This paper presents a novel system for producing the optimum power output from photovoltaic (PV) arrays using dynamic cell reconfiguration. The proposed approach is the first in the literature that creates strings using individual substrings that have been characterized and categorized ensuring maximum power extraction for a given irradiance profile. This optimized and decentralized PV architecture can produce significantly more power than a static equivalent (by an average of 22.6%) and also outperforms the sophisticated alternative known as an irradiance equalized dynamic photovoltaic array (IEq-DPVA) by an average of 13.7% for the relevant tests carried out. This paper identifies the hardware requirements to produce such a system and it describes an algorithm that performs the optimized-string reconfiguration strategy. Finally, a simulator programmed in MATLAB is used to compare the performance of the optimized-string DPVA against an IEq-DPVA in a series of flexibility tests.