The Fuller Number

It is based on the assumption that RYA PY numbers are correct. If we make the reasonable assumption that for most classes they are not too far off, then this formula may help show up where some anomalies are*, or it may be that the formula is being forced to extrapolate beyond the realms of fairness. I wouldn't expect it to work for a foiling Moth, but it should be reasonable for a Hadron.   

* Unless the anomalous boat has been included in the dataset the regression was done on, in which case the formula is infected.

I haven't run any race results using these numbers, but feel free to do so.

The table below is a set of PY numbers calculated by the formula and compared against 2014 RYA numbers. I have divided by 10 and rounded off, because I can, and because I feel that 4 digit numbers are spurious. The delta column is the difference between the two results and the % Delta is the (absolute) percentage difference, obviously. It is usually within 1-2%, with some clear exceptions. For interest, I looked at the official RYA numbers for each class since 1981 (figures courtesy of JimC) and the last column shows how much variance there has been in the official number over those 34 years (in fact, the vast majority actually happens in the last 4 years for nearly all classes, since the system changed). You can see that variance within the RYA's own numbers is as much as the difference between the prediction and 2014 RYA numbers, so improving the accuracy of the prediction is pointless. Note that that is not the same as saying the RYA PYs are accurate to 2%. The RYA's best guess varies 2% year on year, but could still be 10% out (who knows?).

Jim raises some good points. FWIW, this formula gives a slightly better fit than the one I first posted, and it does use a square root of length term. The difference in predicted numbers is only 0.3% though.

PY = exp((8.102887)-(0.49514*(L^0.5))-(0.03296*SA)+(0.001041*((L^0.5)*SA/Disp)))

I wonder whether a simple fudge factor could help this further. Set 4 or 5 categories of boat:
a. Full on dev boats where every ounce of speed is squeezed out e.g. 14s
b. Highly refined racers e.g. 505, Finn, 800 (generally carbon mast, epoxy hull etc)
c. Standard boats e.g. laser, 200 (ali masts)
d. Beach boats e.g. rotomolds

It looks at first glance that if you set a multiplication factor of say a = 0.96, b = 0.99, c = 1.01 and d = 1.05 you would go some way to factoring the 'goodness' of design (in a purist sense) into the equation.

Hi Peaky
The numbers for the 300 EPS seem at odds or is that just to wind up Grf?
When we had Nick Peters in the 300 and Yves Loday in the EPS one evening the 300 won by miles. That was in Solent chop so I would have thought given the advantage to the easier to sail EPS.
Good luck with the maths.
Gordon

Originally posted by iGRF A handicap changing because some physical alteration to the craft has taken place is one thing,

That's a measured rating system, which the Portsmouth yardstick isn't.

Kipling wrote "There are nine and sixty ways of constructing tribal lays,/And every single one of them is right!", and there are probably 69 ways of managing handicap racing, and they're all correct too.

I don't think the current scheme (which of course is in the middle of major change of philosophy) is perfect, or ideal, or anything else, but I do think its probably the most practical system the sport can manage at the moment, and what follows is my personal understanding/interpretation.

Over the years its changed. It started out in the 1950s simply as an average of the handicaps actually in use at as many clubs as possible, with a means of normalising each clubs data. It moved, fairly quickly I think, to club officers saying what they thought the handicap should be, rather than what they had used that season.

That was how it continued until quite recently, based on the club handicap numbers with various increasingly sophisticated numerical tools to make the club numbers as comparable as possible and things like spreadsheets for the clubs to use to adjust their numbers. The criticism of that over the years was that the clubs had become too complacent and were not putting enough effort into recommending variant numbers.

When I look at the historical data I've put together I think there's a little evidence that there weren't as many numbers changing as much in the 2000s as there were in the 1980s. I suspect part of that is pyschological, because 90 -> 91 feels like a smaller change than 900-910...

Anyway, that's changed now, and the system is in the middle of moving away from the method of numbers based on club recommendations to a new method of numbers based on actual club results - the web system. This brings a whole number of advantages, and also a number of disadvantages, because nothing's perfect. The systems aren't as different as they sound, because the actual functionality behind the web system is using the maths in the tools that were made available to the clubs to calculate the equivalent of club recommended numbers, and then aggregate the club numbers in the same way.

So the new system effectively calculates recommendations for every club, whereas before it was considered that many clubs didn't bother to do the calculations and just returned the standard numbers every year. So a big advantage of the new system is that it much more closely represents what's happening on the water, and a feature is that the opinion of the club officer about what the handicap should be, whether beneficial or otherwise, is reduced in influence.

But the key point about this system is, rightly or wrongly, its not based on the size of the boat, the length of the boat or anything else. Its based on what actually happened on the water over the last 3 seasons. So, unlike all the many Yacht handicapping rules which have come and gone over last 50 years, its immune from designers gaming the measurement system, 'cos there ain't one! On the other hand it does mean its impossible for the RYA team to create handicaps for new boats until there's enough data in the system, and if boats don't maintain a certain level of activity they'll be dropped from the published lists.

So basically, whilst in a measured rating system numbers change because the boats have changed, in an observed performance system, which PYS is, numbers change because the performance on the water has changed, and the system doesn't need to attempt to collect data about boat changes. Nothing wrong with either way of doing things: they're just different.



Edited by JimC - 24 Apr 14 at 4:49pm

Originally posted by gordon1277 That was in Solent chop

OTOH a well sailed 300 ought to be able to slice through a chop an EPS would get stopped by. Hard to judge anything by a single instance.