Random Charts & Nonsense

Peter too · Tennis legend

Random Charts & Nonsense

blob · All-time great

This is a good idea.
For once, I'd actually thought of it.
The rankings run from 1-~1300 in any given week.
The compression on a log scale for that range is too much. We don't have much action 10-100, and quite a lot 1000-1300. Also it's not intuitive, or at least it wasn't to me, and I knew what I was supposed to be looking for!
Here's what it looked like

Everything gets lost.

In order to compare the relative positions of ranks 1-10 against each other, I introduced the final chart, with all ten plotted on the same chart.
As the purpose with the micro charts is to make YoY comparison (and show the detailed fluctuations) I'm OK with having the scales differ, although it seems odd, because we're lookng at each one in detail.

Getting the scales to automatically adjust usefully to each chart is tricky though: Excel makes some very strange guesses.
I know how to get around it - plot a dummy series with a colourless line that therefore won' be visible on the chart, but is nonoetheless there. That dummy series has values one multiple of 20 under, and one multiple of 20 over the bounds of anything in the 2017-18 data. Therefore, when plotted, the range of the chart as presented will scale to the dummy series (as it will have the maximum and minimum values on the chart) and everything will be scaled nicely, and the scaling will be uniform across the charts.
But, I need the formulas that Excel has removed to work that all out
Doing it the long way, calculating ranks for each part in turn, and then sorting using formulas, is really hard, and I've struggled to make it work

indiana · Tennis legend

Getting back to the question of comparative retirements in matches between Brits and Peskies, as discussed yesterday I am coming back to it in this thread.

Figures all based on blob's original data in the week 7 Perth thread and this is what I understand and then base my %s on ( please make sure I am understanding this correctly ):

1) From 1606 GB vs Pesky matches 34 were ended by a retirement. 18 of these as a result of a GB retirement and 16 as a result of a Pesky retirement.

2) From 142 GB vs GB matches, 3 of these ended with a retirement.

Now the important point here is that for the GB vs GB matches I am taking that to mean 142 separate matches, therefore clearly with 2 GB players involved 284 GB player involvements.

And it is the number of players involved, 2 for each GB vs GB match, that I think might be the best way to look at things, as distinct to the number of matches.

So anyway, Peskies first. In GB vs Pesky matches 16 retired in 1606 matches. So a retirement rate of 16/1606 per match participant, ie 1.00%.

For the Brits, in the GB vs Pesky matches 18 retired, i.e. 18/1606 = 1.12%. Plus we also have the GB vs GB data and there were 284 participations in these 142 matches with 3 retirements, i.e. a retirement per competitor of 3/284 = 1.06%. In total for the Brits there were 1606 + 284 GB player appearances and a total of 18 + 3 = 21 retirements. So overall for the Brits 21/1890 = 1.11% retirements.

1.11% for Brits as against 1.00% for Peskies in the overall rate of retirements, so Brit retirements running at 11% more, is not significant on that overall number of retirements. And if we look purely at the GB vs Pesky matches with 18 (1.12%) as against 16 (1.00%), an 18 / 16 split in these retirements is also of no real significance at all. A split one way or the other is randomly more likely than a 17 / 17 split.

Now here we are looking just at retirements, but perhaps it is the case that we overestimate the comparative rate of retirements and injuries from Brits, largely because of being so much more aware of the Brits. I know some may think they are taking account of that and still think the GB rate is greater but I do wonder if it really is to any real significance. That though is just speculation about injuries more generally.

blob · All-time great

Last point first, as it's probably more important than the rest: yes, RET's being the only number that match results can provide to show 'fitness' to take the court, or complete a match - I have w/o numbers too.
For the frst week of 2017, I tried to record observable MTO's too! Which didn't last.
So, that's what I used, to see if anything could be told from that (passim ad nauseam)
Availability heuristic, and confirmation bias almost categorically will play a role - even amongst big names, we won't be as aware of injuries or absences in players we don't follow week-to-week, whilst we'll notice periods away by even lowly ranked players (where's Dommie Covington gone, where's Mirabelle, etc)
RET also doesn't speak to how many matches were missed, the severity of the injury, and the season/career impact of the average injury/absence per nation. I have a metric, 'days since last played', which might speak to that, but doesn't necessarily speak to it - the absence could be due to any number of things besides injury/illness.

Otherwise: Numbers correct, as far as I could tell.
I want to treat a 'Player 1' (P1) as one thing, and a 'Player 2' (P2) as another, and speak to the fractions of matches that were decided by RET in favour of each case. Irrespective of who was the opponent, how often did each scenario occur. Deal with match outcomes. Which is what the figures presented did.
I can see the argument (now) to represent 'GB effort/participation' as a discrete thing, whether that effort occured as P1, or P2 (as the second player in all-GB matches). But this changes the rules for all-GB matches.
Of those 142 matches, 3 were decided by RET. To mix it as GB we can, as you say, look at the total number of players involved in those matches - 284 - all GB, and factor a ratio on that.

The fairest way to frame that within your argumanent might be to say:
We measured 3496 performances - 1748 matches each comprising 2 players that gave a 'performance'
Of those 3496 performances:
1890 were by a GB player (1748 as P1 in every single match + an additional 142 as P2 in all-GB matches)
1606 were by a Pesky (3496 - 1890)

Of those 1890 GB performances:
21 ended with the Brit retiring (18 against a pesky + 3 against another Brit)
GB RET in 21/1890 performances = 1.111%

Of the 1606 Pesky performances:
16 ended with the Pesky retiring
Pesky's RET in 16/1606 performances = 0.996%

GB RET variance: 1.111 - 0.996 = +0.115%
Or, not much.

blob · All-time great

Only one $25K this week.
Consecutive clay $25K in Brazil unsurprisingly leads to very similar fields.
MD field this week is very short though, as there were JE & multiple LL, reducing the number of DA as of right that actually took to court.

Oakland2002 · All-time great

I like the weekly comparison of field strengths it puts a relative perspective on things and tempers expectations. Not too relevant his week but thanks.

blob · All-time great

GB 5-9 look at those lines

blob · All-time great

Field Comparisons Week 10:

All >= $25K events:

Comparison of relative strengths in the Australian series of events thus far:

Comparsion of two events field strength in 2017 and 2018

Michael D · All-time great

Charts fascinating as ever. Still a yawning vacancy in the GB #10 position that no-one really wants at the moment. I'd expected Sam to fill it this early part of the year, and indeed she still is the #10 but as a non-player and with no sign yet of returning. Mandy Carreras a bit further back has delayed her return, and after appearing on one entry list is now off everything for the next few weeks at least, Maia and Eden are both struggling with early season form and the transition between 15k and 25k, which they need to make if they really want point rises, and Tara is struggling with singles full stop. All those five remain between 420 and 467 at the moment, but none of them is making any real upwards push at present.

blob · All-time great

I've been trying to figure out why I've been feeling that there's something missing this season, and I think I might finally have an idea: there is not, thus far, a Robyn Beddow/Daniela Konotoptseva/Dommie Covington/Mirabelle Njoze/Suzy Larkin story to follow each week. I've gotten used to having an outlier story of interest running in the background like that.

Michael D · All-time great

blob wrote:
I've been trying to figure out why I've been feeling that there's something missing this season, and I think I might finally have an idea: there is not, thus far, a Robyn Beddow/Daniela Konotoptseva/Dommie Covington/Mirabelle Njoze/Suzy Larkin story to follow each week. I've gotten used to having an outlier story of interest running in the background like that.

It goes on even further. If you look at the GB women player by player lists of all those with rankings there is also no Fran Jones, Emily Arbuthnott (college), Olivia Nicholls, Laura Sainsbury, Laura Deigman, Tiffany William (though she's injured), Manisha Foster, Emma Hurst, Pitak sisters, Ella Taylor (up in a week I think), Erin Richardson, Emile Lindh Gallagher, Summer Yardley, Jasmine Asghar, Gemma Heath, Louise Holtum, Tanshya Dissanayake, Anna Loughlan, Olivia Peet.

These are all GB women, in addition to Mirabelle and Suzy who have points but have not been able to play this year in large part because of the lack of GB 15k tournaments (a few played in the 25k qualifiers, otherwise on the British Tour). At this rate we soon will not have a GB top 25 any longer, barely a top 20, with the lack of opportunities afforded now to women to get on the bottom rungs of the tennis ladder, especially with the fact that for most 3 scoring tournaments are required before a ranking can be obtained.

Coup Droit · Tennis legend

I don't think Blob was meaning ranked women who are not playing for one reason or another.

I think he was talking about needing a 'quirky' player to follow, no matter who - one from the wrong side of the tracks, if you will - the sort who bonds us all together in wishing her luck, with the British love of an underdog (but knowing really that it's going to take a miracle).

Peter too · Tennis legend

Another player who's name still appears occasionally in the entries but often withdraws is Ines James (Janbaksh). She used to put her name down as standby for WTA events and play a round. Starting in 2005 she got a CH of 1017 in 2009. Her last appearance was in Woking last year where she lost 0,0 to Sam Murray.

Michael D · All-time great

Coup Droit wrote:
I don't think Blob was meaning ranked women who are not playing for one reason or another.

I think he was talking about needing a 'quirky' player to follow, no matter who - one from the wrong side of the tracks, if you will - the sort who bonds us all together in wishing her luck, with the British love of an underdog (but knowing really that it's going to take a miracle).

I see the two as being very connected. If the women cannot get on the courts to play... then we won't have the quirky ones either, unless they have bags of money to be self-funded.

Michael D · All-time great

Michael D wrote:
Coup Droit wrote:
I don't think Blob was meaning ranked women who are not playing for one reason or another.

I think he was talking about needing a 'quirky' player to follow, no matter who - one from the wrong side of the tracks, if you will - the sort who bonds us all together in wishing her luck, with the British love of an underdog (but knowing really that it's going to take a miracle).
I see the two as being very connected. If the women cannot get on the courts to play... then we won't have the quirky ones either, unless they have bags of money to be self-funded.

Or to put it another way, I believe that Emily Appleton (sponsored by Sky) live WR 611, Emily WS, live WR 641 and Alicia Barnett live WR 654, are the lowest ranked GB players to have won points so far this year?

Oh Ella Taylor at live rank 1138 has won 1 pt this year reaching the Loughborough 25k L32, and Nell Miller ditto UNR. But those are the only two beyond the cluster in the 600s. Difficult for many quirky players to emerge that way...