Lead PI: Dr. David Edelman
This work package aims to utilise the formalised Information Theory framework (Shannon) to approach the problem of investment under uncertainty. While this approach is not new (originating from Kelly (1956)), there is nevertheless a growing ‘minority’ view that Information Theoretic analysis can yield canonically optimal solutions to many aspects of the classical investment problem, with many unexplored avenues ripe for exploitation. For example, the Information-Theoretic strategy for Investment in simple, well-understood independently repeated games or investments is known to minimise the average waiting time to any fixed financial wealth goal (Edelman (2000)). But what about time-varying games? What about games in which only partial information about probabilistic models is available? Is there a way to transform non-stationary sequences of non-stationary investments or games into stationary ones? If people act optimally in an Information-Theoretic sense within a given type of market, what will the result be? Can one invert this process to infer market sentiment from traded prices? None of the above questions has a well-developed answer in literature, and some of the best progress on a selection of these questions results from ongoing work in this WP.