The applied literature on illicit (“dirty”) money flows increasingly uses gravity equations as a workable empirical strategy to control for characteristics of both “source” and “destination” countries. This literature lacks reliable data as well as solid theoretical underpinnings. We contribute in two ways: i) we exploit the theory of portfolio investments to empirically derive a global picture of anomalous (i.e. unpredicted) money flows over time; ii) we correlate these anomalous flows with money laundering determinants. Our results show: i) theoretical based financial gravity models are a sound way to look at “dirty” money flows; ii) non-compliance with international transparency standards affects the probability of observing anomalous flows in global financial data; iii) a global map of risk (i.e., anomalous financial activities) can be derived from official statistics; iv) moving from the notion of “risky countries” to that of “risky financial bilateral flows” allows to better investigate the push and pull factors of anomalous flows. The full dataset of bilateral anomalies in portfolio investment flows resulting from our empirical work is made freely available for further research.