5)This does not account for levels of opponents
I think it was established earlier on in the thread that a high AM against an opponent also with a high AM is more impressive than a high AM against an opponent who's AM is low.
[B]AMa = (Wa - Aa + FEb - UEa/r)/TP[/B]
Pa = Wa + FEb + UEb
[B]AMa = ((Pa - Aa) - (UEb + UEa/r))/TP[/B]
TP = Total Points in the entire match
A = Aces
FE = Forced Errors
UE = Unforced Errors
W = Winners
P = Points won
r = Rally Length
[B]EMa = (Wa - Aa - Ea/r)/TP[/B]
Pb = Wb + Ea
[B]EMa = ((Wa - Aa) - (Pb - Wb)/r)/TP[/B]
TP = Total Points in the entire match
A = Aces
E = Errors (forced + unforced)
W = Winners
P = Points won
r = Rally Length
Becker may have had an extraordinarily high AM against Milan Srejber at the '86 USO (beat him 3, 2 and 1). Miami Herald reported he made no unforced errors.
No not as far as I know.Interesting feat. Do we have stats on other matches without UE?
I would like to see AM calculated differently.
1. Will discount aces.
2. Will introduce a new factor, "r", which is the rally length. The greater the length of rallies, lesser the effect of UE should be on AM. This will help us in assessing matches like AO '12 Final. Like, "r" will be greater for RG than say for example WC matches. It can be roughly measured as "(duration of match) / (total points)".
3. I will also calculate AM based on their entire errors without the distinction FE and UE. Let me call it EM (Error Margin - nothing much, just a stupid name). This helps in comparing dissimilar matches which doesn't have UE calculation anomalies.
Equation:
But,Code:[B]AMa = (Wa - Aa + FEb - UEa/r)/TP[/B]
Code:Pa = Wa + FEb + UEb
Hence,Code:[B]AMa = ((Pa - Aa) - (UEb + UEa/r))/TP[/B]
Where
Code:TP = Total Points in the entire match A = Aces FE = Forced Errors UE = Unforced Errors W = Winners P = Points won r = Rally Length
Similarly for EM
Equation,
Code:[B]EMa = (Wa - Aa - Ea/r)/TP[/B]
But,
Code:Pb = Wb + Ea
Hence,
Code:[B]EMa = ((Wa - Aa) - (Pb - Wb)/r)/TP[/B]
Where
Code:TP = Total Points in the entire match A = Aces E = Errors (forced + unforced) W = Winners P = Points won r = Rally Length
AMs and EMs of few matches:
r = 4 (I remember the stat shown on TV during the 2011 match during the 2nd set that the average rally length on Federer's serve is 4 and on Nadal's serve is 8. Seemed too much to me then, quite unreal. Think 4 is reasonable)
2010 London
Federer - 32.7%, 10.6%
Nadal - 21.7%, -0.4%
2011 London
Federer - 48.2%, 18.8%
Nadal - 20.1%, -4.6%
2011 Rome, r = 6
Djokovic - 32.1%, 14.1%
Nadal - 18%, 4.5%
r = 5
2010 UO
Nadal - 29.3%, 10.3%
Djokovic - 28.9%, 8.9%
2011 UO
Djokovic - 34.3%, 11.2%
Nadal - 23%, 4.4%
2013 UO
Nadal - 28.3%, 6.6%
Djokovic - 29.3%, 9.5% (oddly, favours Nole for his attacking game)
r = 4
2007 WC
Federer - 33.6%, 4.3% (goes to show Roger didn't play well if we exclude his aces)
Nadal - 36.2%, 7.4%
2008 WC
Nadal - 34.9%, 6.1%
Federer - 33.7%, 6.8%
2012 AO, r = 6 (this is true, I remember the stat. At one stage average length was eight)
Djokovic - 27.5%, 7%
Nadal - 23.1%, 3.1%
2005 AO, r = 4
Safin - 26.3%, 4.2%
Federer - 26.4%, 4.5%
2010 WC, r = 3
Isner - 31.5%, 4.8%
Mahut - 34.1%, 6.5%
Not very perfect, but I think does a better job. r factor is reasonably chosen according to the match. EMs not surprisingly helps in cross-comparing dissimilar matches better.
I would like to see AM calculated differently.
1. Will discount aces.
.
Aggressive Margins in USO finals:
2013 - Nadal 22%, Djokovic 13%
2010 - Nadal 23%, Djokovic 16%
2011 - Djokovic 22%, Nadal 13%
The two players essentially flip-flopped in '11 and '13. Even in the scoreline there was a resemblance; both years the winner ran away with the match 6-1 in the fourth.
2010, slightly, remains the best-played match per the AM's.
Quite possible. AM's for that match were:Sort of what I thought, incredible stats and an impressive level of play - probably his best serving performance. But not his best match ever by any stretch. His performance against Roddick in 2003 is still his best Wimbledon performance IMO.
Seems very reasonable to me; and going by that logic Federer's performance in '03 was a little higher than yesterday's, at least statistically. I put in that caveat because comparing AM's across such a long time span can be problematic; and there are also match-ups issues, in any comparison, of course.I think it was established earlier on in the thread that a high AM against an opponent also with a high AM is more impressive than a high AM against an opponent who's AM is low.
I would like to see AM calculated differently.
1. Will discount aces.
2. Will introduce a new factor, "r", which is the rally length. The greater the length of rallies, lesser the effect of UE should be on AM. This will help us in assessing matches like AO '12 Final. Like, "r" will be greater for RG than say for example WC matches. It can be roughly measured as "(duration of match) / (total points)".
3. I will also calculate AM based on their entire errors without the distinction FE and UE. Let me call it EM (Error Margin - nothing much, just a stupid name). This helps in comparing dissimilar matches which doesn't have UE calculation anomalies.
Equation:
But,Code:[B]AMa = (Wa - Aa + FEb - UEa/r)/TP[/B]
Code:Pa = Wa + FEb + UEb
Hence,Code:[B]AMa = ((Pa - Aa) - (UEb + UEa/r))/TP[/B]
Where
Code:TP = Total Points in the entire match A = Aces FE = Forced Errors UE = Unforced Errors W = Winners P = Points won r = Rally Length
Similarly for EM
Equation,
Code:[B]EMa = (Wa - Aa - Ea/r)/TP[/B]
But,
Code:Pb = Wb + Ea
Hence,
Code:[B]EMa = ((Wa - Aa) - (Pb - Wb)/r)/TP[/B]
Where
Code:TP = Total Points in the entire match A = Aces E = Errors (forced + unforced) W = Winners P = Points won r = Rally Length
AMs and EMs of few matches:
r = 4 (I remember the stat shown on TV during the 2011 match during the 2nd set that the average rally length on Federer's serve is 4 and on Nadal's serve is 8. Seemed too much to me then, quite unreal. Think 4 is reasonable)
2010 London
Federer - 32.7%, 10.6%
Nadal - 21.7%, -0.4%
2011 London
Federer - 48.2%, 18.8%
Nadal - 20.1%, -4.6%
2011 Rome, r = 6
Djokovic - 32.1%, 14.1%
Nadal - 18%, 4.5%
r = 5
2010 UO
Nadal - 29.3%, 10.3%
Djokovic - 28.9%, 8.9%
2011 UO
Djokovic - 34.3%, 11.2%
Nadal - 23%, 4.4%
2013 UO
Nadal - 28.3%, 6.6%
Djokovic - 29.3%, 9.5% (oddly, favours Nole for his attacking game)
r = 4
2007 WC
Federer - 33.6%, 4.3% (goes to show Roger didn't play well if we exclude his aces)
Nadal - 36.2%, 7.4%
2008 WC
Nadal - 34.9%, 6.1%
Federer - 33.7%, 6.8%
2012 AO, r = 6 (this is true, I remember the stat. At one stage average length was eight)
Djokovic - 27.5%, 7%
Nadal - 23.1%, 3.1%
2005 AO, r = 4
Safin - 26.3%, 4.2%
Federer - 26.4%, 4.5%
2010 WC, r = 3
Isner - 31.5%, 4.8%
Mahut - 34.1%, 6.5%
Not very perfect, but I think does a better job. r factor is reasonably chosen according to the match. EMs not surprisingly helps in cross-comparing dissimilar matches better.
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.
And all the Fedal meetings at AOFederer's AMs at AO 2017 :
vs melzer : 22.9%
vs rubin : 23.36%
vs berdych : 35.76%
vs nishikori : 29.17%
vs M. zverev : 42.04% ( due to more of net play )
vs wawrinka : 19.79%
vs nadal : 22.49%
Nadal's AMs at AO 2017 :
vs mayer : 23.03%
vs baghdatis : 16.85%
vs A.zverev : 20.19%
vs monfils : 15.15%
vs raonic : 30.19%
vs dimitrov : 20%
vs federer : 18.68%
And all the Fedal meetings at AO
2009 final
Nadal 19.60%
Federer 19.88%
2012 SF
Nadal 17.75%
Federer 11.96%
2014 SF
Nadal 15.71%
Federer 5.76%
2017 final
Federer 22.49%
Nadal 18.69%
@abmk, what was your take on the court speed this year? I heard some about the court being faster, but I didn't get to see as much of this AO as I would have liked.
Federer had the highest winning percentage for five years consecutive I believe with just over 90% with Djokovic fractionally below 90%.
For Games Won Percentage in their best years Federer and Nadal were in the 61% range and Djokovic also around there. Anything over 60% for Games Won Percentage is amazing. Laver for example in his Grand Slam year of 1969 wasn't at 60% for example.
Guess I would go with Federer.
No doubt that a resistance rating would be very useful. Not sure how to do it. Would a person figure out the average ranking of the players? Would it be based on winning percentage of opponents? I remember Allen Barra did something like that in one of his NFL books using his computer to not only check the competition but who the competition faced.Would be good to have a "resistance" or "competition" rating to further dig into the worth of these numbers in more absolute terms (within the "era") which could be achieved by looking into percentages of games and points won by a greater number of players in a season, as well the specific match-ups encountered by the best players during the season. I know that tennis28.com has some basic stats on % of games won at the Slams by the best five players of the year according to the metric.
Is Federer's ~90% better, worse or as good as Djokovic's ~90%. I know that Federer once had a ridiculous streak of winning finals and beating top-10 players, and that Djokovic holds records (?) for most top-10 players defeated in a season.
No doubt that a resistance rating would be very useful. Not sure how to do it. Would a person figure out the average ranking of the players? Would it be based on winning percentage of opponents? I remember Allen Berra did something like that in one of his NFL books using his computer to not only check the competition but who the competition faced.
So we'd do the same and check the competition and who the competition faced and grade the competition by percentage of points or games won and see how far above the norm the top players are from say the top-20 and the top-50 and see how reliably players can beat a certain threshold of proficiency without faltering. So, maybe when Federer has to play opponents on hard-courts who win 57% of their games he has a 76% chance of winning over his best 3-year span, and we'd compare that to Djokovic. I'm kind of hazy on this because I'm trying to work it out as I go along but hopefully that makes some sense. We could come to some understandings by looking at the quality of opponent faced by looking at a) their typical form on a surface as expressed by percentage of games won for the year on that surface and b) the form displayed for just the tournament in question up to that round.. and somehow weigh these components.
@pc1
https://tt.tennis-warehouse.com/ind...tio-of-fedalovic-qf-opponents-onwards.534658/
I have something on this. It's not fully up to date. It also doesn't take into account the ranking of the players beaten by the QF onwards players. A lower ranked player that comes through several seeds and top 10 players will likely have a lower D/R than a top ranked player who goes through a 'normal' draw.
I remember this thread. Criminal how underappreciated it was in that part of the forum.. lost in a torrent of crap. Will look again.
I didn't even see the thread. Cannot believe I didn't notice it. It's a great thread but it is a thread for stat geeks like me. It may not be for all.@pc1
https://tt.tennis-warehouse.com/ind...tio-of-fedalovic-qf-opponents-onwards.534658/
I have something on this. It's not fully up to date. It also doesn't take into account the ranking of the players beaten by the QF onwards players. A lower ranked player that comes through several seeds and top 10 players will likely have a lower D/R than a top ranked player who goes through a 'normal' draw.
A little while back I started compiling full stats for all slam wins from the big 3. So that includes first serve percentage, serve speeds, winners, errors, forced errors, total points won etc...I did this for every match the slam champion played AND for every match their opponent played. So for example I've captured that same data from all 7 of Roddick's matches at Wimbledon in 2004 and all 6 of Grosjeans etc...
It ended up being a lot of work, I only finished Wimbledon. I did AM's as well.
But yes that thread of mine was underappreciated
I didn't even see the thread. Cannot believe I didn't notice it. It's a great thread but it is a thread for stat geeks like me. It may not be for all.
It sounds great. It's similar to my Games Won percentages which I believe may be a truer indicator of Tennis Strength given relatively similar competition.
While I can find the numbers for GW% it is hard to separate the numbers for the past players for the dominance ratio. For example we know 1984 McEnroe had a Games Won Percentage of 65.32 but we cannot know what percentage of his service games did he win and what percentage of his return games did he win. We did know that with that GW% he was 82-3 for the year which is about right.
D/R may actually be more accurate than GW% if I had to guess.
I didn't even see the thread. Cannot believe I didn't notice it. It's a great thread but it is a thread for stat geeks like me. It may not be for all.
It sounds great. It's similar to my Games Won percentages which I believe may be a truer indicator of Tennis Strength given relatively similar competition.
While I can find the numbers for GW% it is hard to separate the numbers for the past players for the dominance ratio. For example we know 1984 McEnroe had a Games Won Percentage of 65.32 but we cannot know what percentage of his service games did he win and what percentage of his return games did he win. We did know that with that GW% he was 82-3 for the year which is about right.
D/R may actually be more accurate than GW% if I had to guess.
Ultimately some years are going to produce varying amounts of great performers. Some years might have particularly strong competition where 5 or 6 of the quarterfinalists are playing great ball, but ultimately there can only be one winner. Other years or Slams might be quite weak and only have a couple of strong performers. Some years might have 5 or 6 strong performers but have 1 particularly anomalous super performer who is a rare next level up for the event. In Slams where a player has won it by being unusually brilliant against a draw that yielded many strong performers, the peak would be better proved especially if that player had to dispose of a couple of those strong performers along the way.
Clearly there are some useful candidates to consider here, namely: AM, DR, GW%.
Which is best?
Why?
Find out in the next episode of Dragon Ball Z.
@NatF I think D/R is a fabulous stat. I have to use it to see how it works with current players.
What if we're vigilant though and record GW% as follows: We could come to some understandings by looking at the quality of opponent faced by looking at a) their typical form on a surface as expressed by percentage of games won over their best period on that surface and b) form on a surface as expressed by percentage of games won over a more recent period and c) the form displayed for just the tournament in question up to that round.. and somehow weigh these components.
Overall I think DR could be best because it discriminates the least against big servers??? Unsure.
@NatF
Didn't you produce stats on the strength of recent tennis seasons regarding the Slam events, or was that someone else? I remember it compared competition across a span of something like 2003-2014 or so.
It's hard to do a PER stat in tennis because I think in tennis there are fewer stats to measure. I think you would have to really have a stat keeper at every match which is very tough because of the amount of players in the ATP. Perhaps there could have % of forehands hit. Winners on forehand and backhand. Serve winners and % etc.I'm wondering if some sort of Performance Efficiency Rating equivalent could be used in tennis, as is used in basketball.
I guess it doesn't work nearly as well because match-ups aren't guaranteed.
Is there anything in particular you want to test?
The problem with D/R is that
1) It tends to favour players who win a lot on serve
2) It's also not necessarily fair to those players that coast on return because of their confidence in the serve.Player This probably evens out over time.
I don't think 1) is so much of a problem, it's more important to win on serve than return.
See above what I wrote to pc1.
For your main paragraph we could infact come up with our own measurement weighing all those factors. The issue is how do we subjectively weigh each of those 3 metrics?
I want to see how it relates to winning percentage first and foremost but how the ratio translates against players of better or lesser D/R ratios.Is there anything in particular you want to test?
The problem with D/R is that
1) It tends to favour players who win a lot on serve
2) It's also not necessarily fair to those players that coast on return because of their confidence in the serve.Player This probably evens out over time.
I don't think 1) is so much of a problem, it's more important to win on serve than return.
See above what I wrote to pc1.
For your main paragraph we could infact come up with our own measurement weighing all those factors. The issue is how do we subjectively weigh each of those 3 metrics?
From what I've seen Gary is incorrect.The problem I'm seeing is that there are inherent biases in all three measures. GW% runs into the same problem as D/R. I don't know if it's more or less fair than D/R to players who win a lot on serve. @Gary Duane talks about how being 90-30 on serve-return is not the same as being 80-40, even though they reach the same total.
I want to see how it relates to winning percentage first and foremost but how the ratio translates against players of better or lesser D/R ratios.
The problem I'm seeing is that there are inherent biases in all three measures. GW% runs into the same problem as D/R. I don't know if it's more or less fair than D/R to players who win a lot on serve. @Gary Duane talks about how being 90-30 on serve-return is not the same as being 80-40, even though they reach the same total.
From what I've seen Gary is incorrect.
Yes in the D/R ratio it would be changed but I think it's not a problem in GW%. The winning percentages seem close in either case from what I've seen.https://tt.tennis-warehouse.com/ind...-strongly-predicts-winning-percentage.461224/
A thread on just that. How familiar are you with tennisabstract? I would suggest using that to test it yourself.
Well 90-30 would be 30/10 = 3 and 80/40 = 2 so yes there is a marked difference. I do think the serve is the more important shot than the return so I do think if it leans one way it should be towards the serve. At the end of the day you have control of your serve so it's a better measure of how well you're playing than if an opponent hits an ace or a DF.
That's a great question. I have to think about that. Unfortunately I have some important work to do on my most hated day...Monday.Do you think a player who is brilliant on return and rallies is just as likely to coast in bothering to produce incisive and potent first serves as a great server is likely to coast some return games? Also, do you think a more return heavy style is just as likely to dominate in the biggest matches against the best opponents as one who is serve heavy? I know that in the past you have questioned out of the legends who had a better serve and who had a better return, and that you felt it was quite evenly split. Obviously, there's more to the serve and return game than just the serve and just the return.