Redress.
Say a boat's series *scores* are:-
2.0 DNF 4.0 2.0 DNF
the boat is given redress for both DNFs as average scores of all
races excluding the race in question.
2.0 RDGAveAll 4.0 2.0 RDGAveAll
to exclude the race in question in one case, is to include the
other RDG.
There is a common sense way to do this (ignore all other RDGs when
calculating a RDG) - but - are the RRS specific about it and of
not why not.
Regarding redress, the RRS provide "Guidance" not "rules." There are so many possible situations that might arise that making "rules" for redress simply wouldn't be practical.
Personally, I think I would tend to say that the default should be "average of all the races except the race in question." However, if there is another RDG then I would ignore that although if that redress is some kind of average it probably won't make any significant difference. Bottom line - RDG is always at the discretion of the PC. So, it cannot be programmed in definitively and there always needs to be options. I would suggest that the "average points" need three options - "races before", "all races except" and "specified races."
Art
On 7/20/2012 12:49 AM, Colin Jenkins wrote:
Redress.
Say a boat's series *scores* are:-
2.0 DNF 4.0 2.0 DNF
the boat is given redress for both DNFs as average scores of all races excluding
the race in question.2.0 RDGAveAll 4.0 2.0 RDGAveAll
to exclude the race in question in one case, is to include the other RDG.
There is a common sense way to do this (ignore all other RDGs when calculating a
RDG) - but - are the RRS specific about it and of not why not.--
Cheers,
Colin J
http://sailwave.com
thanks Art,
Sailwave has those 3 variants - i've just been pondering the
implementation while doing the new Sailwave.
what it currently (windows version) does is:-
1. apply for the 'single race' redresses points - i.e. redress
of X points.
2. apply the 'races before' variants starting with the lowest
race - computable then.
3. apply the 'selected races' variant - some ambiguity here -
depends on order of application.
4. apply the 'average all' variant.
but in hindsight i think it needs simplifying to always ignore
other redress points - its explainable then and checkable.
On 20/07/2012 09:20, Art Engel wrote:
Cheers,
Colin J
Regarding redress, the RRS provide "Guidance" not“rules.” There are so
many possible situations that might arise that making
“rules” for
redress simply wouldn’t be practical.Personally, I think I would tend to say that the defaultshould be
“average of all the races except the race in question.”
However, if
there is another RDG then I would ignore that although if
that redress
is some kind of average it probably won’t make any
significant
difference. Bottom line - RDG is always at the discretion
of the PC. So,
it cannot be programmed in definitively and there always
needs to be
options. I would suggest that the “average points” need
three options -
“races before”, “all races except” and “specified races.”Art On 7/20/2012 12:49 AM, Colin Jenkins wrote: > > > Redress. > > Say a boat's series *scores* are:- > > 2.0 DNF 4.0 2.0 DNF > > the boat is given redress for both DNFs as averagescores of all races excluding
> the race in question.
>
> 2.0 RDGAveAll 4.0 2.0 RDGAveAll
>
> to exclude the race in question in one case, is to
include the other RDG.
>
> There is a common sense way to do this (ignore all
other RDGs when calculating a
> RDG) - but - are the RRS specific about it and of not
why not.
>
> –
> Cheers,
> Colin J
> http://sailwave.com
>
>
>
No virusfound in this message.
Checked by AVG - www.avg.com
Version: 2012.0.2178 / Virus Database: 2437/5142 - Release Date:
07/19/12
If the redress is avg of ALL other races, then it will always be the average of only the non-redress races, no matter how many redress races there are (hooray math!). HOWEVER, if the redress is avg of all PRIOR races, then the first redress must be established, then the second redress is calculated, and so forth, as the regatta proceeds. I.e., each redress builds upon the score assigned to any prior redress. You cannot simply ignore a prior redress race, as that score is legitimately "earned" by the competitor based on the circumstances of that race.
Using Colin's example...
AveAll
2.0, RDG(2.7), 4.0, 2.0, RDG(2.7) = 13.4
AveBefore, incl RDG
2.0, RDG(2.0), 4.0, 2.0, RDG(2.5) = 12.5
AveBefore, excl RDG (penalizes the competitor!)
2.0, RDG(2.0), 4.0 2.0, RDG(2.7) = 12.7
I think I might have a slightly different take on what is the appropriate redress.
The goal, I believe, is to make the "bad" race (the race for which a boat is getting redress) have no impact. In other words, you are trying to undo the badness, as it were. But, each RDG given is merely an approximation of that and probably not a very good approximation. Hence, I would tend to ignore all RDG and just use the other races. Using that standard, in the analysis below I would say that the "best" redress is the AveAll (although it is mathematically the same, I would say AveAll, excl all RDG). The AveBefore, excl RDG doesn't penalize the competitor, it merely benefits the competitor less than AveBefore, incl RDG. But, both are a slight distortion of the goal.
Why then would anyone consider using AveBefore?
There is another important consideration when giving redress - How will all competitors know where the redress-competitor is in the standings? This is important in the last race and also usually for the last day of a multi-day regatta or series. How else can you know which competitor to cover? Based on this consideration the PC wants to award RDG that won't include the last race or the races on the last day. That makes the RDG fairest to ALL the competitors. So, I would lean toward awarding RDG of AveBefore, excl RDG with the either the last race or all the races on the last day excluded (with any RDG excluded as well).
Colin - I don't think double redress happens often enough to worry a lot about. But, I should think one easy way to handle it would be to have a fifth average option that would calculate an average based on specified races [or, based on all races EXCEPT specified races] (then there would presumably be a box to list races to include [or exclude], like when you specify your discard profile). Otherwise, in the case of double redress the scorer is going to have to recalculate by hand after new races are added. [The "average all races except" would probably be slightly more versatile as presumably fewer races would need to be listed.]
Art
On 7/20/2012 9:57 AM, kett_cummins wrote:
If the redress is avg of ALL other races, then it will always be the
average of only the non-redress races, no matter how many redress races
there are (hooray math!). HOWEVER, if the redress is avg of all PRIOR
races, then the first redress must be established, then the second
redress is calculated, and so forth, as the regatta proceeds. I.e., each
redress builds upon the score assigned to any prior redress. You cannot
simply ignore a prior redress race, as that score is legitimately
"earned" by the competitor based on the circumstances of that race.Using Colin's example...
AveAll
2.0, RDG(2.7), 4.0, 2.0, RDG(2.7) = 13.4AveBefore, incl RDG
2.0, RDG(2.0), 4.0, 2.0, RDG(2.5) = 12.5AveBefore, excl RDG (penalizes the competitor!)
2.0, RDG(2.0), 4.0 2.0, RDG(2.7) = 12.7
Just to add a further complication - which average do we use?
I had a long discussion once with an international judge who was adamant that using median points rather than mean points provided a more realistic redress! Being one of the world' natural number-crunchers he then took the result from from several years of a major cruiser racing series (which he had on a spreadsheet) and devised a test where each result was replaced by redress (either mean or median) and the result compared with the actual result. Statistically, median points in the series gives a result closer to the actual result.
Nit-picking? Possibly. But if the point of giving redress is to re-establish a score as close to that which a boat would have scored then this detail might be worth considering.
Gordon
(not one of the world's natural number-crunchers).
On 21/07/2012 00:04, Art Engel wrote:
I think I might have a slightly different take on what is the
appropriate redress.The goal, I believe, is to make the "bad" race (the race for which a
boat is getting redress) have no impact. In other words, you are trying
to undo the badness, as it were. But, each RDG given is merely an
approximation of that and probably not a very good approximation. Hence,
I would tend to ignore all RDG and just use the other races. Using that
standard, in the analysis below I would say that the "best" redress is
the AveAll (although it is mathematically the same, I would say AveAll,
excl all RDG). The AveBefore, excl RDG doesn't penalize the competitor,
it merely benefits the competitor less than AveBefore, incl RDG. But,
both are a slight distortion of the goal.Why then would anyone consider using AveBefore?
There is another important consideration when giving redress - How will
all competitors know where the redress-competitor is in the standings?
This is important in the last race and also usually for the last day of
a multi-day regatta or series. How else can you know which competitor to
cover? Based on this consideration the PC wants to award RDG that won't
include the last race or the races on the last day. That makes the RDG
fairest to ALL the competitors. So, I would lean toward awarding RDG of
AveBefore, excl RDG with the either the last race or all the races on
the last day excluded (with any RDG excluded as well).Colin - I don't think double redress happens often enough to worry a lot
about. But, I should think one easy way to handle it would be to have a
fifth average option that would calculate an average based on specified
races [or, based on all races EXCEPT specified races] (then there would
presumably be a box to list races to include [or exclude], like when you
specify your discard profile). Otherwise, in the case of double redress
the scorer is going to have to recalculate by hand after new races are
added. [The "average all races except" would probably be slightly more
versatile as presumably fewer races would need to be listed.]Art
On 7/20/2012 9:57 AM, kett_cummins wrote:
> If the redress is avg of ALL other races, then it will always be the
> average of only the non-redress races, no matter how many redress races
> there are (hooray math!). HOWEVER, if the redress is avg of all PRIOR
> races, then the first redress must be established, then the second
> redress is calculated, and so forth, as the regatta proceeds. I.e., each
> redress builds upon the score assigned to any prior redress. You cannot
> simply ignore a prior redress race, as that score is legitimately
> "earned" by the competitor based on the circumstances of that race.
>
> Using Colin's example...
>
> AveAll
> 2.0, RDG(2.7), 4.0, 2.0, RDG(2.7) = 13.4
>
> AveBefore, incl RDG
> 2.0, RDG(2.0), 4.0, 2.0, RDG(2.5) = 12.5
>
> AveBefore, excl RDG (penalizes the competitor!)
> 2.0, RDG(2.0), 4.0 2.0, RDG(2.7) = 12.7
>
If only the decision lay in the Maths. There are other factors to consider and nothing is absolutely correct.
Mike b IJ
Just to add a further complication -
which average do we use?
I had a long discussion once with an international judge who was
adamant that using median points rather than mean points provided
a more realistic redress! Being one of the world’ natural
number-crunchers he then took the result from from several years
of a major cruiser racing series (which he had on a spreadsheet)
and devised a test where each result was replaced by redress
(either mean or median) and the result compared with the actual
result. Statistically, median points in the series gives a result
closer to the actual result.
Nit-picking? Possibly. But if the point of giving redress is to
re-establish a score as close to that which a boat would have
scored then this detail might be worth considering.
Gordon
(not one of the world's natural number-crunchers).
On 21/07/2012 00:04, Art Engel wrote:
I think I might have a slightly different take on what isthe
appropriate redress.The goal, I believe, is to make the "bad" race (the racefor which a
boat is getting redress) have no impact. In other words,
you are trying
to undo the badness, as it were. But, each RDG given is
merely an
approximation of that and probably not a very good
approximation. Hence,
I would tend to ignore all RDG and just use the other
races. Using that
standard, in the analysis below I would say that the
“best” redress is
the AveAll (although it is mathematically the same, I
would say AveAll,
excl all RDG). The AveBefore, excl RDG doesn’t penalize
the competitor,
it merely benefits the competitor less than AveBefore,
incl RDG. But,
both are a slight distortion of the goal.Why then would anyone consider using AveBefore? There is another important consideration when givingredress - How will
all competitors know where the redress-competitor is in
the standings?
This is important in the last race and also usually for
the last day of
a multi-day regatta or series. How else can you know which
competitor to
cover? Based on this consideration the PC wants to award
RDG that won’t
include the last race or the races on the last day. That
makes the RDG
fairest to ALL the competitors. So, I would lean toward
awarding RDG of
AveBefore, excl RDG with the either the last race or all
the races on
the last day excluded (with any RDG excluded as well).Colin - I don't think double redress happens often enoughto worry a lot
about. But, I should think one easy way to handle it would
be to have a
fifth average option that would calculate an average based
on specified
races [or, based on all races EXCEPT specified races]
(then there would
presumably be a box to list races to include [or exclude],
like when you
specify your discard profile). Otherwise, in the case of
double redress
the scorer is going to have to recalculate by hand after
new races are
added. [The “average all races except” would probably be
slightly more
versatile as presumably fewer races would need to be
listed.]Art On 7/20/2012 9:57 AM, kett_cummins wrote: > If the redress is avg of ALL other races, then itwill always be the
> average of only the non-redress races, no matter how
many redress races
> there are (hooray math!). HOWEVER, if the redress is
avg of all PRIOR
> races, then the first redress must be established,
then the second
> redress is calculated, and so forth, as the regatta
proceeds. I.e., each
> redress builds upon the score assigned to any prior
redress. You cannot
> simply ignore a prior redress race, as that score is
legitimately
> “earned” by the competitor based on the circumstances
of that race.
>
> Using Colin’s example…
>
> AveAll
> 2.0, RDG(2.7), 4.0, 2.0, RDG(2.7) = 13.4
>
> AveBefore, incl RDG
> 2.0, RDG(2.0), 4.0, 2.0, RDG(2.5) = 12.5
>
> AveBefore, excl RDG (penalizes the competitor!)
> 2.0, RDG(2.0), 4.0 2.0, RDG(2.7) = 12.7
>
At the Olympics we've seen a redress option used which I think Art would agree with but I don't think Sailwave can manage. That's average points for all races before the last day.
This is a good option because it avoids the perversion of redress seen at one high profile event last summer where a boat given redress was then sailed down the fleet in the last race so as to make both their last race score and their redress score worse.
I guess one would present this as average of first N races [exc DNC] or average of all but last N races [ex DNC].
Jim C
--- In sailwave@yahoogroups.com, Art Engel <artengel123@...> wrote:
I think I might have a slightly different take on what is the
appropriate redress.