Analysis of last 10 yrs of GS draws finds top two seeds had easier first rounds than statistically probable
grif
Sakari Lund
<griff...@hotmail.com> wrote:
Interesting. It sounds correct too. I remember a lot of cases like
they mention in the article, Jenkins, Britton for the top seeds at
USO. Whereas in the other slams they could have really tough first
round draws.
undecided
TT
>
> http://espn.go.com/espn/otl/story/_/id/6850893/espn-analysis-finds-top-seeds-tennis-us-open-had-easier-draw-statistically-likely
Interesting stuff. It's still 1/300 chance though...could happen.
Anything could. But combine that with how often Murray faces Nadal etc...
TT
lol at:
"Any way you want to look at these, there is significant evidence here
that these did not come from a random draw,"
+
"Only three of OTL's 1,000 simulations produced first-round opponents as
easy as those the top two men's seeds have actually faced on average
over 10 years in the U.S. Open. In none of the 1,000 simulations did OTL
get the extreme results found in 10 years of actual opening matchups for
the top two women's Open seeds."
Why only 1000 simulations though...slow computers?
Ted S.
> Interesting. It sounds correct too.
The only question I'd ask is why they did a Monte Carlo simulation.
Since they've got all the information, the statisticians should simply
have been able to calculate the actual percentage of time you'd get 20
players averaging where they did, instead of running the simulation.
--
Ted Schuerzinger
tedstennis at myrealbox dot com
If you're afraid of the ball, don't sit in the front row. --Anastasia
Rodionova
jdeluise
On 15-Aug-2011, "Ted S." <justac...@gmail.com> wrote:
> The only question I'd ask is why they did a Monte Carlo simulation.
> Since they've got all the information, the statisticians should simply
> have been able to calculate the actual percentage of time you'd get 20
> players averaging where they did, instead of running the simulation.
I'd say it's an appeal to people's belief that computers are infallible.
Superdave
i have been saying this for a long time.
Rafa had lots of help with his CHEATING.
jdeluise
On 15-Aug-2011, Superdave <DaveHa...@remail-it.net> wrote:
> i have been saying this for a long time.
>
> Rafa had lots of help with his CHEATING.
Sure sure.... but who has won the most US Open titles in the last ten years
as a top seed?
drew
> http://espn.go.com/espn/otl/story/_/id/6850893/espn-analysis-finds-to...
Interesting article.
In light of the stats for the other majors, there would appear to be
something fishy with the USO draw. When you
consider how much weight is put on ratings and how the USO tries so
hard to get the men's semi and final done on
consecutive days on the weekend (rather than doing the right thing and
keeping them 2 days apart) it doesn't take too
much imagination to consider that the organizers rig the draw to make
it very unlikely that the top seeds will lose in the
first round.
Early says it hasn't made a significant difference in the
results....and that much may be true but if they can move the
goalposts
just a little bit in favour of the best players.....without getting
caught.....why not? They know that it can only help the top players,
even
when a first round loss is unlikely.
drew
>
> Interesting stuff. It's still 1/300 chance though...could happen.
>
> Anything could. But combine that with how often Murray faces Nadal etc...
Sure it 'could' happen. But alarm bells should go off with this
finding. At the very least
it warrants further investigation and even more rigorous analysis.
My guess is that it won't
happen....I wish I'd run those stats.....I'd take the payola and
shutup.
drew
It's more the quality of the simulation. But given that the results
of the simulation as applied
to the FO and AO draws showed strong agreement with predicted results,
I'd say let's take a closer
look. Something doesn't seem right.
In Canada we found that store owners were keeping winning lottery
tickets. The probability of store
owners cashing as many tickets as they did was 'statistically
improbable'. Sure enough, it was determined
that the cashiers were keeping the winning tickets and scanning
losing tickets....pocketing the winnings.
Statistically possible yes....but very fucking unlikely usually means
something is fishy.
jdeluise
On 15-Aug-2011, drew <dr...@technologist.com> wrote:
> It's more the quality of the simulation. But given that the results
> of the simulation as applied
> to the FO and AO draws showed strong agreement with predicted results,
> I'd say let's take a closer
> look. Something doesn't seem right.
I can understand using computer simulations to model things that are
difficult to understand or predict otherwise (too many variables, variables
that aren't well understood). A statistician shouldn't have many problems
with this particular problem.
MBDunc
> On Mon, 15 Aug 2011 14:11:22 -0700 (PDT), grif
>
>
> >http://espn.go.com/espn/otl/story/_/id/6850893/espn-analysis-finds-to...
>
> Interesting. It sounds correct too. I remember a lot of cases like
> they mention in the article, Jenkins, Britton for the top seeds at
> USO. Whereas in the other slams they could have really tough first
> round draws.
It is possible that it is OK (it is statistically not that
improbable).
However USO is actually the only slam caught from fixing the draw
(1996 when they made a non-seed draw first and then hand picked seeded
players to fill the slots to ensure Sampras/Agassi to opposite halves
-> draw was then redone).
.mikko
Ted S.
> It's more the quality of the simulation. But given that the results
> of the simulation as applied to the FO and AO draws showed strong
> agreement with predicted results, I'd say let's take a closer look.
> Something doesn't seem right.
I wasn't suggesting their conclusions are wrong, just that using a Monte
Carlo simulation for proof is unneccessary.
Sadly, JDelouise is probably right about the belief in the infallibility
of computers. We've been down this road with the line-guessing system
already. The number of commentators who claim the line-guessing system
is more accurate than the marks left on clay is depressing.
Ted S.
> Early says it hasn't made a significant difference in the
> results....and that much may be true but if they can move the
> goalposts just a little bit in favour of the best players.....without
> getting caught.....why not? They know that it can only help the top
> players, even when a first round loss is unlikely.
It also theoretically gives them an easier opening-round match, which
should pay dividends later in the tournament in the form of less
fatigued players.
Superdave
I give up. Tell me..
TT
Bryan brothers?
Superdave
Those "gay guys" ?
They should join the US Army !
TT
Pelle Svanslös
>
> http://espn.go.com/espn/otl/story/_/id/6850893/espn-analysis-finds-top-seeds-tennis-us-open-had-easier-draw-statistically-likely
"USTA Pro Circuit Director Brian Earley, who has been the U.S. Open
tournament referee since 1992 and presides over the draw" says:
"What would the U.S. Open gain by fixing the draw in this way? I believe
the U.S. Open would gain nothing," Earley said. "I think that that would
be a risk that the U.S. Open would never take. Never."
Well, they fixed the draw in 1996. That is, they did the draw first and
then set the seeds. And then had to redo the draw.
What would the USO have gained in doing what they did? TV audience,
that's what. By making it as likely as possible that Sampras and Agassi
would be ther in the latter rounds.
Brian Earley sounds like a CHEAT.
kaennorsing
> On Tue, 16 Aug 2011 00:33:11 +0300, Sakari Lund wrote:
> > Interesting. It sounds correct too.
>
> The only question I'd ask is why they did a Monte Carlo simulation.
> Since they've got all the information, the statisticians should simply
> have been able to calculate the actual percentage of time you'd get 20
> players averaging where they did, instead of running the simulation.
That's what Swift did: "After being presented with the "Outside the
Lines" analysis, Swift conducted his own study of the opponents of the
top two seeds and found that only four times in 1 million simulations
did he come up with an average ranking equal to or easier than what
was actually observed in the men's and women's draws over the last 10
years."
That's pretty shocking, though not entirely unbelievable... What's
most incredible imo is to assume this unlikelihood is pure chance or a
'computer malfunction'.
kaennorsing
Computers are not infallible, but they don't cherry-pick... Just
facilitate the 'needs'.
kaennorsing
> 16.8.2011 2:01, TT kirjoitti:
>
> > 16.8.2011 0:11, grif kirjoitti:
>
>
> > Interesting stuff. It's still 1/300 chance though...could happen.
>
> > Anything could. But combine that with how often Murray faces Nadal etc...
>
> lol at:
>
> "Any way you want to look at these, there is significant evidence here
> that these did not come from a random draw,"
>
> +
>
> "Only three of OTL's 1,000 simulations produced first-round opponents as
> easy as those the top two men's seeds have actually faced on average
> over 10 years in the U.S. Open. In none of the 1,000 simulations did OTL
> get the extreme results found in 10 years of actual opening matchups for
> the top two women's Open seeds."
>
> Why only 1000 simulations though...slow computers?
Because it's a large enough sample to warrant representative numbers.
But remember that's only the men, the ladies draw was even more
unlikely... And combining those in the same organisation over the 10
years as actually happened is even more unlikely:
kaennorsing
> On Mon, 15 Aug 2011 18:02:13 -0700 (PDT), drew wrote:
> > Early says it hasn't made a significant difference in the
> > results....and that much may be true but if they can move the
> > goalposts just a little bit in favour of the best players.....without
> > getting caught.....why not? They know that it can only help the top
> > players, even when a first round loss is unlikely.
>
> It also theoretically gives them an easier opening-round match, which
> should pay dividends later in the tournament in the form of less
> fatigued players.
Of course. Earley's assumptions on how the USTA couldn't possibly
benefit and any rigging is impossible is either silly or disingenuous.
number_six
> http://espn.go.com/espn/otl/story/_/id/6850893/espn-analysis-finds-to...
In the interests of transparency, they should release the code of the
software they are using.
Since that won't happen, they should abandon it and go to a *physical*
draw in the traditional manner.
Their dilemma is, when the results of the physical draw diverge from
those produced by the computer draw, their earlier system will be
*presumed* to have been dishonest.
They never should have abandoned the physical draw in the first place.
Superdave
wrote:
they had to otherwise Rafa couldn't win anything. he can't win without cheating
no?
jdeluise
On 16-Aug-2011, kaennorsing <ljub...@hotmail.com> wrote:
> That's pretty shocking, though not entirely unbelievable... What's
> most incredible imo is to assume this unlikelihood is pure chance or a
> 'computer malfunction'.
Maybe he used a crappy random number generator. There are a lot of
potential problems when running a simulation. Seriously, what's the
point... this can't be a difficult problem?
jdeluise
On 16-Aug-2011, kaennorsing <ljub...@hotmail.com> wrote:
> Computers are not infallible, but they don't cherry-pick... Just
> facilitate the 'needs'.
Computer programs inherit the bugs of their programmers. I don't doubt the
results of the simulation, and I wouldn't be surprised if the USO was fixing
draws. However, too many people are willing to accept the output of a
computer program at face value. The worst kind of bug is a bug so subtle
that the output appears correct in most conditions but not all. In this
particular case, I'm guessing it's harder to set up and run the simulation
*accurately* than to simply run the numbers.
Ted S.
> Maybe he used a crappy random number generator.
Anybody who uses deterministic methods to obtain random numbers is, of
course, living in sin. (John von Neumann)
Whisper
> On Aug 15, 2:11 pm, grif<griffin_...@hotmail.com> wrote:
>> http://espn.go.com/espn/otl/story/_/id/6850893/espn-analysis-finds-to...
>
> In the interests of transparency, they should release the code of the
> software they are using.
>
Found it;
Case
If opponent_rank_status = 'clown' then = Federer_opponent else
Nadal_opponent
End
undecided
jdeluise
On 25-Aug-2011, undecided <cos...@gmail.com> wrote:
> oopsie, obviously you're not in IT. Lousy pseudo code but correct
> intentions.
So, you'd rather Djok be on Nadal's side?