Re: UK 'Points' System

On Apr 13, 10:33 pm, Chris Kerr <cj...@xxxxxxxxx> wrote:
Peter Ford wrote:

However, 10 minutes thought leads me to suggest a basic system which
would work reasonably well:

Every rower has a ranking: a real number between 0 and 100. The
ranking of a crew is the average of the rowers in it.
Regattas would be encouraged to take entries just by boat type and
sex, and then subdivide them as follows:
Given some number of entered men's 4+s, divide them into roughly equal
sized categories to give sensible sized categories (say 5-8 for a
river regatta, 12-18 for Dorney, etc.).
Place crews into these categories by taking eg the highest ranked 8
crews, then the next highest 8, etc.
Race against similar crews and have fun, somebody gets a pot.
At the end of the regatta, send the full results off to Lower Mall as
now.
A computer calculates updates to the rankings based on the results the
crew got, and the rankings of the crew they beat and lost to. The
updated ranking is then accessible to competitors via eg their log-in
to the BR website.

The main point of this is: the details of the updating calculations
don't need to be simple, or particularly fixed; changing them from
year to year would be easy. It wouldn't matter that the competitors
didn't know months in advance whether they were going to be able to
enter IM1 2- at Shrewsbury, or IM3 8+ at Marlow; they would enter and
know that they would race a reasonable number of crews of a similar
standard (subject to entering a sufficiently popular boat class, as
now), and one of them would win. Crucially, there could be no huge
upsets where a crew suddenly won a couple of events at Evesham regatta
and had 16 extra points to deal with; the ranking (once settled in)
would be relatively stable to race results, because *all* of the
results would be considered, not just the winning crew.

There are various missing details that would take decisions and
judgement (the first to spring to mind: how to 'set-up' the system,
mapping from current points; how to integrate Junior and Vet racing
into the system), but I don't think any of them represent fundamental
flaws.

Thoughts?

Peter Ford

I think that's the best idea I've heard yet - here are some of my ideas for
possible improvements:

1) If you're going to use a real number, why limit it to an interval? You
could easily just use arbitrary real numbers (or perhaps, to avoid putting
people off, just use positive reals). However, if you are going to limit the
number to within a range, I'd say from an aesthetic point of view it should
be between 0 and 1 - if you want to be able to quote a number between 0 and
100 then perhaps you could express it in percent.

True. I thought that people would find it easiest to discuss small two-
digit numbers, rather than 0.something
What turned out to be best might be dependent on what you decided to
start people on when switching over from the current system.

2) Why stick with one ranking system? If you make available a database of
past results, any regatta organiser who disagrees with the ARA's ranking
algorithm can simply use their own method. We could even do controlled
trials to see which ranking method is the best predictor of future
performance, and which classification system gives the closest races.

An interesting idea, but I think this would be thought of as more of a
step forward in the organisation needed, to have a database available.
Whether it would be I don't know.
3) If people like knowing in advance which category they are going to be in,
you could always call [0,10) Novice, [10,30) IM3, [30,50) IM2, [50,70) IM1,
[70,90) Senior and [90,100) Elite (or some other set of intervals, depending
on how the ranking system is calculated).

No, I think this weakens the idea substantially. The main thrust of my
idea (in my mind, at least), is that people shouldn't have to care
about their ranking. Similar to stated by Tink, what I want from a
system is:

I get to race, preferably (if I'm successful) over a total distance
somewhere near 4k per event during a day's regatta racing (although I
do like the repechages when they're run at Dorney; maybe that counts
as not being successful...)

My competitors are neither much faster nor much slower than me.

I don't have to think about what crew and boat type to enter in order
to satisfy the first two criteria, I enter what I'd like to race, with
the crew-mates I want to race with.

Under the current system, it's possible to satisfy the first two
criteria, but only by carefully managing how many points you have.

4) I don't think there's any need to have separate ranking systems for
juniors and veterans - it won't take many people occasionally racing in two
different categories to even out the scores between age groups.

Yes, I think the rankings should work fine being updated across all
races, I was just noting the fact that you would want to keep the
junior categories and veteran categories as alternative events, as
now; a junior crew could either choose to enter their age category or
Open, where they would be placed according to their ranking.

Everyone will have their own favourite algorithm, but here are a few
criteria that I think any ranking system should aim to fulfil (although some

* It should be possible for someone's score to decrease if they have stopped
training and are no longer competitive at their old level.

* Losing your first race shouldn't put you on a lower score than someone who
has never raced.

* It shouldn't be possible to get into an easier category for an important
regatta by deliberately losing less important events. [Perhaps these events
could rank by your highest ever score rather than your current one.]

* Beating many other crews should give you more points than beating one
other crew.

* Beating one crew of your own standard or better should give you more
points than beating lots of crews that are all much worse than you.

* If two approximately equal crews race each other at lots of consecutive
regattas and each one wins half of the time, their points total shouldn't
increase significantly (with the possible exception of novices where you
could argue that the race experience makes them better).

* If crew A races at lots of regattas all season while crew B only races a
couple of times but beats A every time, crew B should have a higher score
than A.

* If crew A is usually much better than crew B but A has an equipment
failure, crashes, catches a crab etc, neither A nor B should end up too far
from their original status.

Yes, I'd agree with all of them.
The other crucial factor is how to arrange for averaging to work well;
if George Nash jumps in the 6 seat of the St Catharine's boat, the
other 7 of whom have a ranking of say 50, how many points he has
should work out so that you're matched against the "60" crews you can
now race.
In this way, how many points you should get for being really good (or
really bad...) has to be dependent on your effect on an average boat;
the problem with this is that less data is going to be available for
this than on how more evenly distributed crews will perform, I would
imagine.
.

Relevant Pages

• Re: Regatta Programmes
... >> Regatta of the season and I wonder if anybody has ... >> due to boat, crew and equipment clashes. ... > the 2000m race in 2 minutes please let us race" events occurring. ... I've organised a number of regattas, and my opinion is that boat clashes are ...
(rec.sport.rowing)
• Re: UK Points System
... than a Nether Wallop Regatta crew - so if you beat the Dorney crew, ... end of every regatta. ... racing after the first weekend in July, or at local regattas, is "pot- ... Every rower has a ranking: a real number between 0 and 100. ...
(rec.sport.rowing)
• Re: UK Points System
... For crews, yes you'd have to use the crew average, and this system ... end of every regatta. ... racing after the first weekend in July, or at local regattas, is "pot- ... Every rower has a ranking: a real number between 0 and 100. ...
(rec.sport.rowing)
• Re: UK Points System
... Every rower has a ranking: a real number between 0 and 100. ... Race against similar crews and have fun, ... At the end of the regatta, send the full results off to Lower Mall as ... crew got, and the rankings of the crew they beat and lost to. ...
(rec.sport.rowing)
• Re: Hereford Regatta off (UK)
... the 300+ entries of a small regatta are actually likely to ... is the best way to make a decent day of it. ... I would always much rather race in my ... If I lose to crew ...
(rec.sport.rowing)