Random Charts & Nonsense

Oakland2002 · All-time great

Random Charts & Nonsense

Coup Droit · Tennis legend

All European ratings are based on a set formula depending on the ranking of the person you beat, and some countries have negative points if you lose to someone ranked worse than you. The rankings are done in ranges so, effectively, if you are ranked 400 and you beat someone 300-400 you get a times one bonus. If you beat someone ranked 200-300 you get a times two bonus, 150-200 a time three. (And it works in reverse for the negatives). I think a bonus system adds a nice touch while not messing up the basic structure.

Peter too · Tennis legend

Risers and fallers since 25/12/2017 ordered by percentage rise.

Name	Order	%	2017	Latest	Change	Best	DOB
Gabriella Taylor	1	47%	332	178	154	178	Mar-98
Harriet Dart	2	19%	532	433	99	238	Jul-96
Emily Appleton	3	16%	733	617	116	617	Sep-99
Maia Lumsden	4	14%	524	452	72	420	Jan-98
Katy Dunne	5	12%	267	236	31	236	Feb-95
Laura Robson	6	8%	229	211	18	27	Jan-94
Eden Silva	7	7%	501	465	36	464	Mar-96
Heather Watson	8	4%	75	72	3	38	May-92
Ella Taylor	9	4%	1185	1139	46	1135	Jun-95
Alicia Barnett	10	4%	682	658	24	658	Oct-93
Suzy Larkin	11	2%	746	732	14	627	Jun-92
Freya Christie	12	1%	542	536	6	286	Nov-97
Samantha Murray	13	1%	422	418	4	165	Oct-87
Lara Deigman	14	1%	1122	1113	9	629	May-93
Mirabelle Njoze	15	1%	1018	1010	8	599	Feb-97
Olivia Nicholls	16	1%	1022	1014	8	943	Oct-94
Tiffany William	17	1%	1038	1030	8	1030	Jul-94
Aleksandra Pitak	18	1%	1223	1214	9	1208	Apr-01
Emily Arbuthnot	19	1%	718	713	5	551	Oct-97
Emma Hurst	20	1%	1185	1179	6	870	Jun-96
Alice Gillan	21	0%	1086	1085	1	1037	Aug-98
Sarah Beth Grey	22	0%	602	603	-1	587	Aug-95
Francesca Jones	23	-1%	681	690	-9	679	Sep-00
Jodie Burrage	24	-2%	537	546	-9	537	May-99
Laura Sainsbury	25	-2%	1034	1056	-22	818	Oct-96
Katie Swan	26	-2%	300	307	-7	289	Mar-99
Manisha Foster	27	-5%	1101	1153	-52	635	Aug-93
Emily Webley-Smith	28	-12%	541	607	-66	240	Jul-84
Naomi Broady	29	-16%	116	134	-18	76	Feb-90
Amanda Carreras	30	-24%	441	545	-104	236	May-90
Tara Moore	31	-34%	351	470	-119	145	Aug-92
Johanna Konta	32	-63%	9	14	-5	4	May-91

indiana · Tennis legend

Thanks, Peter, interesting.

Just a couple of things. Harriet's ranking figures are wrong ( though her best is fine ) and no Katie B ?

-- Edited by indiana on Sunday 25th of March 2018 03:50:06 PM

blob · All-time great

Peter

Good contribution.

I couldn't figure out a way to chart-ify this sort of table.
It also seemed weird to have Jo last given the small incremental change at the top end has massive impact. I tried to normalise this over some sort of curve function, but couldn't do it because my math only goes about as far as Sesame Street teaches you.
I also can't think in this instance, and all others, how do you deal with UNR players? I give them 9999 which works for handing them, but extends scales out a long way, and makes proportional calculations like this very misleading. So, generally I have just excluded them for ease and laziness

but they should be counted somehow.

I'd like to do it on a rolling 52 week basis.

blob · All-time great

I'll look at the additional quantitative scoring system over the coming days.
I'm still trying to get the National Strength ranking factored by age of player working

Peter too · Tennis legend

I hope this is OK now. cry

Adding Katie Boulter made the post too long so I had to remove the "order" column.

Gabriella Taylor	47%	332	178	154	178	Mar-98
Harriet Dart	22%	306	238	68	238	Jul-96
Emily Appleton	16%	733	617	116	617	Sep-99
Maia Lumsden	14%	524	452	72	420	Jan-98
Katy Dunne	12%	267	236	31	236	Feb-95
Laura Robson	8%	229	211	18	27	Jan-94
Eden Silva	7%	501	465	36	464	Mar-96
Heather Watson	4%	75	72	3	38	May-92
Ella Taylor	4%	1185	1139	46	1135	Jun-95
Alicia Barnett	4%	682	658	24	658	Oct-93
Suzy Larkin	2%	746	732	14	627	Jun-92
Freya Christie	1%	542	536	6	286	Nov-97
Samantha Murray	1%	422	418	4	165	Oct-87
Lara Deigman	1%	1122	1113	9	629	May-93
Mirabelle Njoze	1%	1018	1010	8	599	Feb-97
Olivia Nicholls	1%	1022	1014	8	943	Oct-94
Tiffany William	1%	1038	1030	8	1030	Jul-94
Aleksandra Pitak	1%	1223	1214	9	1208	Apr-01
Emily Arbuthnot	1%	718	713	5	551	Oct-97
Emma Hurst	1%	1185	1179	6	870	Jun-96
Alice Gillan	0%	1086	1085	1	1037	Aug-98
Sarah Beth Grey	0%	602	603	-1	587	Aug-95
Francesca Jones	-1%	681	690	-9	679	Sep-00
Jodie Burrage	-2%	537	546	-9	537	May-99
Laura Sainsbury	-2%	1034	1056	-22	818	Oct-96
Katie Swan	-2%	300	307	-7	289	Mar-99
Katie Boulter	-4%	200	208	-8	189	Aug-96
Manisha Foster	-5%	1101	1153	-52	635	Aug-93
Emily Webley-Smith	-12%	541	607	-66	240	Jul-84
Naomi Broady	-16%	116	134	-18	76	Feb-90
Amanda Carreras	-24%	441	545	-104	236	May-90
Tara Moore	-34%	351	470	-119	145	Aug-92
Johanna Konta	-63%	9	14	-5	4	May-91

-- Edited by Peter too on Sunday 25th of March 2018 06:29:12 PM

Peter too · Tennis legend

One way to cope with new entires would be to set an initial ranking equal to one more than the lowest ranking in the WTA table.

blob · All-time great

Peter too wrote:
One way to cope with new entires would be to set an initial ranking equal to one more than the lowest ranking in the WTA table.

But the number of ranked players varies considerably over time; there is a general long term trend towards an increasing number of ranked players.
Would this not provide a false impression that beating an UNR coded as WR[last + 1] at one point in time was somehow equivalent to beating an actual ranked player of that rank 6 months down the line? If so, is that equivalency actually real? For example, at the extreme, a 14yo local WC compared to a player that demonstrably at least earned three points in separate events. I'm not sure.

Also, I think when the new tour is introduced next year, the likelihood seems that the total number of ranked players will go down (considerably?). This seems to create a lot of players as UNR with a great deal of variation between them. It's an additional complication.

I can certainly appreciate the simplicity and elegance of this solution. As long as everyone knew that was where the line was drawn it would be fine enough. But, that is not the way tennis stats tend to work in my experience. We want to deconstruct the rankings of every player (been injured, ill, otherwise not playing, junior, college, only been playing soft events etc - always a reason why a ranking isn't really apparently a 'true' ranking).

I wish I were smarter, and less prone to over-thinking and prevarication; just make the decision and plough on.

Half-Empty.

Strongbow · All-time great

You would need to use VBA to go through each row and apply a specific modifier to each rankings band. Have VBA do the hard work and automatically generate the graph or chart

blob · All-time great

Here is the chief reason why I think a metric of value predicated on seed rankings alone is problematic: Seeds don't win a third of tournaments.
The scope for this table is every ITF event played in calendar year 2017:

And heres how that means the regions compare to each other per level

So, ignoring the non-seed part of the draw leaves out an important factor determining the srength of the event.
This is why in the weekly field strength comparisons, I've started plotting the scatters of median ranks of seeds v non seeds on scatter plots to show the relative placement of each event compared to it's peers at it's level. The intersection of seeds and non-seeds is the useful marker.

The other method certainly shows a facet of it, along a consistent scale. But, is it the right scale?

blob · All-time great

Strongbow wrote:

You would need to use VBA to go through each row and apply a specific modifier to each rankings band. Have VBA do the hard work and automatically generate the graph or chart

If I knew VBA
My charts are built in such a way that they do update automatically with out any VBA or macros. I use dynamic ranges linked to tables. As soon as I paste in new data, the formulas update, and any Pivot Tables or charts linked to them also update. e.g. paste in the new weekly rankings, the 'Strongest Nation' table is ready 2 seconds later, after it re-calculates to include the new data added to the table.

Strongbow · All-time great

Getting a new "pewter" soon which will have vba - I'll share any code I do for tennis smile

blob · All-time great

We're winning more than half the matches we play
We did this up to this point in 2017, but through the clay, grass seasons, and to the end of the year, we fell well behind. Can we maintain it this year?

No GB Rank updates in the second week of Miami, as the rankings haven't changed

Raw numbers don't compare to last year well, but that's due to the extra GB even in 2017.

In terms of productivity per result/entry though, we're well up.

blob · All-time great

Week 13 Field comparisons:

2018 YTD Regional Strength per level

When there's more a larger number of usable data points, I might break this down by country too, but I suspect country variation will not be as pronounced as Regional variation.