31 Jul 2004 18:20:25
Hops
Toronto SF rankings

Final is rematch of last years' SF, the only time Roddick has defeated
Federer. re: the gap between #2 and #3 - it is amazing, wg. I have
rankings going back ten years, this is the largest it's ever been:

Roddick 5185 over Coria 3195 gap 1990 62%

previous high

2/17/03 Agassi 4495 over Ferrero 3070 gap 1425 46%


1 (1) 5955 Federer, Roger*
2 (2) 5185 Roddick, Andy*
3 (3) 3195 Coria, Guillermo
4 (4) 2625 Moya, Carlos
5 (6) 2465 Nalbandian, David
6 (5) 2340 Henman, Tim
7 (7) 2330 Ferrero, Juan Carlos
8 (8) 2095 Schuettler, Rainer
9 (9) 1995 Gaudio, Gaston
10 (10) 1945 Hewitt, Lleyton
11 (11) 1890 Agassi, Andre
12 (12) 1630 Grosjean, Sebastien
13 (13) 1575 Massu, Nicolas
14 (14) 1430 Safin, Marat
15 (15) 1385 Srichaphan, Paradorn
16 (17) 1325 Chela, Juan Ignacio
17 (16) 1319 Pavel, Andrei
18 (25) 1250 Kiefer, Nicolas
19 (19) 1220 Gonzalez, Fernando
20 (18) 1210 Fish, Mardy
21 (23) 1200 Kuerten, Gustavo
22 (21) 1170 Spadea, Vincent
23 (20) 1168 Hrbaty, Dominik
24 (22) 1165 Novak, Jiri
25 (24) 1120 Schalken, Sjeng
26 (26) 1110 Ljubicic, Ivan
27 (27) 1075 Robredo, Tommy
28 (28) 1061 Ancic, Mario
29 (29) 1050 Dent, Taylor
30 (30) 1035 Lopez, Feliciano
31 (33) 1014 Johansson, Joachim
32 (31) 984 Canas, Guillermo
33 (34) 925 Mirnyi, Max
34 (32) 911 Verdasco, Fernando
35 (36) 878 Arazi, Hicham




31 Jul 2004 20:39:36
Robert B. Waltz
Re: Toronto SF rankings

In article <IrqdnY66K4b82JHcRVn-pg@giganews.com >,
"Hops" <kev8128nossssspam@yahoo.com > wrote:

> Final is rematch of last years' SF, the only time Roddick has defeated
> Federer. re: the gap between #2 and #3 - it is amazing, wg. I have
> rankings going back ten years, this is the largest it's ever been:
>
> Roddick 5185 over Coria 3195 gap 1990 62%
>
> previous high
>
> 2/17/03 Agassi 4495 over Ferrero 3070 gap 1425 46%

Although we should note that a little of that gap is fluke,
because of the presence of an extra Masters Series for
Roddick. Come two weeks from now, it will probably be a
little closer to the Agassi/Ferrero number. Of course,
that's still at the high end of the scale. But even *that*
is unlikely to last.

The other side of it is the strong clumping of the
players from #4 to #7, who could, I believe, have
ended in *any* order after this week. And Coria is
wearing back toward that group. It's not so much.
that #1 and #2 are vastly exceeding the average
(though their results are, I think, better than the
average), but that the mid-level guys are so close
together.

Another point: The data from before Required and Optional
isn't really relevant. Prior to that, of course, players
could only count 14 events, which meant that their maximum
totals were about 800 points less than they are now. Of
course, no one ever equalled the maximum -- but the
principle does stand, and it will be precisely the top
players who will be most likely to have big points in
tournaments #15-#18.

--
A man can always maintain his convictions in the face of apparently
hostile evidence if he is prepared to make the necessary ad hoc
assumptions. But... a proposition whose validity we are resolved
to maintain in the face of any experience is not a hypothesis at all,
but a definition. -- A. J. Ayer


31 Jul 2004 19:06:17
Hops
Re: Toronto SF rankings


"Robert B. Waltz" <waltzmn@skypoint.com > wrote in message
news:waltzmn-7295BF.20393631072004@corp.supernews.com...
> In article <IrqdnY66K4b82JHcRVn-pg@giganews.com>,
> "Hops" <kev8128nossssspam@yahoo.com> wrote:
>
> > Final is rematch of last years' SF, the only time Roddick has defeated
> > Federer. re: the gap between #2 and #3 - it is amazing, wg. I have
> > rankings going back ten years, this is the largest it's ever been:
> >
> > Roddick 5185 over Coria 3195 gap 1990 62%
> >
> > previous high
> >
> > 2/17/03 Agassi 4495 over Ferrero 3070 gap 1425 46%
>
> Although we should note that a little of that gap is fluke,
> because of the presence of an extra Masters Series for
> Roddick.


Yes.


Come two weeks from now, it will probably be a
> little closer to the Agassi/Ferrero number. Of course,
> that's still at the high end of the scale. But even *that*
> is unlikely to last.

extreme data points, by defintion, usually do not last. : )


> The other side of it is the strong clumping of the
> players from #4 to #7, who could, I believe, have
> ended in *any* order after this week. And Coria is
> wearing back toward that group. It's not so much.
> that #1 and #2 are vastly exceeding the average
> (though their results are, I think, better than the
> average), but that the mid-level guys are so close
> together.


It's Federer and Roddick. Since they switched systems in 2000:

avg. score of #1/#2: 4163
Federer/Roddick: 5645 (after tomorrow)

avg. score of #4-#7: 2640
Moya/Nalb./Henman/JCF: 2440





> Another point: The data from before Required and Optional
> isn't really relevant. Prior to that, of course, players
> could only count 14 events, which meant that their maximum
> totals were about 800 points less than they are now. Of
> course, no one ever equalled the maximum -- but the
> principle does stand, and it will be precisely the top
> players who will be most likely to have big points in
> tournaments #15-#18.


fwiw, results 1995-1999:

avg. score of #1/#2: 4141

avg. score of #4-#7: 2820

(unsure if best 14 applied for all those years)







31 Jul 2004 22:01:43
Robert B. Waltz
Re: Toronto SF rankings

"Hops" <kev8128nossssspam@yahoo.com > wrote:

[ ... ]

> Come two weeks from now, it will probably be a
> > little closer to the Agassi/Ferrero number. Of course,
> > that's still at the high end of the scale. But even *that*
> > is unlikely to last.
>
> extreme data points, by defintion, usually do not last. : )

Of course not. But there are two ways an extreme can fail
to last: It can regress toward the mean, or something even
more extreme can come along. :-)

> > The other side of it is the strong clumping of the
> > players from #4 to #7, who could, I believe, have
> > ended in *any* order after this week. And Coria is
> > wearing back toward that group. It's not so much.
> > that #1 and #2 are vastly exceeding the average
> > (though their results are, I think, better than the
> > average), but that the mid-level guys are so close
> > together.
>
>
> It's Federer and Roddick. Since they switched systems in 2000:
>
> avg. score of #1/#2: 4163
> Federer/Roddick: 5645 (after tomorrow)
>
> avg. score of #4-#7: 2640
> Moya/Nalb./Henman/JCF: 2440

But that's not quite the same thing. You're arguing that #4-#7
are about where they usually are, or a little behind. This
I do not question. But it's not the point. What I'm talking
about is not the mean but the standard deviation. You could
have a mean of, say, 2500 by having players at 2650 2550
2450 2350 -- or by having players at 3100 2700 2300 1900.
Same average, much bigger gap from #2 to #4. All I'm pointing
out is that the gap can increase at both ends. Most of it is
Federer and Roddick moving far ahead of everyone else. But
the other part is that everyone else is falling: Coria
failed to defend three optional titles, Moya failed to defend
Umag, Henman inevitably failed to defend Washington, etc.

> > Another point: The data from before Required and Optional
> > isn't really relevant. Prior to that, of course, players
> > could only count 14 events, which meant that their maximum
> > totals were about 800 points less than they are now. Of
> > course, no one ever equalled the maximum -- but the
> > principle does stand, and it will be precisely the top
> > players who will be most likely to have big points in
> > tournaments #15-#18.
>
>
> fwiw, results 1995-1999:
>
> avg. score of #1/#2: 4141
>
> avg. score of #4-#7: 2820
>
> (unsure if best 14 applied for all those years)

Interesting. So I'm half right. :-) But I guess I'm not
surprised that #4 through #7 were stronger under Best 14 --
because they weren't dragged down by inability to count
their results at lesser events.

Somewhere in here, there is a column idea. Now I just have
to find it. Thanks for the thoughts. :-)

--
A man can always maintain his convictions in the face of apparently
hostile evidence if he is prepared to make the necessary ad hoc
assumptions. But... a proposition whose validity we are resolved
to maintain in the face of any experience is not a hypothesis at all,
but a definition. -- A. J. Ayer


31 Jul 2004 23:18:06
Michael Lockhart
Re: Toronto SF rankings

"Robert B. Waltz" <waltzmn@skypoint.com > wrote in message
news:waltzmn-1E6C57.22014331072004@corp.supernews.com...
> "Hops" <kev8128nossssspam@yahoo.com> wrote:
> [ ... ]
>
> > Come two weeks from now, it will probably be a
> > > little closer to the Agassi/Ferrero number. Of course,
> > > that's still at the high end of the scale. But even *that*
> > > is unlikely to last.
> >
> > extreme data points, by defintion, usually do not last. : )
>
> Of course not. But there are two ways an extreme can fail
> to last: It can regress toward the mean, or something even
> more extreme can come along. :-)

We know which of those is more probable, of course. :-)

> > > The other side of it is the strong clumping of the
> > > players from #4 to #7, who could, I believe, have
> > > ended in *any* order after this week. And Coria is
> > > wearing back toward that group. It's not so much.
> > > that #1 and #2 are vastly exceeding the average
> > > (though their results are, I think, better than the
> > > average), but that the mid-level guys are so close
> > > together.
> >
> >
> > It's Federer and Roddick. Since they switched systems in 2000:
> >
> > avg. score of #1/#2: 4163
> > Federer/Roddick: 5645 (after tomorrow)
> >
> > avg. score of #4-#7: 2640
> > Moya/Nalb./Henman/JCF: 2440
>
> But that's not quite the same thing. You're arguing that #4-#7
> are about where they usually are, or a little behind. This
> I do not question. But it's not the point. What I'm talking
> about is not the mean but the standard deviation. You could
> have a mean of, say, 2500 by having players at 2650 2550
> 2450 2350 -- or by having players at 3100 2700 2300 1900.
> Same average, much bigger gap from #2 to #4. All I'm pointing
> out is that the gap can increase at both ends. Most of it is
> Federer and Roddick moving far ahead of everyone else. But
> the other part is that everyone else is falling: Coria
> failed to defend three optional titles, Moya failed to defend
> Umag, Henman inevitably failed to defend Washington, etc.

You know, I read this post wrong and thought it was Hops who had written it.
I thought, "Wow, he sounds here exactly like Robert Waltz!" Then I noticed
that you were indeed the author. Funny how you can tell a person from an
internet post. So, what *is* up with the standard deviation? Is it higher
now, or lower, or the same?

Michael




31 Jul 2004 20:37:02
Hops
Re: Toronto SF rankings


"Michael Lockhart" <ml1000@bellsouth.net > wrote in message news:RgZOc.3288



> > >
> > > It's Federer and Roddick. Since they switched systems in 2000:
> > >
> > > avg. score of #1/#2: 4163
> > > Federer/Roddick: 5645 (after tomorrow)
> > >
> > > avg. score of #4-#7: 2640
> > > Moya/Nalb./Henman/JCF: 2440
> >
> > But that's not quite the same thing. You're arguing that #4-#7
> > are about where they usually are, or a little behind. This
> > I do not question. But it's not the point. What I'm talking
> > about is not the mean but the standard deviation. You could
> > have a mean of, say, 2500 by having players at 2650 2550
> > 2450 2350 -- or by having players at 3100 2700 2300 1900.
> > Same average, much bigger gap from #2 to #4. All I'm pointing
> > out is that the gap can increase at both ends. Most of it is
> > Federer and Roddick moving far ahead of everyone else. But
> > the other part is that everyone else is falling: Coria
> > failed to defend three optional titles, Moya failed to defend
> > Umag, Henman inevitably failed to defend Washington, etc.
>
> You know, I read this post wrong and thought it was Hops who had written
it.
> I thought, "Wow, he sounds here exactly like Robert Waltz!" Then I
noticed
> that you were indeed the author. Funny how you can tell a person from an
> internet post. So, what *is* up with the standard deviation? Is it
higher
> now, or lower, or the same?


I can run some numbers and throw up a few graphs, but first realize that the
top x players in tennis do not constitute a normal distribution curve - they
are the _far right hand portion_ of the curve. IOW, this part:



x |
x |
x | # of players
x |
x |
x |
-------------------------------------
Ranking Points





It is quite common for the #1 player to be 1000 points ahead of the #2
player. The #9 player is never 1000 points ahead of the #10 player.




01 Aug 2004 11:03:06
Robert B. Waltz
Re: Toronto SF rankings

"Hops" <kev8128nossssspam@yahoo.com > wrote:

[ ... ]

> > You know, I read this post wrong and thought it was Hops who had written
> it.
> > I thought, "Wow, he sounds here exactly like Robert Waltz!" Then I
> noticed
> > that you were indeed the author. Funny how you can tell a person from an
> > internet post. So, what *is* up with the standard deviation? Is it
> higher
> > now, or lower, or the same?
>
>
> I can run some numbers and throw up a few graphs, but first realize that the
> top x players in tennis do not constitute a normal distribution curve - they
> are the _far right hand portion_ of the curve. IOW, this part:
>
>
>
> x |
> x |
> x | # of players
> x |
> x |
> x |
> -------------------------------------
> Ranking Points
>
>
>
>
>
> It is quite common for the #1 player to be 1000 points ahead of the #2
> player. The #9 player is never 1000 points ahead of the #10 player.

You could perhaps handle this by plotting the log of the points.
Might be interesting. But I doubt it's worth it. (Yes, even *I*
doubt it. :-)

I guess, ultimately, the point I'm trying to make is this: There
are always a certain number of gaps in the Top Ten (or anywhere
else). An individual gap will grow or shrink, but the overall
picture doesn't change much.

If anyone is interested, we should also observe that the tennis
curve probably *won't* be truly "normal" (i.e. it won't be
a proper fraction of a bell-shaped curve), even taking into
account the fact that points are awarded discretely (you can't
earn .00143 points). The reason is economic: The top players
can make a living, #1000 can't. So you get a lot more
turnover at the lower levels.

--
A man can always maintain his convictions in the face of apparently
hostile evidence if he is prepared to make the necessary ad hoc
assumptions. But... a proposition whose validity we are resolved
to maintain in the face of any experience is not a hypothesis at all,
but a definition. -- A. J. Ayer


01 Aug 2004 21:20:28
Sakari Lund
Re: Toronto SF rankings

On Sat, 31 Jul 2004 18:20:25 -0700, "Hops"
<kev8128nossssspam@yahoo.com > wrote:

>Final is rematch of last years' SF, the only time Roddick has defeated
>Federer. re: the gap between #2 and #3 - it is amazing, wg. I have
>rankings going back ten years, this is the largest it's ever been:
>
>Roddick 5185 over Coria 3195 gap 1990 62%

Would be a lot closer if Coria managed to win one of his match points
at FO final.


01 Aug 2004 17:18:46
Hops
Re: Toronto SF rankings


"Sakari Lund" <sakari.lund@pp.inet.fi > wrote in message
news:410d5f5a.3898072@news.inet.fi...
> On Sat, 31 Jul 2004 18:20:25 -0700, "Hops"
> <kev8128nossssspam@yahoo.com> wrote:
>
> >Final is rematch of last years' SF, the only time Roddick has defeated
> >Federer. re: the gap between #2 and #3 - it is amazing, wg. I have
> >rankings going back ten years, this is the largest it's ever been:
> >
> >Roddick 5185 over Coria 3195 gap 1990 62%
>
> Would be a lot closer if Coria managed to win one of his match points
> at FO final.

Not much, 300 points. gap would still be record.