Quick PY Question

Oh no I only opened it to get one out and now the bloody things fallen over...

3 or 4 individuals surely? Otherwise it's nothing short of a personal handicap?

Originally posted by pondmonkey 3 or 4 individuals surely? Otherwise it's nothing short of a personal handicap?

It would have to be more than 4.
And it would have to be random, uncorrelated individuals for such a small sample to reflect any sort of potential of the boat.
The trouble is, people do not buy boats at random, the people who are more serious about racing are very likely to buy into a class or fleet. And they will buy what their friends have, so you get a group of like-minded people racing the same class.

But it is always easy for some spreadsheet operator with no clue about statistics to believe they can put everything right, justifying gross changes with a belief that their local water is strangely different from any other in the UK.

When you sail at a small club, with only 4 or less of most classes sailing, people need to accept the numbers and get on with the racing. Once the numbers are up for debate, it gets personal, and a significant fraction will be upset.

Each year we get changes in the numbers, largely I suspect caused by flawed analysis and data skew.
The RS400 vs Laser 4000 has changed in the same direction the last few years. My boat hasn't changed, the 4000 has not changed, so either some of the numbers are wrong or the uncertainty is far greater than the changes.

 Of course the RS facelift does not exactly add clarity to the situation, but my boat is unchanged. SMODs may have step changes in their rules, more open classes will develop continuously to a greater or lesser extent. Obsolescent classes get slower on average as the sails etc age.

Originally posted by SoggyBadger I'm extremely sceptical about the validity of these 4-dgit numbers even given the sample size the RYA have to work with.

I think the source of the problem was that back in 1980 or whenever it was this was going to be the second renumbering in not too many years because they'd got too little granularity at the low wne of the scale. So by multiplying the numbers by 9.something they made d**** sure that wasn't going to happen again. But I quite agree that one point adjustments are pretty much meaningless. Normally we run a minimum of 5 point adjustments at my club.

Originally posted by Peaky http://www.sciencebuddies.org/science-fair-projects/project_ideas/Soc_participants.shtml You have every right to be sceptical. To have a 95% chance of being within 1% of the true PY, you need a sample size of 10,000. To get to single digit accuracy on a 4 digit PY scheme you need 1 million samples.

Great find Peaky. So it seems that at club level even if you could get 10 boats (and that would have to be 10 boats of each class entered in the race) then you'd still have a margin of error of just over 30%.