Following the work of F. Pozzi et al.  who provided evidence that stocks lying at the periphery of a stock network tend to outperform those lying at the center, in this work we attempt to create an algorithmic stock selection and portfolio construction strategy based on stock network analysis. To achieve our goals, we do create two portfolios including N equally weighted central and peripheral stocks for various numbers of stocks included in each portfolio and for varying portion of money invested in the aforementioned portfolios aiming to increase diversification and decrease risk exposure. Our results appear to be promising since throughout back testing our strategy, we achieved a return of 41.6%, investing in a total number of 10 stocks, over the period 2004-2005, at the same time that S&P 500 performed 9.3% and our benchmark portfolio had 12.9% return. In fig. 1. we present some preliminary results that depict the performance of our approach versus the benchmark portfolio. For the purpose of our research we used data from Thomson Reuters DataStream database for 185 stocks traded continuously on S&P 500 over the period 2002-2012. Our final results will also include the performance of our strategy during periods of financial crisis and turbulence.