential than his opponent. Or in other cases the victor does have a better differential, but not by the margin reflected in the scoreline.

But the AM method reflects the results and scorelines of these 3 matches very well. In the 2008 match, Nadal was at 31.5% and Federer at 30.3%. Nadal was just 1.2% "better" than Federer, which reflects the fact that he barely won the match 9-7 in the fifth.

In 2007 Federer is at 33.1% and Nadal at 31.0%. The difference there is 2.1%. That looks like a little bit more of a margin of victory, though not much: and it was still a close five-set battle but Federer pulled through by a more comfortable 6-2 margin in the fifth.

In 2006 Federer is at 30.5% and Nadal at 22.4%. That looks like a comfortable victory, but not a blowout: and in fact Federer took that match in four sets. Nadal's performance there is the only one in the three matches not around the 30% level, though his figure of 22.4% is a bit deceptive because it partly reflects his very poor start in that match (he was bageled in the first set, while making a battle of the next three sets).

The AM method will not necessarily work so well with other matches, of course. But the fact that it tracks so well with the scorelines of these 3 matches is a plus for using this method.

[abmk]

The YEC 2011 match by federer vs nadal does really stand out IMO...... the surface there isn't a fast one ....... that was an extremely clean performance by federer ....

here are the AMs for a few more matches :

federer-murray YEC 2010 RR - 25.5%
federer-djokovic YEC 2010 SF - 27.7%

federer-hewitt wimbledon 2004 QF - 32.2%
federer-hewitt wimbledon 2005 QF - 31.8%
federer-grosjean wimbledon 2004 SF - 31%
federer-haas wimbledon 2009 SF - 34.2%

federer-henman USO 2004 SF - 32.3%
federer-nalbandian USO 2005 QF - 29.5%
federer-lopez USO 2007 QF - 36.7%

while fed-hewitt @ USO 2004 F was 'only' 31.1 %, for the 1st and last set combined, it was 47%

gonzalez-haas AO 2007 SF- 45.7% !!
tsonga-nadal AO 2008 SF - 34.25%
nadal-murray AO 2010 QF - 25.4%

edberg courier USO 91 final - 36.9%

nadal-almagro RG 2008 - 30.8%
soderling-davydenko RG 2009 - 27.3%

[abmk]

oh and I am assuming you calculated the RG 2008 final stat for nadal from the official website ? well those stats are totally wrong ...I don't know what the hell the statistician was doing !

I did the stats for the first set:

federer: 3 winners, 17 UEs, 7 forced errors
nadal : 6 winners, 4 UEs, 8 forced errors

they have :

federer : 9 winners, 12 UEs, 3 forced errors
nadal: 15 winners, 3 UEs , 3 forced errors

while you could say UEs are subjective, the winner counts are way off

federer won 52 points in that match, they have him @ 31 winners, nadal at 7 UEs, therefore federer only forced 14 errors from nadal ?

he had 31 winners , but forced only 14 errors from nadal?

[piece]

Great stuff. Thanks for all the work you've put into this abmk and (especially) krosero.

This measure certainly captures a lot of the important aspects of a match, but doesn't track all that well with my subjective judgements of player performance in the match (though it's doesn't do too badly either). Maybe it's the inflated stats on grass that are messing with my intuitions here.

[abmk]

like krosero already mentioned, aggressive margin depends a lot on the speed/bounce of the surface .....

it'd be good to bunch together the matches based on the surface - maybe tourneys ?

[urban]

Dont know about this method, although i appreciate the input by Krosero. Following the results, Federer had his peak level mostly 2008-12, with the Roddick matches at Wim 2003-5 in between. It is somewhat corresponding with my impressions, that Federer played more serve and volley against Roddick and Philippoussis 2003 at Wim.
Tennis level is always relative to opponent, surface and playing style. I would assume, that most players hit more winners on grass, that one player plays more for winners than another, who is defending better and basically tries to play one more ball back, to win; that is is more difficult to hit winners against fast players than against slower ones and so on.

[abmk]

Originally Posted by urban

Dont know about this method, although i appreciate the input by Krosero. Following the results, Federer had his peak level mostly 2008-12, with the Roddick matches at Wim 2003-5 in between. It is somewhat corresponding with my impressions, that Federer played more serve and volley against Roddick and Philippoussis 2003 at Wim.

what makes you think that federer had his peak level b/w 2008 and 2012 ?

that's just from the list of matches krosero has. it isn't a comprehensive list by any means ...

just that he probably has more stats of matches in 2008-12

Originally Posted by urban

Tennis level is always relative to opponent, surface and playing style. I would assume, that most players hit more winners on grass, that one player plays more for winners than another, who is defending better and basically tries to play one more ball back, to win; that is is more difficult to hit winners against fast players than against slower ones and so on.

yeah, which is why I said federer's performance vs nadal @ the YEC in 2011 stands out, it was on a medium surface against an elite defender ....

so does gonzalez's vs haas in the AO 2007 SF ....

[NatF]

That 2003 Wimbledon Semi-final was raved about by the commentators, they claimed it was possibly the highest level of tennis they'd ever seen.

Federer's numbers completely eclipse the likes of Nadal and Djokovic using this method, although as previously stated it's an uncomplete list.

[pc1]

Krosero,

Do you feel aggressive margins are more important or aggressive ratios?

One problem I have is let's say a player is very aggressive as for example Rod Laver. Laver faces a ball machine like Bjorn Borg, well Laver is probably going to hit less winners and make more errors. Obviously the surface has to be taken into account.

I think from some of your examples that is why I believe Federer is so rarely upset by players outside of the top few His forehand is almost the equivalent of a powerful volley putaway by today's standards. When he gets a short ball it is rare players outside of Nadal, Djokovic and Murray that they had a chance to win the point. Just a theory but a theory from observation.

Nadal, Murray and Djokovic don't have the same type of putaway forehand that I believe Federer has.

Over the course of a year or a few years, if all the matches were charted I can easily see this as a super valuable statistic. Great job.

[abmk]

Originally Posted by pc1

Krosero,

Do you feel aggressive margins are more important or aggressive ratios?

by aggressive ratio you mean : no of points won by a player through forcing/total number of points in the match


where no of points won by forcing = winners+errors forced from opponent ?

Originally Posted by pc1

One problem I have is let's say a player is very aggressive as for example Rod Laver. Laver faces a ball machine like Bjorn Borg, well Laver is probably going to hit less winners and make more errors. Obviously the surface has to be taken into account.

yeah, the surface matters a lot here , which is why it'd be better to arrange it by the surface ....

[abmk]

I have a match for nadal with AM greater than 40%. Would like to see if anyone can guess it before I post it .

Hint: a match vs a player within the top 50 ....

[abmk]

at wimbledon : 92 QF vs stich or 99 F vs agassi ?

Having actually watched those 2 matches , I feel sampras played a tad bit better in the stich match ( I can see the sampras fans angrily shouting here ! ) - especially returning wise ...

don't have the stats with me for either of the matches now , but I feel sampras' AM would be better in the 92 QF ....

want to see how many of those who watched both those matches agree with me !

[pc1]

Originally Posted by abmk

by aggressive ratio you mean : no of points won by a player through forcing/total number of points in the match


where no of points won by forcing = winners+errors forced from opponent ?



yeah, the surface matters a lot here , which is why it'd be better to arrange it by the surface ....

I mean if a player hits 45 winners and has 15 errors, that would be a 3 to 1 ratio. Another player may hit 15 winners and have only 5 errors but has the same 3 to 1 ratio.

If you do plus-minus the first guy is far better at plus 30 than the second guy at plus 10.

I think ratio may be better.

One other thing we should considering is normalizing the information for the eras. They do that in the National Football League and Major League Baseball all the time. I wrote an article for a well known NFL magazine a few years ago normalizing information on the teams stats of NFL teams from the 1940's onward to the present. The results were in line with the relative dominance of the teams. In the wood era there were far more errors than winners. We should perhaps consider ratios for the different eras also. So era like the 1970's and 1980's had a mix of different racquets and equipment. So used wood into the 1980's and switched to a more modern racquet. Some have argue that Chris Evert was at a huge disadvantage against Martina Navratilova for a while when Navratilova was beat Evert all the time because Evert used wood and Navratilova the most modern racquet. One Evert switched to a modern racquet, it was competitive against.

Some statisticians have used standard deviations as a way of normalizing information. We obviously don't have this information available but it would be wonderful if some researcher was able to find as many matches as possibly to get a large statistical sample.

[krosero]

Thanks all for the comments, will post replies as a I have time.

Just a few quick replies now.

Originally Posted by abmk

oh and I am assuming you calculated the RG 2008 final stat for nadal from the official website ? well those stats are totally wrong ...I don't know what the hell the statistician was doing !

I did the stats for the first set:

federer: 3 winners, 17 UEs, 7 forced errors
nadal : 6 winners, 4 UEs, 8 forced errors

they have :

federer : 9 winners, 12 UEs, 3 forced errors
nadal: 15 winners, 3 UEs , 3 forced errors

while you could say UEs are subjective, the winner counts are way off

federer won 52 points in that match, they have him @ 31 winners, nadal at 7 UEs, therefore federer only forced 14 errors from nadal ?

he had 31 winners , but forced only 14 errors from nadal?

Yes those official stats are messed up, but fortunately they don't have an impact on the AM. As long as the unforced errors can be trusted -- and they do look ok -- then the remaining points can be regarded as either forcing points or as clean winners, without worrying about how many exactly were clean winners. That's the problem with the official stats: the figures that they have for Winners are actually the number of points that each player won, in total, on serve (ie, Federer won 31 points on his own serve, Nadal won 46 on his own).

In 2007 the RG site also had a problem counting Winners, and I don't know how the Winners were calculated that year. Fortunately, again, it doesn't have an impact on the AM, so long as the Unforced Errors look reliable, which they do.

Originally Posted by abmk

at wimbledon : 92 QF vs stich or 99 F vs agassi ?

Having actually watched those 2 matches , I feel sampras played a tad bit better in the stich match ( I can see the sampras fans angrily shouting here ! ) - especially returning wise ...

don't have the stats with me for either of the matches now , but I feel sampras' AM would be better in the 92 QF ....

want to see how many of those who watched both those matches agree with me !

The official Wimbledon stats in '99 have Unforced Errors that are completely unreliable. Those figures look like the total number of errors of every kind, forced and unforced. They've got Sampras at 51 unforced errors. NBC put Sampras at only 22, which looks right. Unfortunately I don't know the UE count that NBC had for Agassi, so AM's can't be calculated.

Do you have any idea what the UE counts were for the Stich match?

Or the UE count for Becker in the '95 Wimby final?

Originally Posted by abmk

by aggressive ratio you mean : no of points won by a player through forcing/total number of points in the match


where no of points won by forcing = winners+errors forced from opponent ?

That would just be a measure of how aggressive a style a player has, and it would be clearly advantageous to big hitters.

I thought he meant:
(a player's points won by making winners and by forcing opponent into errors) / (a player's unforced errors).

PC1?

[krosero]

PC1, thanks for your answer, I see you posted it while I was composing my post. Will reply later.

[Feather]

Great effort krosero, great effort. Hats off to you for the post and thanks a lot for sharing it

[RoddickAce]

Originally Posted by krosero

To have a high Aggressive Margin does not mean that you have to be what we normally think of as "an aggressive player." A guy who makes relatively few winners and few errors, like Nadal, can have just as high an Aggressive Margin as a guy who makes a ton of winners and errors. What matters is whether you can win points but not pay too high a cost in errors. Whoever does better at that balancing act has the higher Aggressive Margin and is almost always the winner of the match.

I think your analysis is quite interesting, but I disagree with the bolded statement. It also depends on what you define as peak. Are we referring to a period of time? ie. 2005-2006 for Federer. Or are we referring to peak levels in certain matches? If you are referring to peak level in certain matches, then that is fine, but players like Berdych, Tsonga and Del Potro will have very high margins for certain matches. If we are drawing a conclusion from the sample about a period of time, then there are some points to consider (see the reasons below).

To clarify, I don't think there is anything wrong with biasing towards more aggressive players since you are effectively trying to measure how aggressive a player relative to his/her errors.

Reasons for bias towards aggressive players:

1) Aggressive players play high risk high reward tennis, over the wide spread of matches played throughout one's career, they are more likely to have wins with a high margin of aggressively won points over unforced errors, than their defensive counterparts. This can cause skewed results. For example, let's say Isner and Nadal play 150 matches against the field during the defined "peak period" (150 matches is roughly 2-2.5 years).

• Isner -Being a risky player, Isner always has a high winner/forced error, high unforced ratio. Over his 150 matches, his aggressive margin is 10%, because of his high unforced error count. However, he gets on fire 10 of the 150 (6.67%) matches he played, and has a high margin of 50% for his top 10 scores, supposedly representing his peak period.

• Nadal -Being a defensive player, Nadal has a lower winner, lower unforced ratio. Over his 150 matches, his aggressive margin is 25%. Because of his steady play, this 25% held true for each of his 200 matches. Thus, his top 10 matches also average 25%, which is 1/2 of what Isner achieved in his top 10 matches.

This clearly does award players that "we normally think of as "an aggressive player."

To reduce this bias, perhaps one can define a peak period, say all matches in 2005 and 2006, and then select a random sample of matches to calculate the aggressive margin, rather than selecting the top aggressive margins over a large amount of years.

2) Also, using a direct calculation of % points won aggressively subtracting the % points lost due to unforced errors will be advantageous for the aggressive player.

ie: In a match, Isner won 50% of points aggressively, but lost 30% of points due to unforced errors. The aggressive margin is 20%. The RATIO is 50/30 = 1.67 times more aggressively won points than errors.

Nadal won 15% of points aggressively, but lost 2% of points due to unforced errors. The aggressive margin is 13%. The RATIO is 7.5 times more aggressively won points than errors.

Which statistic is actually more impressive in terms of "balancing aggressive shots and errors" as you said you were trying to find out from the stats?

[Moose Malloy]

Or the UE count for Becker in the '95 Wimby final?

I had Becker with 7. I know that's unofficial, but I'm curious what the AM would be in that match. That goes for any of our stats where we calculated UE's(Laver-Ashe etc)
You could just mention that the AM was calculated with unofficial stats when posting some of these(I doubt Laver ever played a match where officials were calculating winners/ue's, may as well use ours just for fun)

I guess the winner # isn't really necessary in calculating any AM's, just the UE count for both players.

[fed_rulz]

Great post Krosero! I like the AM method of evaluating peaks.

Here's a question: are service winners included in the opponents forced error column? IMO, this method would be more accurate if serve-related stats are excluded in calculating the AM.