On the performance of network science metrics as long-term investment strategies in stock markets

Abstract

Firms and individuals have always searched for investment strategies that perform well and are robust to market variations. Over the years, many strategies have claimed to be effective but few resist the effect of time, that is, most of them become outdated. It turns out that markets have a “self-correcting ability”; the secretive/novel nature of strategies firms employ cannot win forever; other firms eventually implement competing strategies causing the market to adjust. Nowadays, most investment firms “sell” to their clients two approaches: high reward and low reward. Unfortunately the possibility of high reward is generally coupled with low robustness (volatility) and if one wants high robustness the yields are low (low reward). In this paper, we use an approach based on network characteristics extracted from historical market data. Network Science has argued that all complex systems have an underlying network structure that explains the behavior of the system. With this in mind, we propose a long-term investment strategy that builds a network from historical investment data, and considers the current state of this network to decide how to create portfolios. We argue that our approach performs better than standard long-term approaches.