New Posts New Posts RSS Feed: New PY Website
  FAQ FAQ  Forum Search   Register Register  Login Login

New PY Website

 Post Reply Post Reply Page  <1 910111213 14>
Author
Andymac View Drop Down
Really should get out more
Really should get out more


Joined: 04 Apr 07
Location: Derbyshire
Online Status: Offline
Posts: 833
Post Options Post Options   Quote Andymac Quote  Post ReplyReply Direct Link To This Post Posted: 07 Nov 12 at 1:31am
I did qualify my example by saying that they were racing around the same course at the same time.
 
I totally agree with you that a large separation time between races starting or finishing could lead to a corruption in results/data.
 
Originally posted by sargesail


Now I would be the first to say that I have no qualms with the Draycote Dash methos of producing a single set of results for three fleets which start over a 6 minute period...
To me if the separation is more than you could reasonably bring together the handicap and fleet times and produce one set of results then the data is flawed...
 
With the caveat that the comparable races are conducted on the same course at the same time, I believe you are in agreement with me.
 
 
 
Back to Top
pondmonkey View Drop Down
Really should get out more
Really should get out more
Avatar

Joined: 12 Aug 11
Online Status: Offline
Posts: 2202
Post Options Post Options   Quote pondmonkey Quote  Post ReplyReply Direct Link To This Post Posted: 07 Nov 12 at 7:06am
At the same time, same start, same course- yep, it's effectively one big handicap race with extracted fleet results. Okay there's a 'mindset' angle I could push, but I won't- that's jus argumentative.   

However our data is NOT collected from the boats starting at the same time- so I think we're agreeing it's flawed, but maybe disagree that it's 'within tolerance' of usable/fair data. I do acknowledge that the RYA is desperate for any data, therefore maybe this is better than nothing, but that's hardly a ringing endorsement for the scheme!!!!

Oh God I'm sounding like a stats man....

Edited by pondmonkey - 07 Nov 12 at 7:08am
Back to Top
JimC View Drop Down
Really should get out more
Really should get out more
Avatar

Joined: 17 May 04
Location: United Kingdom
Online Status: Offline
Posts: 5419
Post Options Post Options   Quote JimC Quote  Post ReplyReply Direct Link To This Post Posted: 07 Nov 12 at 8:13am
Chaps, you may not know it, but in this discussion you are getting deep into statistical theory - far deeper than I for one have the mathematics to safely go and also far deeper than you actually have enough information to usefully discuss the topic.

As a rule of thumb more data normally beats selected data in statistical analysis unless you know *exactly* what you are doing, so my money is that what Draycote are doing will produce more accurate PY numbers than they were getting before.

If you want to contest this it ought to be straightforward enough as the sets of data are reasonably identifiable. What you need to do is get all the data, which I imagine is published on their website, separate the two bodies, that of times extracted from class races and times extracted from handicap races, and run a series of comparisons to see if the variations between them are statistically valid.

I have no intention of doing the research it would take *me* as a very amateur statistician even to work out how to do the analysis but it would probably be on the lines of starting with standard deviation and variance and going from there. Without doing this, well, I suppose internet forums are just pub chat...
Back to Top
pondmonkey View Drop Down
Really should get out more
Really should get out more
Avatar

Joined: 12 Aug 11
Online Status: Offline
Posts: 2202
Post Options Post Options   Quote pondmonkey Quote  Post ReplyReply Direct Link To This Post Posted: 07 Nov 12 at 9:12am
Originally posted by JimC

Chaps, you may not know it, but in this discussion you are getting deep into statistical theory - far deeper than I for one have the mathematics to safely go and also far deeper than you actually have enough information to usefully discuss the topic.

As a rule of thumb more data normally beats selected data in statistical analysis unless you know *exactly* what you are doing, so my money is that what Draycote are doing will produce more accurate PY numbers than they were getting before.

If you want to contest this it ought to be straightforward enough as the sets of data are reasonably identifiable. What you need to do is get all the data, which I imagine is published on their website, separate the two bodies, that of times extracted from class races and times extracted from handicap races, and run a series of comparisons to see if the variations between them are statistically valid.

I have no intention of doing the research it would take *me* as a very amateur statistician even to work out how to do the analysis but it would probably be on the lines of starting with standard deviation and variance and going from there. Without doing this, well, I suppose internet forums are just pub chat...


All good points Jim, and for clarity it's all a bit moot, because ATM, I understand the handicap fleet just use the numbers off the shelf, so the issues around timing everyone, the added burden to OD duty etc only exist for a theoretical study.

Like most clubs I reckon there'd be concern from some if 'localised ratings' were used. A lot of the handicap fleet appear to be moving on too... But then we're very fortunate to have a very strong class racing ethos (I must be getting old to say that!!) and a newly growing moth fleet for the adrenaline overstocked, so I'm fortunate to have options.

Edited by pondmonkey - 07 Nov 12 at 9:15am
Back to Top
Oli View Drop Down
Really should get out more
Really should get out more
Avatar

Joined: 23 Mar 05
Location: essex
Online Status: Offline
Posts: 825
Post Options Post Options   Quote Oli Quote  Post ReplyReply Direct Link To This Post Posted: 07 Nov 12 at 4:03pm

as promised a couple of pages back, list of changes for classes throughout the season.

 
Back to Top
craiggo View Drop Down
Really should get out more
Really should get out more
Avatar

Joined: 01 Apr 04
Location: United Kingdom
Online Status: Offline
Posts: 1723
Post Options Post Options   Quote craiggo Quote  Post ReplyReply Direct Link To This Post Posted: 07 Nov 12 at 9:41pm
So for those interested in the suggested changes for a tidal estuary, here are my clubs latest figures from PYS.

Back to Top
RS400atC View Drop Down
Really should get out more
Really should get out more


Joined: 04 Dec 08
Online Status: Offline
Posts: 2727
Post Options Post Options   Quote RS400atC Quote  Post ReplyReply Direct Link To This Post Posted: 12 Nov 12 at 12:08pm
Originally posted by craiggo

So for those interested in the suggested changes for a tidal estuary, here are my clubs latest figures from PYS.

Looking at those figues, particularly the Scorpion, you seem to have a serious problem.
You have significant data, yet you are suggesting a PY 8% different from what you are using?
That is an immense difference, how do expect people to take your racing seriously if you spread the word that the py's are that badly wrong?

I suspect you have some good sailors sailing Scorpions?

Back to Top
Oli View Drop Down
Really should get out more
Really should get out more
Avatar

Joined: 23 Mar 05
Location: essex
Online Status: Offline
Posts: 825
Post Options Post Options   Quote Oli Quote  Post ReplyReply Direct Link To This Post Posted: 12 Nov 12 at 12:23pm
looking at the numbers id say that either the river/courses are very biased towards symmetric classes, and/or that the better sailors are in the symmetric classes.
Back to Top
RS400atC View Drop Down
Really should get out more
Really should get out more


Joined: 04 Dec 08
Online Status: Offline
Posts: 2727
Post Options Post Options   Quote RS400atC Quote  Post ReplyReply Direct Link To This Post Posted: 12 Nov 12 at 12:40pm
Originally posted by slop_idol

looking at the numbers id say that either the river/courses are very biased towards symmetric classes, and/or that the better sailors are in the symmetric classes.


Looks like the RS700 figures would challenge one of those possibilities?

It seems to me the current adjustment scheme really does not work in the real world where ability varies from class to class.
Back to Top
Peaky View Drop Down
Really should get out more
Really should get out more


Joined: 15 Nov 08
Location: United Kingdom
Online Status: Offline
Posts: 2571
Post Options Post Options   Quote Peaky Quote  Post ReplyReply Direct Link To This Post Posted: 12 Nov 12 at 12:46pm
Not for the first time in my life, I am a little confused.

To keep things simple, consider a two boat race. These boats, A and B, race each other 1000 times. Although different classes, A and B were initially given the same PY. 1000 races later (high confidence index) it is clear that A is faster, even when the helms swapped (so not a crew skill factor thing). In this latest version of the PYS, how do you know whether A should have its PY dropped, or B's increased? Perhaps a bit of both?

How do you ensure that you don't get the situation whereby Club X drops A's PY and Club Y raises B's?

The Confidence Index also confuses me. It should be based upon more than just the number of results generated by that class. Imagine that 10 boats in Class A race against 1 boat in Class B. AIUI, that would give Class A a higher Confidence Index than Class B. However, if a new class, Class C, gets introduced, it may become clear that actually it is A that is wrongly handicapped. The CI should be based on the number of classes raced against as well as the number of results.

Edited by Peaky - 12 Nov 12 at 12:46pm
Back to Top
 Post Reply Post Reply Page  <1 910111213 14>

Forum Jump Forum Permissions View Drop Down

Bulletin Board Software by Web Wiz Forums® version 9.665y
Copyright ©2001-2010 Web Wiz