At GP's, the super sunday series are capped at 409 players to maintain the 9 round limit. I have been told that in this scenario everyone with x-1 or better should be top 8. However, if I calculate with swiss triangle, this is the result:
1 players at 9-0
8 players at 8-1
28 players at 7-2
68 players at 6-3
100 players at 5-4
100 players at 4-5
68 players at 3-6
28 players at 2-7
8 players at 1-8
0 players at 0-9

Am I missing something here?

At 409, it's *possible* that there will be 8 players at 8-1 and 1 at x-0,
but unlikely. Your calculation assumes that pair-downs will always win.
Notice how the bottom of your bracket has only 8 players total between 0-9
and 1-8? That's the other possibility if you assume that pair-downs will
always lose.

Even if I do a simple calculation that ignores the possibility of early
draws, I get about a 20% likelihood of there being 8 x-1s and 1 x-0.

I don't know exactly what math they've used to get their numbers, but
basically they arrive at some acceptable percentage where x-1 or better
*should* make top 8.

2016-11-02 6:17 GMT-04:00 Olivier Wattel <

A few notes on this:


- This also exists at the 8-to-9 round cutoff, at 225-226 players.


- There's no official documentation that states that an X-1 record will guarantee anything at any event. Top 8 cuts cut to the top 8 players, not a record. (yes, we use X-1 as a guide, but it's currently a guide in the same way that we say matches are “best-2-out-of-3” when that isn't technically correct either)


- I rigged up a Swiss triangle a while back which could do random distributions of pair-down wins (i.e. it could be set so that a pair-down lost a percentage of the time) and found:

Given a 410-player tournament where pair-downs lose 50% of the time, approximately 5.7% (5676 in 100k simulations) end Round 9 with more than eight players at X-1 or better.

Given a 410-player tournament where pair-downs lose 30% of the time, approximately 47.2% (47164 in 100k simulations) end Round 9 with more than eight players at X-1 or better.

Given a 226-player tournament where pair-downs lose 50% of the time, approximately 2.4% (2401 in 100k simulations end Round 8 with more than eight players at X-1 or better.


- I found the “X-1 Guarantee” round thresholds to be 8-16-32-64-128-224-384-736


- I also did some work during this time in order to find the effect of byes on these thresholds, in order to see if there were better “multipliers” for these players. (the last paragraph of Appendix E of the MTR) The problem with byes is that any additional player will affect the total number of pair-down matches in an idealized (no-draws) triangle, so the effect of adding a bye to a player can change the effect on a tournament in an unpredictable manner. (at least, at the level of math that I'm familiar with)


More information in this document. (from a proposal submitted back in February of this year)

Edited Jordan Baker (Nov. 2, 2016 08:51:02 AM)

I sent a similar proposal to Jordan's to WotC at the end of last year (but much less comprehensive, Jordan yours is amazing!)

The short version of the response was that they're aware of the potential issue, but that it would be a big deal involving a lot of parties inside WotC to get it updated in all of the many places it appears, and messaged correctly to everyone running events. Missing any of these updates would open the door to “bad times”.

I believe it's unlikely their position has changed in the last year, but you can always ask again.