Characterising Score Distributions in Dice Games

Isaksen, Holmgaard, Togelius, Nealen

dice randomization probability closeness

@article{isaksen:gapd-2016,
  author={Aaron Isaksen and Christoffer Holmg{\aa}rd and
          Julian Togelius and Andy Nealen}, 
  title={Characterising Score Distributions in Dice Games},
  journal={Game and Puzzle Design Journal},
  year=2016,
  volume=2,
  number=1,
  pages={24--37}
}

Qualitatively works through impact on closeness and predictability of various combinations of numbers of dice, number of dice faces, and tiebreaker rules.

Looks at:

• Expected closeness
• Win bias
• Tie percentage

Closeness and tie percentage decrease with the number of faces

Closeness decreases with the number of dice

Sorting the dice tends to decrease closeness, while unsorted pairs increases ties

Arbitrary tiebreaker creates a win bias

Can tradeoff rolling fewer dice versus one side rolling a few higher sided dice to balance tie percentage versus win bias