Members Login
Username 
 
Password 
    Remember Me  
Post Info TOPIC: Strongest Nation?


Tennis legend

Status: Offline
Posts: 31420
Date:
Strongest Nation?


Still very much liking the 'base' average of nations' best 10 players.

As I said before 10 does seem to me about the best number for combining strength of very best players with depth. And while yes say 9, 11 or 12 would clearly be similarly good it is natural to take the round number.

And we have anyway been given so many options to play about with things in line with our own preferences for numbers.

Yes, the very top players moving points doesn't effect things - eg clearly specifically Andy. But I am much more interested in seeing the base figures and how we progress as a nation, not really in trying to find ways to make us seem to position better.

Generally loving sporting stats, more complex weighted info may well interest me but ultimately simpler can be best and I will ultimately come back to the original straight average of rankings in following our progress and be great if that GB average can show improvement over time.

Thanks again for all the work done here.



__________________


All-time great

Status: Offline
Posts: 5568
Date:

The Bug Blatter Beast makes a good point, and provides a statistically more rigorous approach.
A raw average of ranking, especially with small N (sample size) is flawed, which is why I set the minimum to 5, not, say 2.
But, as Indiana points out, Average of ranking is at least easily understood by most people, and probably more fun and interesting to them because of that.

The thing is, that looking at the results for top N from 5 to 30, a base average actually serves pretty well.

As to the actual formula proposed, putting in the factor for weighting based on players rank seems a bit tricky.
Firstly as a subjective measure (though such factors are common necessities in statistics).
Secondly progression, worth, value, contribution, throughout the ranks is not uniform, and to get good benchmarks for the weighting would take a deal of research (especially for me, not being as au fait with the intricacies of the ATP 'pyramid' as with that on the WTA) to set usefully.
Ideally, I would love to look at every match by every player, the ranks of the players involved, the level it was played at, the round, the score etc
From that, adjudge each performance on a meaningful basis - add all that up to gague actual worth of every single ranking point earned, and show which countries players most regularly exceeded expectation on that scale as well as their actual ranking etc.
I don't have a) the access to that data or, b) the time to do that, though I wish I did, as I think it would be fun and fascinating to see a data-driven system where you could justify every nuance.

So, instead, I've used JonH's original suggestion of average ranking, allowing for users to vary the size of N as they choose.

I don't think that any one chart is going to tell the whole story, so, I've tried to view it from several different angles.
Therefore, I also added the three other ATP views, to show relative positions of the actual players, and illustrate the variable gap sizes.
I think this is very illustrative of strength to supplement the raw average:
eD4Ewcc.png

Again, you choose N yourself, and the countries, and dates concerned. This is Top 10 for the latest rankings for France, Germany, Belgium & GB.

At some point you would expect, or hope, that, all other things being equal, GB would 'catch up' to, or close the gap with, other similar countries, but this is not the case generally.

There is also something to be said in 'Strongest Nation' as to resources. Belgiums performance with their comparatively meagre population is remarkable.
So, add to that the Population data, tab, and, for example, pound for pound, Belgium punches far harder than GB, throughout all of the scales. GB also lags behind it's peers (ESP, FRA, GER, ITA etc) even at the top end, and progressively the further down the scale to look at strength in depth.

Overall from the 4 charts, I hope you get a useful picture of what is going on, and, crucially, from one data set I can download each week to just update the ranks, and show variations over time.

If someone wants to derive a meaningful weighting factor, then I can do a version using RBBOT's rigorous formula.

So many interesting approaches, and ideas. If the data were all easier to come by, I'd love to build a proper model.



__________________


Challenger level

Status: Offline
Posts: 2420
Date:

Just to leave Andy to one side, lets's see with the women...

On the current version of this (live) website...

live-tennis.eu/en/wta-live-ranking

.. our two top ten active women are Konta and Carreras. Assuming all other players stay on the same ranking points score around them, then if Konta won her current tournament, Oz Open, she would rise 8 places, and raise the GB score by 0.8; if Carreras won her current tournament, 50k qualies, she would rise 9 places, and raise the average by 0.9.

Not even Amanda's mum thinks that GB is stronger because she qualified for a 50k, as opposed to Jo winning a Grand Slam. The aggregate points score (2000 vs 5) would seem to me more relevent and informative.

__________________


All-time great

Status: Offline
Posts: 6109
Date:

First of all , massive thanks and kudos to Alibahbah for taking a little idea and turning it into some fantastically rigorous statistical analysis. Much appreciated and I will play around with a lot.

My original intent was to amuse myself and create a talking point, which I guess has been accomplished. At one level, I wasnt actually trying to find the strongest nation, rather being quite simplistic, taking the top ten, as Indy suggests, gave a reasonable proxy to both top level accomplishments and depth, whilst being easy to work out (dividing by ten) given I was using pen and paper.

I think the thing this does is gives everyone the chance to review and assess and discuss, which a discussion board is for, I guess.

 

My second thought is to the power of the resource - if I was a journo, there is some rich data here that a paper or ATP or WTA could turn into something interesting. I have mentioned previously that I have a friend who is a sportswriter for the Guardian (albeit Football) but a journo like Kevin Mitchell (one of the best tennis writers IMO currently, for the Guardian of course) may well see an interesting article in some of this data. ABB, would you have a problem if I pointed my friend, David Conn, from the Guardian to this data and suggested his colleague take a look?

 

thanks again!

 

 



__________________
JonH


All-time great

Status: Offline
Posts: 5568
Date:

By all means, I think there's some fascinating things that could be done with tennis data - in the way the sabermetricians have done with Baseball.
If only it were easier to get the data, and every match, at least ATP/WTA had full, consistent stat feed data, people around the world would come up with amazing things.
Currently, you have to be steven-level to do that; that's a very high bar for entry

__________________


All-time great

Status: Offline
Posts: 5683
Date:

Huge thanks to ABB - you are truly amazing.

Picking up on one of your points: Belgium is fascinating - especially as not only is it a small country, but its top players are also relatively small. Three of Belgium's top five are under 6' and the remaining two are 6'0. That's quite a contrast with most countries. (And when you think back to the Rochus brothers, surely the shortest players to succeed on the men's tour in the past few decades ....) Curious ... and interesting that Dan Evans, now making his breakthrough, was coached by a Belgian ...

__________________


All-time great

Status: Offline
Posts: 5568
Date:

RBBOT wrote:

[...]
edit:
A better formula is:
score% = (scale ^ -rank/slope) / (scale ^ -1/slope) * 100%
[...]
e.g. scale = 3, slope = 125 gives
#10 = 93%, #50 = 65%, #100=42%, #300=7%, #500 = 1%


I'm trying to model this.
Is the formula what you intended?
Here's my Excel formula (in yellow), and results:
wqF3Lgc.png

I've omitted the final term "*100%" because anything *100% of itself is itself.
Should this be something else?

Where:
[Scale_Val] is a named range linked to the purple cell

[Slope_Val] is a named range linked to the red-brown cell

[@Rank] is an Excel Table reference to the value in the column 'Rank' in whichever row the formula is in. e.g. for Rank 1, [@Rank] = 1; for Rank 10, [@Rank] = 10, etc.


The resultant values don't fit your intended curve.
What am I doing wrong?



__________________


All-time great

Status: Offline
Posts: 5568
Date:

Updated with rankings data of 30/01/2017: https://public.tableau.com/views/20170130ATPWTACountryStrengthsv2_3/ATPCountryRankings?:embed=y&:display_count=yes

For pure averages of ranking, based on top 10 players per country:

ATP

  • Italy move past Australia into 7th place - a big rise by Seppi, and steady improvement from the Italian #'s 5-10 all contributing.
  • GBR stay resolutely in 13th place, but improve their average ranking from 211 to 206. Improvements by Norrie & Broady the major contributors.
    • Things look unlikely to change for GBR for the time being. They have a 26 point average ranking cushion to Brazil, in 14th place, but also are 32 average ranking places behind Belgium, who are currently in 12th.

WTA

  • The exploits of Ash Barty & Destanee Aiava sees Australia leap 2 spots, to #14, and improving their average ranking by fully 20 points in the process.
  • GBR sit at #15, nudged down one by the Aussie surge; though average ranking is fractionally improved from 216 to 215.


__________________


All-time great

Status: Offline
Posts: 6109
Date:

Thanks Aliblahblah - Spain and France have interestingly drawn closer for #1 position, even though both fell in the average ranking since the pre Aus Open rankings . That is despite Nadal having his wonder tournament.

Overall, a lot of the players from Spain at the top seem to be getting a bit older, I wonder if their relative strength could drop off over this year and we could see france take top spot on this measure.

Re GB, it would be good to see the average improve to within 200, that would be a good initial benchmark.

__________________
JonH


All-time great

Status: Offline
Posts: 6109
Date:

Within the ladies, although we GB rank fell slightly, overall we are a fair bit closer to several runners and riders around the 200 mark, so maybe a chance to move up on that front. Again, hitting an average of 200 would a a good benchmark, but good that we are moving slightly forwards on mens and womens. Overall been a reasonable start to the year for GB men and women despite Andy losing early (which of course has no impact on the rankings) 



__________________
JonH


All-time great

Status: Offline
Posts: 5568
Date:

Updated: https://public.tableau.com/views/20170206ATPWTACountryStrengths/ATPCountryRankings?:embed=y&:display_count=yes

Not much change on the ATP.
France and Spain still 1 & 2 respectively.
3 & 4 USA & Argentina again switch places; a weekly occurence, it seems.
Only real movement in the top 30 nations is Belarus, who drop 3 places from 28 to 31. Interstingly, this is driven by big losses for their players ranked 7, 8 & 9 - the model reflects strength in depth.
GBR stay in thirteenth position for the fifth week in a row. Avg. rank slides one point from 206 to 207.

WTA
All very quiet. Top 13 places remain as previously; USA miles ahead of everyone in the top spot.
GBR stay in fifteenth place, as with the ATP, average ranking slips one point from 215 to 216.



__________________


All-time great

Status: Offline
Posts: 6109
Date:

Thanks Aliblahblah. Just looking more closely I hadnt spotted the large rise for the Australian's on the WTA, significant rise from start of season. My first reaction was home court advantage, but of course every year they have home court advantage so that wouldnt have changed, clearly they are making a strong move.

Interesting re Belarus and glad the model is showing how those lower movements can impact on things. Most of the GB players in the top 10 currently are playing Challenger events these days so I guess we are unlikely to see mass shifts, more a gradual movement as one or two players have good or bad weeks? Presumably for Belarus their 7-9 ranked players where in Futures and maybe one or two events in the local region came off together and impacted the overall ranking?

 



__________________
JonH


All-time great

Status: Offline
Posts: 6109
Date:

AliBlahBlah wrote:

Updated: https://public.tableau.com/views/20170206ATPWTACountryStrengths/ATPCountryRankings?:embed=y&:display_count=yes

Not much change on the ATP.
France and Spain still 1 & 2 respectively.
3 & 4 USA & Argentina again switch places; a weekly occurence, it seems.
Only real movement in the top 30 nations is Belarus, who drop 3 places from 28 to 31. Interstingly, this is driven by big losses for their players ranked 7, 8 & 9 - the model reflects strength in depth.
GBR stay in thirteenth position for the fifth week in a row. Avg. rank slides one point from 206 to 207.

WTA
All very quiet. Top 13 places remain as previously; USA miles ahead of everyone in the top spot.
GBR stay in fifteenth place, as with the ATP, average ranking slips one point from 215 to 216.


 On the WTA side, GB are just ahead of Ukraine, but if you look closely, that is almost entirely because of the difference between Laura Robson and her Ukrainian equivalent. Most of the way through, the Ukraine players are ahead of their UK equivalents.

In the ATP, France and Spain are effectively neck and neck and France appear stronger if one compares each rank in the top 10 - probably a matter of time before they take over in the lead.    



__________________
JonH


All-time great

Status: Offline
Posts: 6109
Date:

AliBlahBlah wrote:

Updated: https://public.tableau.com/views/20170206ATPWTACountryStrengths/ATPCountryRankings?:embed=y&:display_count=yes

Not much change on the ATP.
France and Spain still 1 & 2 respectively.
3 & 4 USA & Argentina again switch places; a weekly occurence, it seems.
Only real movement in the top 30 nations is Belarus, who drop 3 places from 28 to 31. Interstingly, this is driven by big losses for their players ranked 7, 8 & 9 - the model reflects strength in depth.
GBR stay in thirteenth position for the fifth week in a row. Avg. rank slides one point from 206 to 207.

WTA
All very quiet. Top 13 places remain as previously; USA miles ahead of everyone in the top spot.
GBR stay in fifteenth place, as with the ATP, average ranking slips one point from 215 to 216.


 Based on this week so far, GB Top 10 average will improve to 205, Ed Corrie contributing well to that. If he gets further, we will see a decent improvement on the week as a top 10 average



__________________
JonH


All-time great

Status: Offline
Posts: 5568
Date:

13/02 Updated: https://public.tableau.com/views/20170213ATPWTACountryStrengths/ATPCountryRankings?:embed=y&:display_count=yes

ATP
Serbia overtakes Japan to claim 9th place by a razors edge.
Otherwise all deathly quiet at the top.
There is still less than one average ranking place between Spain and France in top spot, but Spain cling on.
France are #1 nation if you make the number of players used for the average 4,5, or 12-30 though. Only for a sample of 6-11 are Spain top.

GB stay in 13th for the 6th consecutive week (i.e. since the beginning)
Avg. rank improves from 207 to 205. Evans, Edmunds, Bedene, Corrie contributing to negate falls further down.
Behind us, Brazil also improve two ranking points to 233; ahead of us, Belgium stay on 176 average.
We're not gong to move for about 6 months, exceptional circumstances aside.

WTA
Italy tie Spain to claim equal 9th place.
Australia continue to power ahead, moving up another place, to 13th, having started in 17th. No other country has moved more than one place either way.

GB stay in 15th place.
Avg. Ranking of top 10 players improves from 216 last week, to 214 now - Naomi, Mandy & Katie B the main determinants there.
One place ahead, the Belgian avg. rank stays constant at 208. One place behind us, Ukraine drop from 218 to 220 (look at the weird split in the Ukranian player distribution!)

GB stay right on the median line when all ranked players are considered. We are exactly average.



__________________
« First  <  Page 42  >   Last »  sorted by
 
Quick Reply

Please log in to post quick replies.

Tweet this page Post to Digg Post to Del.icio.us


Create your own FREE Forum
Report Abuse
Powered by ActiveBoard