Member-only story
Game and Auction Theory for Data Scientists
Part 4 of 6: Games of Incomplete Information, what changes when the other side knows something you cannot observe: adverse selection, strategic responses to a published model, and the bid function that leads to auctions. Python code provided.
Diogo Ribeiro is a mathematician, researcher and lecturer at the Faculdade de Media Artes e Design, Universidade Técnica do Porto. ORCID 0009–0001–2022–7072. GitHub diogoribeiro7. Gitlab diogoribeiro7.
In the first three parts of this series every player knew the game. The firms in Part 1 knew each other’s costs, the fraudster in Part 2 knew what every channel was worth, and the cooperative games of Part 3 were defined by a table everyone could read.
Most situations a data scientist models are not like that. A customer knows how risky they are and the insurer does not. An applicant knows how easily they could move a score and the lender does not. A bidder knows what an object is worth to them and the other bidders do not. In each case one side holds information about itself that the other cannot observe, and acts on it.
This part introduces the framework for such games and works through three of them, in order of increasing symmetry. In the first, an insurer reprices each year on the customers who bought the year before, and the share of the population it insures…
