Remove this ad

#21 [url]

Jul 6 16 4:25 PM

Surely it is within your power to arrange a Pugwash Cup involving HMS Clyde and the islands of your choice.

Quote    Reply   
Remove this ad
avatar

TheRoonBa

Posts: 5,508 Site Admin

#22 [url]

Jul 6 16 5:04 PM

I've just discovered a great competition known as the "Navy Cup" between HMS ships and the like, complete with results going back to 2008. I reckon that this will allow me to calculate some sort of ranking for Tristan da Cunha and South Georgia as well as Saint Helena.

After hauling in some Malaysian FC Cup results, Hassanal Bolkiah Trophy U-21 tournament results and a Burma v Burma U-21 friendly match result, I can now (again somewhat shakily) calculate a ranking for both Christmas Island and Cocos Islands (seemingly, they are both slightly better than Micronesia and Kiribati, with Christmas Island slightly better again than Cocos Islands).

Quote    Reply   
avatar

TheRoonBa

Posts: 5,508 Site Admin

#23 [url]

Jul 6 16 7:49 PM

And now, after entering all the "Navy Cup" results (around 200), I can give estimated rankings for Tristan da Cunha (slightly worse than Micronesia) and South Georgia (slightly worse than Sark).

Marshall Islands can be ranked if I can find out the HMS Daring v Kwajalein score from 2013 - HMS Daring rank somewhere just below Niue, so finding the result would give at least some idea of the strength/potential of the Marshall Islands football scene.


Saint Helena can also be ranked if I can find the Falkland Islands v HMS Montrose result of 2012.


Quote    Reply   

#25 [url]

Jul 6 16 11:15 PM

This morning I had a filling at the dentist - I'd say I enjoyed that more than the first half of Portugal-Wales.

Your impressive albeit tenuous ranking extrapolations would be simplified if you just classified all teams in one of the following categories:

1. Reasonably good (e.g. Germany)
2. Solid but impotent (e.g. Switzerland)
3. Decent on a good day (e.g. Hungary)
4. Poor (e.g. England)
5. @%#! (e.g. India)

This kind of ranking has the added benefit of being impervious to actual results, which are unreliable indicators and generally just get in the way.

Quote    Reply   

#26 [url]

Jul 7 16 10:44 PM

nfm24 wrote:
1. Reasonably good (e.g. Germany)
2. Solid but impotent (e.g. Switzerland)
3. Decent on a good day (e.g. Hungary)
After tonight's match I am downgrading Germany to level 2.  France were already in level 1 along with Iceland.

Quote    Reply   
avatar

TheRoonBa

Posts: 5,508 Site Admin

#30 [url]

May 26 17 9:30 PM

This is all very helpful. smiley: laugh


If I take the invisible pill, maybe I can keep the story going?  Or maybe the AlkaSeltzer will destroy it? 


I wish we lived in simpler times...

Quote    Reply   

#31 [url]

May 27 17 10:46 AM

Given that eating only N jelly beans, when there are 2N available, is clearly suboptimal, what happens if you eat both simultaneously?

Matrix-wise, is it something mathematical or computational (or both)?

Quote    Reply   
avatar

TheRoonBa

Posts: 5,508 Site Admin

#32 [url]

May 27 17 11:05 AM

Mathematical, I think.

Problems are teams with all-time records in the form:
Won X Lost N
Won N Lost X
(where N<X and N is either 1 or zero).


So, teams who have played few games, and have never lost, or teams who have played few games, and have never won.

However, the mathematical problem is encased in a computational problem, which makes fixing it harder. 

Apart from this problem, the rankings are looking pretty good.

Last Edited By: TheRoonBa May 27 17 6:50 PM. Edited 2 times.

Quote    Reply   

#33 [url]

May 28 17 5:05 PM

I see. Logically I'd say any ranking of such teams is spurious, unless winning margin is a reliable measure. But what is the matrix aspect? Are you trying to invert a matrix with too many zeroes in it or something like that?

Quote    Reply   
avatar

TheRoonBa

Posts: 5,508 Site Admin

#34 [url]

May 28 17 9:29 PM

I'm solving iteratively to a certain degree of accuracy for a matrix of all matches, and also taking into account winning margin (which is generally a reliable measure, especially in football) as well as "age" and competitiveness of match. Teams who lose all or almost all matches seem to be too low, and vice versa for teams who win all or most matches (as in, a crap team who has won 2 matches against crapper teams may find itself at the top of the rankings, ahead of Brazil, Argentina, Germany et al.) I think it may be a computational problem rather than a mathematical problem, but as I am not an expert in either, it is hard to tell :-) I'll get there!

Quote    Reply   

#35 [url]

May 29 17 12:54 AM

Hmm. If the problem is due to lack of data on some teams, maybe you can fudge round it by (say) adding fictional stabilization terms, in this case putting in "enough" additional results such as a 1-0 win in 2071 BC. The actual impact of which are negligible but which act as a computational AlkaSeltzer.

Quote    Reply   

#36 [url]

May 29 17 1:35 PM

supposition

Your problem sounds like you have too little bias on the strength of the opponent.

If a very crappy team wins against a very very crappy team this should earn far less "Points" than a win against a decent team and 100 losses against decent teams.


(You should not only consider the difference between the teams concerned but also the absolute "Points" the opponent has in the Moment before the match concerned)

Last Edited By: Stevan May 29 17 1:42 PM. Edited 1 time.

Quote    Reply   
Remove this ad
Add Reply

Quick Reply

bbcode help