Find another good heuristic to reduce first round

[janLo]

The possible distributions for the first round can be calculated by

>>> import math
>>> def permcount(teams):
...   num = 1
...   for cnt in reversed(range(2,(((teams / 3) * 2) + 1),2)):
...     num = num * ((math.factorial(cnt) / math.factorial(cnt - 2)) / 2)
...   return num
...
>>> permcount(teamcount)

Which gives a 11 digit long number for only 24 teams - thats too much. The Firstround_team_selection already uses heuristics like "clever" sorting to get good solutions fast or slicing by max distance or teamcount per host - but I think it there must be a better solution.