ThePortForum.com

PhilW wrote:

Either way, enough houses declaring to call this a general declaration, which it makes it the first '1' year to be generally declared in the two centuries since the system began..

I guess 1931 had too few declaring to count as 'general' and 1991 was split 1991/2.

It is interesting looking at the number of times (or % of times) in which years ending with each digit are declared, to see if there might be any decision bias; The following is a rough count of general declarations for the last century (1901-2000, or 1900-1999, whichever you prefer) for years ending with each digit, with split declarations counted as a half - there might be argument about a few of the years as to whether they were 'general' or not, but the idea holds:

xxx0 5
xxx1 0.5
xxx2 1
xxx3 1
xxx4 2
xxx5 4
xxx6 1
xxx7 3.5
xxx8 1.5
xxx9 0

Overall therefore, it seems far more likely to declare in years ending in 5 and 0.. possible decision bias, though I'm sure the case could be argued for this being a coincidence, especially with 7 also have a decent count. Ok, so lets have a look at the 19th century instead:

xxx0 5
xxx1 2
xxx2 1
xxx3 3
xxx4 3
xxx5 2
xxx6 1
xxx7 3
xxx8 3
xxx9 0

These totals use TPF's list of declared years, as I do not know whether any of them were/were not general declarations. Again, years ending in zero are substantially ahead, though 5's do not feature so prominently. Interestingly again no years ending in 9 declared in this century either.

RAYC wrote:Uncorked have published their pricing on the Symington 2011.

uncle tom wrote:The first thing one notices is the total absence of consecutive declarations. Are the vines so exhausted after a good year, that they cannot repeat the exercise without taking a rest? The answer, I think, is no.

jdaw1 wrote:If 30% of years are declared, what is the probability that a given year is part of a consecutive declaration?

Easy maths. Pick a year that is declared: say x. There is a 70% chance that x”“1 wasn’t declared, and a 70% chance that x+1 wasn’t declared, so a 49% chance that neither neighbour was declared. Hence a 51% chance that one (42%) or both (9%) were declared.

Thus if 30% of years are declared independently of other years, about half of all years should be part of a consecutive (not split, consecutive) declaration. Which suggests that a year being declared is not independent of neighbouring years being declared. As Tom said.

jdaw1 wrote:

uncle tom wrote:the total exclusion of '9' years - try working out the probability of that happening by chance

Let’s assume that each year has a one-in-four chance of being declared, all years identically independently distributed. Then the probability that, during some particular century, no 9s are declared is (3/4)^10 ≈ 5.63%, Improbable, but not special. But the wrong question. What is the probability that there is a digit, whether 9 or something else, such that in some particular century, no year ending in that digit is declared? The answer is obviously 1 - ( ( 1 - ((3/4)^10)) ^ 10 ) ≈ 43.988%. That’s quite likely.

So the existence of a digit such that, during the last century, no years ending in that digit were declared, proves, in round numbers, nothing.

Let me rephrase: what is the probability that, during the nineteen-eighties, only ’80, ’83 and ’85 were general declarations? Answer: 0.2086%. Less than one percent?! Does that prove a conspiracy? Well, what is the probability that, during the nineteen-eighties, exactly three years were general declarations? Answer: 25%. So be careful about probabilities: a more specific question (‟nines?”) has a lower probability, whereas a question better capturing the whole class of perceived unusualness (‟some digit such that!”) has a higher probability.

jdaw1 wrote:

uncle tom wrote:two recorded centuries when there has been no signficant declaration with a year ending in a 9, and that makes the odds of no '9' years come out at something less than 1000:1

Assuming a constant declaration frequency of one-in-four, the probability that there is a digit such that, over two particular centuries, no year ending in that digit has been declared, is 3.126%. That alone fails a 99% test. And if the probability of declaration was constant at one-in-five, there is a 10.949% chance of observing this outcome, and one-in-six takes that to 23.2%.

So the hypothesis that there is a shippers’ prejudice against some particular digit is statistically unproven.

Much stronger is the hypothesis that they don’t like consecutives.

Which sevens? ’27, ’77, ’97? 1907 no, ’17 I don’t know, ’37 no; ’47 not generally, ’57 no, ’67 not generally, ’87 not generally. Three or four in the last ten decades. Snorey dull.