This is a fair dice. NOT as fair as a casino dice, but 3x fairer (is that a word?) than the real store-bought dice I tested it against. Now to explain the math and testing. 

Dice have 6 sides. This means that in the perfect world, each side would get rolled 16.667% of the time. Casino dice, which are very accurate, are off by an average of 0.5% (standard deviation). Store-bought dice aim to be off by 1-3% standard deviation. My dice (when I tested it on the published profile) was off by 0.96%. Pretty cool, right? I'll explain my methods now.

The pips on a dice take out a little bit of material from the dice. A standard 16mm dice has 4096 mm^3 of material, and the pips on my dice take out 187 mm^3. That sounds like a lot, but it's only 4%. After that, the center of mass isn't quite in the center. Most are off by about an average of 0.05mm, which is what I measured mine at. The real dice I tested weighed about 6.1g, while mine weighed 4.8. That doesn't matter as much, dice range from 4-12g. 

Out of 656 rolls, the store bought dice rolled:

101 ones (15.4%)

120 twos (18.29%)

119 threes (18.14%)

73 fours (11.13%)

135 fives (20.58%)

108 sixes (16.46)

with an average standard deviation of 2.96 (population)

3D printed dice out of 757 rolls…

114 ones (15.06%)

124 twos (16.38%)

126 threes (16.64)

135 fours (17.83%)

127 fives (17.83%)

131 sixes (17.305%)

with a standard deviation of 0.86% population.

Here's pictures of all the versions, the green being the official and the others being before. The painted one is real.