[FSPA] League questions - player opinions wanted

[Wei-Hwa Huang]

On Fri, May 10, 2024 at 10:29 AM Dave Hubbard via FSPA <fspa at fspazone.org>
wrote:

> Being able to play everyone (or close as can be) would be great, but I
> think I would prefer a system like Pinburgh where you slowly squeeze the
> gaps between seeds.  Like in a theoretical 40-person league, week 1 might
> be 1-19-20-40, week 2 would be 1-9-10-20, week 3 would be 1-4-5-10...
> something like that that slowly converges to 1-2-3-4, 5-6-7-8, etc... at
> week 10.  Not sure how you administer that with a variable number of league
> players, and smaller leagues are going to break down.  Balanced throughout
> is fine and arguably the most fair, but does have the potential to obviate
> a week 10 playoff push if the top seed happens to draw 3 bottom feeders,
> for example.
>

What you're basically looking for is a Swiss system, with elimination-style
seeding.  One approach for making this work for pinball league play would
be something like this:

Sort everyone by the number of matchpoints they've earned.  Use performance
in past seasons as a tiebreaker.
Split the list into some number of tiers.  One tier for week 1, Two tiers
for week 2.  For a really large group the progression would go 4, 8, 16,
etc., but for a pinball league size probably just doing 1, 2, 3, 4, 4, 4,
4, ... would be better.
Within each tier, pair up the top and bottom.  Repeat until you are out of
people to pair.  Sort those pairs by the best score, and now pair up the
pairs (top v. bottom) to make groups of 4.

If your numbers aren't multiples of 4 you'll need some massaging, but this
general approach should work.


-- 
Wei-Hwa Huang, onigame at gmail.com
-----------------------------------------------------------
Mad scientists are often wrongfully accused of using science for evil.
We're using evil for science; that's totally different!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.fspazone.org/pipermail/fspa/attachments/20240510/2c975611/attachment.htm>