Individualized prognostic predictions in people at clinical high risk are crucial to tailor suitable interventions and personalized prevention. Furthermore, in recent years, the synergy between fast-pace technical sophistication in neuroscience (e.g. neuroimaging and neurophysiological) and novel bio-statistical tools (e.g. machine learning algorithms) has accelerated the development of more inclusive predictive models and magnified the potential for such individualized risk stratification enriching classical psychopathological tools. However, the clinical translation of such research insights is still circumscribed and, despite incremental optimization of assessment tools, increasingly accepted criteria to characterize at risk mental states and tumultuous advance in the field, the prediction of psychosis at such individual level remains a not fully accomplished target.