Hi Art
Having just recovered from curing my broken PC I have just got round to
looking at your scoring formula which I believe has the right general
characteristics and is along the same lines as Rinderle B and CHIPS although
rather more “aggressive” at the bottom end (i.e. for few starters).
Yes I have quite an interest in scoring systems having gone through such
an analysis in depth in 2003 and settled on CHIPS as the fairest and most
acceptable HPS method, well at least in my opinion. (You may find it useful to look at the Chipstead SC website
– www.chipsteadsc.org.uk, link
reproduced here since you mentioned a broken link - and also the paper by
Malcolm Clark, http://www.styvechale.net/chip.pdf
, which gives a detailed mathematical critique of CHIPS).
Rinderle B gives the same score for being last in a race irrespective of
the number of starters – which is a massive difference from one of the
principles of conventional Low Point Scoring where the score for coming last
normally equals the number of starters.
And of course the score of the last boat is important in this type of
HPS system because the points for all the intermediate positions are slotted in
between first and last. One of the features of CHIPS is that it preserves that
particular characteristic of LPS in allocating a fair score for last as a
function of the number of starters which importantly means that everyone else then
also scores fairly. But of course
the view of what is fair is totally subjective judgement – there is no such
thing as a “correct” scoring
system, they are all arbitrary.
When we came up with CHIPS a few years ago the big consideration was how
aggressive should the scheme be, and we received a great deal of stick because
competitors were unhappy if their score for a particular position could be
compromised because someone who had done “worse” – i.e. one or even 2 places
further down the fleet in another race in the series – were allocated a better
score simply because of the presence of more starters in the race. It is mainly
for this reason that we went through three versions (CHIPS 1, 2 and 3) as
described in my “All about CHIPS” paper, before we homed in on CHIPS 3 as being
the optimum. The key
characteristic was that we needed to “flatten” the curves so that the effect of
the number of starters became less dominant (i.e. less aggressive).
To illustrate this with 3 examples of why we decided that CHIPS gives a
fairer outcome and how we addressed the aggressiveness issue, it is worth
comparing the scores for your formula vs CHIPS. I will use the notation 2(5) to mean the score for coming 2nd
in a race with 5 starters, and compare the Art Engel method with CHIPS: -
(i) Art Engel: 1(3) =
8.5 is the same as 2(9) = 8.5 which means in a race with 6 more starters a 2nd
will equate to a 1st with 3 starters - i.e. A difference of 6 boats.
CHIPS: 1(3) = 90
which is beaten by 2(11) = 90.1, i.e. a difference of 8 or more boats is
required for a 2nd to beat a 1st.
(ii) Art Engel:
2(3) = 5.5 is the same as 3(5) = 5.5 which means only 2 more starters
are required for a 3rd to score the same – Difference of 2 boats
CHIPS: 2(3) =
77.5 is only beaten by a 3rd when there are 8 or more boats (78.2),
i.e. a Difference of 5 boats.
(iii) Art Engel: 2(5) = 7.3 is beaten by 3(9) = 7.5, a difference of 4 boats.
CHIPS: 2(5) =
82.1 is beaten by 3(11) = 82.8, a difference of 6 boats.
Our sailors objected strongly to a system in which they appeared to be
excessively penalised when fewer boats raced and so using CHIPS we flattened
the curves to reduce the influence of the number of boats while preserving the
other advantages of the scheme. And we considered it essential to preserve fair
scores all the way down the fleet not just at the top.
In one of your posts I think you mentioned the benefits of a percentage
system. CHIPS aims at being a percentage system by not only making the max
possible score for winning a race 100 it works on the basis that if there is a
large turn-out then the performance of each helm can be judged as being related
to the performance for the club as a whole and spans the range 100 to 5 from 1st
to last, leaving zero for a DSQ. As
the number of competitors reduces the range of scores reduces at both the top
and bottom ends which makes sense since the turn-out is less representative of
the club as a whole, while the scheme conforms to the HPS principles in which one
scores more by beating more boats and in the case of CHIPS, at the back of the
fleet one also scores more if one is beaten by fewer boats. (The latter occurs
in your system but in a very limited manner, i.e. a small effect, while in
Rinderle B there is no recognition at all that one should score better at the
back of the fleet if one is beaten by fewer boats).
The other issue raised by our club members is that the system is more complicated,
making it more difficult to work out what one needs to achieve to beat ones
competitors. In an attempt to address this, on the Chipstead website, and also in
the Sailwave Files area, I placed an Excel file that works out exactly what one
needs to do to beat ones competitors.
The other factor is that CHIPS substantially reduces the number of series
ties, and in this respect works better than Art Engel, and of course this is
one of the big problems with LPS that is not handled well by RRS A8.1, 8.2.
Sorry about this long message but I can rant on for hours about scoring
systems.
Kind regards
Geoff
···
-----Original Message-----
From: sailwave@yahoogroups.com
[mailto:sailwave@yahoogroups.com]On Behalf Ofart.engel
Sent: 10 February 2008 04:08
To: sailwave@yahoogroups.com
Subject: [sailwave] High Point
Scoring Systems
I would be interested in a discussion of
some of the more complicated
scoring systems, which all seem to be high point systems. I mean
Cox-Sprague, Renderle B, etc. [I know this User Group isn’t supposed to
be about that. I would happily take the discussion elsewhere if anyone
has a suggestion.]
We have a series of summer races - 20 weeks and 5 or 6 classes of appox.
12-20 boats in each class. Traditionally, we have had a single overall
winner among all the boats and classes (usually 1 one-design class and
the rest handicap classes, PHRF TOD). So, we need a scoring system that
can “fairly” score between and among boats in different classes. In
2007
our biggest start of the year was 18 boats and the smallest was 6. So,
we have to accommodate class sizes that vary from race to race.
In 2002, our club took a look at scoring systems and put together our
own, which we think does a pretty good job. In the last few days, there
was some reference to the Renderle B and Chips systems. At the same
time, our club is questioning whether we might come up with a better
system for 2009 (it is undoubtedly too late to do anything for 2008).
That makes me wonder if there we might be able to improve the system we
are using now.
Our Sailwave formula would be: 1+( ( ( (r-p)+.5)/r ) *9 )
All abbrev. except DNC get 1 point; DNC and boats that don’t enter get
zero points.
[We don’t use Sailwave but I just inputted our scoring system into
Sailwave so that another club in our harbor could use it.]
Basically, our formula tries to score each boat based on its relative
position in the class with boats that come out to the starting area
getting at least 1 point and boats that stay ashore or don’t enter
getting zero points.
I looked at the paper that Geoff Burrell referred to (the link is broken
but the PDF file is available under “Files” in this User Group). I
was
pleased to find that I think our system is “fairer” than either
Rinderle
B or Chips. Of course, what is “fair” depends on what you think the
goal
of a scoring system should be.
Anyone interested?