A Hierarchical Approach to Quantitative Portfolio Optimization for Technology Development Project Portfolios (OPTIM-H)

Abstract

The use of Mean-Variance Portfolio Optimization (MVO) in Modern Portfolio Theory (MPT) has been a long-standing method to guide investment decisions for market-traded assets like stocks and bonds. Recent research shows that portfolio optimization developed using MPT could prove useful in investment decisions for technology projects. Traditionally, empirical data from past projects and statistically driven technology trends are used to predict the risk-return model necessary for MPT. This thesis introduces a new methodology, Optimizing Portfolios in Technologies Investments Methodology with Hierarchy (OPTIM-H), which extends MPT to make investment decisions within a hierarchical organizational structure of technology projects. An integrated dataset was developed to demonstrate this methodology, combining 19,000 data records from Techport and Small Business Innovation Research (SBIR) datasets. The dataset captures investment trends and maturity pathways across 17 taxonomy areas, revealing that most projects begin at Technology Readiness Levels (TRLs) 2–4, with average funding amounts near \$300,000. OPTIM-H effectively distinguishes between broader technology groups and their subcategories, showing the impact of community interest on investment decisions. Furthermore, this work investigates k-means clustering as a tool for classifying technology projects for targeted investment, with the analysis identifying seven clusters and achieving a mean utility score of 0.595 with a standard deviation of 0.651.