Image logo Bob conference 2018.

Talk: 16:35-17:20 (English)

Solving NP-Hard Problems: An Example of Universal Portfolios

The portfolio selection is an interesting problem: it can be considered as a combinatorial optimization, a variation of the knapsack problem, as a Mean-Variance optimization problem, or as a neural network problem. It can be solved in a variety of ways, using an adapted algorithm from the knapsack problem, or as weights for the neural network.

In this talk, we explore the portfolio selection problem by thinking of it as an adapted knapsack problem, and optimize the portfolio asset allocation between multiple risky assets and risk-free asset process using reinforcement learning techniques. With this approach, the investment weights between the multiple risky assets and risk-free asset are computed, which also solves the asset allocation problem.

Jenny Hung

Jenny is a passionate data scientist who love to get her hands dirty with data and code. Jenny studied Mathematics and Statistics at Simon Fraser University in Canada and went onto work on her Masters in Computer Science from Georgia Tech. She believes the reason data cannot be converted into actionable intelligence reflects only our lack in mathematical /statistical understanding, and not the limitation of statistics and mathematics itself. Jenny’s interests are in machine learning, data mining, applied statistics, and pattern discovery.