This appendix describes the probability model and the test of the pairing selection of the insurance network's core.

First step: The calculation of each clan's probability distribution Pi (x) is a solution to the classic probability problem of getting a given allocation of colored tiles from repeated single draws from a bag full of colored tiles. Obviously, the higher the number of tiles of a given color in the bag, the higher the probability that the color will be drawn. For the purpose of our test, imagine that each Genoese family is a color and that each of the tiles is an underwriting family tie in a given policy. Thus the most active underwriting clans are more likely to be part of each family distribution of an insurance contact.

The “tiles-drawing probability” problem has a well-known mathematical solution. However, the theoretical solution is not easily operationable for such a large, diversified “bag” of underwriting as the one in our sample. Indeed, our bag contains 1,168 tiles (insurance ties) unequally distributed among 26 colors. Remember that for each clan, one needs to figure out the probability of having a given number of colors (families) for a given number of draws (number of policies contracted).