A measure of statistical complexity based on predictive information with application to finite spin systems

► We introduce 'binding information' as a entropic/statistical measure of complexity. ► Binding information (BI) is related to earlier notions of predictive information. ► We derive upper and lower bounds of BI relation to entropy and multi-information. ► Parity processes found to maximise BI in finite sets of binary random variables. ► Application to spin glasses shows highest BI obtained with high-order interactions.