Winning percentage minus opponent's ranking

Lew II

Lew II

G.O.A.T.
example: 2015 Djokovic won 93.18% of matches facing opponents ranked 17.8 on average ---> 93.18 - 17.8 = 75.38

> 60.0 scores:

2015 Djokovic - 75.38
2011 Djokovic - 74.81
1984 McEnroe - 74.77
2013 Nadal - 72.56
2014 Djokovic - 70.51
2013 Djokovic - 70.36
1985 Lendl - 70.11
2012 Djokovic - 68.81
2016 Murray - 68.06
2017 Federer - 67.03
2016 Djokovic - 66.44
1986 Lendl - 66.40
1979 Borg - 66.13
2006 Federer - 65.95
2005 Federer - 65.79
2015 Federer - 65.54
2007 Federer - 65.51
1980 Borg - 65.01
2012 Nadal - 64.90
2012 Federer / 2018 Nadal - 64.34
1978 Connors / 2008 Nadal - 64.17
2004 Federer - 63.40
2009 Federer - 63.36
2015 Murray - 63.23
2014 Federer - 62.98
1977 Connors - 62.60
2009 Nadal - 62.50
2019 Nadal - 62.46
1979 Connors - 62.31
2011 Federer - 62.01
1978 Borg - 61.76
2010 Federer - 61.13
2017 Nadal - 60.60
2009 Murray - 60.31
 
Last edited:
Another day of Lew II at the office, being secretly financed by the NFSI (Novak Family Service of Intelligence)


source.gif


P. S.: According to this thread Federer in 2010 with 1 Slam was better than Nadal in 2010 with 3 Slams.
 
Last edited:
Sport said:
Another day of Lew II at the office, being secretly financed by the NFSI (Novak Family Service of Intelligence)


source.gif


P. S.: According to this thread Federer in 2010 with 0 Slams was better than Nadal in 2010 with 3 Slams.
Click to expand...
Federer won the 2010 Australian Open.
 
I think it's useful to adjust results for quality of opposition, but I wonder about how useful this particular method is.

Suppose you have two players, Player A and Player B, who each played 10 matches. Player A played 10 matches against world number 1, and Player B played 10 matches against world number 20. Let's say Player A beat world number 1 8/10 times, and Player B beat world number 20 10/10 times. So we have:

Player A: 80 - 1 = 79
Player B: 100 - 20 = 80

But pretty clearly, beating the best player in the world 80 percent of the time is more impressive than beating world number 20 every time.

The issue is that playing ability as a function of ranking isn't linear (e.g., the distance between #1 and #10 is wider than the distance between #10 and #20), so perhaps subtracting average ranking doesn't appropriately account for that nonlinearity. Maybe it would be more informative to compute an adjusted score for each match that weights opponent ranking (or Elo rating, whichever seems better).
 
daggerman said:
I think it's useful to adjust results for quality of opposition, but I wonder about how useful this particular method is.

Suppose you have two players, Player A and Player B, who each played 10 matches. Player A played 10 matches against world number 1, and Player B played 10 matches against world number 20. Let's say Player A beat world number 1 8/10 times, and Player B beat world number 20 10/10 times. So we have:

Player A: 80 - 1 = 79
Player B: 100 - 20 = 80

But pretty clearly, beating the best player in the world 80 percent of the time is more impressive than beating world number 20 every time.

The issue is that playing ability as a function of ranking isn't linear (e.g., the distance between #1 and #10 is wider than the distance between #10 and #20), so perhaps subtracting average ranking doesn't appropriately account for that nonlinearity. Maybe it would be more informative to compute an adjusted score for each match that weights opponent ranking (or Elo rating, whichever seems better).
Click to expand...
There is no official ELO rating in tennis, so the ATP official ranking is always better.

For instance, according to the unofficial ELO rating of the pro-Djokovic webpage Ultimate Tennis Statistics, as of November 2019 Del Potro has a higher ELO rating than Zverev

Ultimate Tennis Statistics - Elo Ratings

Weekly tennis Elo ratings
www.ultimatetennisstatistics.com

On the other hand, according to the official ATP ranking, Zverev is #6 in the world while Del Potro is outside the top 50.

What do you think is more reliable? There is no way Del Potro is right now in better shape than Zverev.
 
Last edited:
Sport said:
There is no official ELO rating in tennis, so the ATP official ranking is always better.
Click to expand...

It may be true that official ranking is better (I'm not sure), but I don't think the fact that there is no "official" Elo is a good reason to prefer ranking. As long as there existed an Elo that was methodologically sound and predictive, it would seem like fair game to me.
 
Sport said:
P. S.: According to this thread Federer in 2010 with 0 Slams was better than Nadal in 2010 with 3 Slams.
Click to expand...

> 60.0 score in Grand Slams (2000s only):

2015 Djokovic 76.94
2011 Djokovic 76.55
2007 Federer 75.20
2013 Nadal 74.33
2009 Federer 69.76
2006 Federer 69.53
2013 Djokovic 69.19
2010 Nadal 67.55
2012 Djokovic 67.49
2017 Federer 67.44
2016 Murray 66.66
2012 Murray / 2016 Djokovic 66.30
2005 Federer 64.81
2014 Djokovic 64.80
2011 Nadal 64.46
2015 Wawrinka 64.10
2013 Murray 64.07
2007 Nadal 63.86
2017 Nadal 63.70
2004 Federer 63.65
2012 Nadal 63.40
2009 Del Potro 62.70
2011 Federer 62.63
2019 Nadal 62.01
2008 Nadal 61.91
2015 Federer 61.42
2012 Federer 61.36
2009 Nadal 60.94

2010 Federer with 59.16 doesn't make the list.
 
Last edited:
Sport said:
There is no official ELO rating in tennis, so the ATP official ranking is always better.

For instance, according to the unofficial ELO rating of the pro-Djokovic webpage Ultimate Tennis Statistics, as of November 2019 Murray has a higher ELO rating than Tsitsipas.

Ultimate Tennis Statistics - Elo Ratings

Weekly tennis Elo ratings
www.ultimatetennisstatistics.com

On the other hand, according to the official ATP ranking, Tsitsipas is #7 in the world while Murray is outside the top 100.

What do you think is mroe reliable? There is no way Murray is right now in better shape than Tsitsipas.
Click to expand...

Change Elo weightings to adjust for recent form, then. Or penalize periods of inactivity. This isn't an intractable problem.

I get your point, though. It's true that vanilla Elo has some flaws.
 
daggerman said:
Change Elo weightings to adjust for recent form, then. Or penalize periods of inactivity. This isn't an intractable problem.

I get your point, though. It's true that vanilla Elo has some flaws.
Click to expand...
That's not my point. Only the official ranking should be used, and so only the ATP rnaking is reliable. Unofficial ratings created by fans are not reliable.
 
daggerman said:
I think it's useful to adjust results for quality of opposition, but I wonder about how useful this particular method is.

Suppose you have two players, Player A and Player B, who each played 10 matches. Player A played 10 matches against world number 1, and Player B played 10 matches against world number 20. Let's say Player A beat world number 1 8/10 times, and Player B beat world number 20 10/10 times. So we have:

Player A: 80 - 1 = 79
Player B: 100 - 20 = 80

But pretty clearly, beating the best player in the world 80 percent of the time is more impressive than beating world number 20 every time.

The issue is that playing ability as a function of ranking isn't linear (e.g., the distance between #1 and #10 is wider than the distance between #10 and #20), so perhaps subtracting average ranking doesn't appropriately account for that nonlinearity. Maybe it would be more informative to compute an adjusted score for each match that weights opponent ranking (or Elo rating, whichever seems better).
Click to expand...
The average I used is the geometric mean, which gives more weight to small numbers.

For example between 1 and 10 the arithmetic mean is 5.5 while the geometric mean is 3.16.
 
Lew II said:
example: 2015 Djokovic won 93.18% of matches facing opponents ranked 17.8 on average ---> 93.18 - 17.8 = 75.38

> 60.0 scores:

2015 Djokovic - 75.38
2011 Djokovic - 74.81
1984 McEnroe - 74.77
2013 Nadal - 72.56
2014 Djokovic - 70.51
2013 Djokovic - 70.36
1985 Lendl - 70.11
2012 Djokovic - 68.81
2016 Murray - 68.06
2017 Federer - 67.03
2016 Djokovic - 66.44
1986 Lendl - 66.40
1979 Borg - 66.13
2006 Federer - 65.95
2005 Federer - 65.79
2015 Federer - 65.54
2007 Federer - 65.51
1980 Borg - 65.01
2012 Nadal - 64.90
2012 Federer / 2018 Nadal - 64.34
1978 Connors / 2008 Nadal - 64.17
2004 Federer - 63.40
2009 Federer - 63.36
2015 Murray - 63.23
2014 Federer - 62.98
1977 Connors - 62.60
2009 Nadal - 62.50
1979 Connors - 62.31
2011 Federer - 62.01
1978 Borg - 61.76
2010 Federer - 61.13
2017 Nadal - 60.60
2009 Murray - 60.31
2019 Nadal - 60.18
Click to expand...
lol, fedr sucks :D
(didn't read... but got it right, didn't i?) :giggle:
 
Sport said:
That's not my point. Only the official ranking should be used, and so only the ATP rnaking is reliable. Unofficial ratings created by fans are not reliable.
Click to expand...

Well I don't think that's an especially compelling point. If somebody created a reliable Elo, it's by definition reliable.
 
daggerman said:
Well I don't think that's an especially compelling point. If somebody created a reliable Elo, it's by definition reliable.
Click to expand...
Nope, there is no such thing as an unoffical reliable rating in a sport which already has an official rating system.
 
In Grand Slams:

2015 Djokovic 76.94
2011 Djokovic 76.55
2007 Federer 75.20
1978 Borg 74.94
2013 Nadal 74.33
1978 Connors 73.66
1984 McEnroe 71.54
2009 Federer 69.76
2006 Federer 69.53
2013 Djokovic 69.19
1977 Borg 68.41
2010 Nadal 67.55
2012 Djokovic 67.49
1988 Wilander 67.45
2017 Federer 67.44
1981 McEnroe 66.74
 
What shall we do with our percentages today, class? I know, let's subtract an average value for a disconnected metric so that another illusory feather in djokovic's cap can be perceived to coexist with those buried under months of statistical garbage that have regularly delved unto similarly flattering statistics using the metrics separately anyway.
 
Last edited:
Lew II said:
example: 2015 Djokovic won 93.18% of matches facing opponents ranked 17.8 on average ---> 93.18 - 17.8 = 75.38

> 60.0 scores:

2015 Djokovic - 75.38
2011 Djokovic - 74.81
1984 McEnroe - 74.77
2013 Nadal - 72.56
2014 Djokovic - 70.51
2013 Djokovic - 70.36
1985 Lendl - 70.11
2012 Djokovic - 68.81
2016 Murray - 68.06
2017 Federer - 67.03
2016 Djokovic - 66.44
1986 Lendl - 66.40
1979 Borg - 66.13
2006 Federer - 65.95
2005 Federer - 65.79
2015 Federer - 65.54
2007 Federer - 65.51
1980 Borg - 65.01
2012 Nadal - 64.90
2012 Federer / 2018 Nadal - 64.34
1978 Connors / 2008 Nadal - 64.17
2004 Federer - 63.40
2009 Federer - 63.36
2015 Murray - 63.23
2014 Federer - 62.98
1977 Connors - 62.60
2009 Nadal - 62.50
1979 Connors - 62.31
2011 Federer - 62.01
1978 Borg - 61.76
2010 Federer - 61.13
2017 Nadal - 60.60
2009 Murray - 60.31
2019 Nadal - 60.18
Click to expand...

I think that this is great. It is very elegant way to combine success with the level of opposition. Contact UTS, discuss with them and see whether this can be included into their statistics as Lew II index (or index that can use your real name). I am serious.
 
Lew II said:
example: 2015 Djokovic won 93.18% of matches facing opponents ranked 17.8 on average ---> 93.18 - 17.8 = 75.38

> 60.0 scores:

2015 Djokovic - 75.38
2011 Djokovic - 74.81
1984 McEnroe - 74.77
2013 Nadal - 72.56
2014 Djokovic - 70.51
2013 Djokovic - 70.36
1985 Lendl - 70.11
2012 Djokovic - 68.81
2016 Murray - 68.06
2017 Federer - 67.03
2016 Djokovic - 66.44
1986 Lendl - 66.40
1979 Borg - 66.13
2006 Federer - 65.95
2005 Federer - 65.79
2015 Federer - 65.54
2007 Federer - 65.51
1980 Borg - 65.01
2012 Nadal - 64.90
2012 Federer / 2018 Nadal - 64.34
1978 Connors / 2008 Nadal - 64.17
2004 Federer - 63.40
2009 Federer - 63.36
2015 Murray - 63.23
2014 Federer - 62.98
1977 Connors - 62.60
2009 Nadal - 62.50
1979 Connors - 62.31
2011 Federer - 62.01
1978 Borg - 61.76
2010 Federer - 61.13
2017 Nadal - 60.60
2009 Murray - 60.31
2019 Nadal - 60.18
Click to expand...
Murray>Federer

Interesting
 
I suppose this gives some view on an average of the two stats, but that means little. It just shows who's near the top of both measures.
 
TripleATeam said:
I suppose this gives some view on an average of the two stats, but that means little. It just shows who's near the top of both measures.
Click to expand...

I disagree. I think that Lew II has found a great way to describe success in relation to opposition. Attempt with ELO is also worthwhile.
 
Lew II said:
Everyone can combine stats and make new interesting stats. ATP did that too. Search for the serve, return and under pressure ratings.
Click to expand...

Lol OK. I mean, sure anyone can combine stats but it doesn't make it meaningful or insightful. For starters you've got a 1:1 correlation between the two...

But anyway I've said my bit. I've given this more time than it deserves.
 
Monfils, Fognini , Berretini in top 10.
Fringe players of prior era perched in top 4 along with players like Raonic, Nishikori.

Why wouldn't this stat be great in this decade ???
 
Mainad said:
He's a stats nerd like me only much much better. Will always have a soft spot for a good old fashioned stats nerd. :)
Click to expand...

Then you have to absolutely suck in stats.

I mean, has the guy ever posted stats where Djokovic is not the number one?

Simple plan called "Cherry Picking":
1) use a metric
2) is Djokovic the number one according to that metric? If so, post, else delete.

You have to admire his commitment though, but to somehow say he is a good stats nerd is ridiculous based on what I see.
 
2011-16 Djokovic 71.12
2008-13 Nadal 63.85
2004-09 Federer 63.36

Scary peak Nole. He brought tennis to a new level.
 
You must log in or register to reply here.

Similar threads

MichelaDe
Seasons with 20+ wins over top10s: Djokovic more than Federer, Nadal, Sampras and Borg combined
Replies
35
Views
1K
TearTheRoofOff
TearTheRoofOff
Lew II
Most/least frequent Slam combo
Replies
8
Views
612
Lew II
Lew II
Leandro2045
Most seasons with multiple slam finals in Open era history.
Replies
20
Views
642
RS
RS
Lew II
Years with >2600/2500/2400 Elo ratings
2
Replies
70
Views
3K
TMF
T
Lew II
Most Grand Slam wins over Elo >2400/2200 in a year
Replies
6
Views
475
Breakpointerer
Breakpointerer
Back
Top