Online payment fraud: from anomaly detection to risk management

Financial Innovation

Table 6 Simulated moments and quantiles for online banking based on 100’000 Monte Carlo simulations

in CHF 1’000	Mean	Std. Dev.	90%	99%	99.9%
Total no detection	2’023	1’056	3’443	5’083	6’505
Total FPR 0.4 Pct	1’659	892	2’865	4’258	5’524
Total FPR Opt 0.4 Pct	800	535	1’518	2’403	3’114
Beta no detection	358	89	475	588	679
Beta FPR 0.4 Pct	293	81	400	504	585
Beta FPR Opt 0.4 Pct	206	65	291	376	442
GPD no detection	773	537	1’455	2’512	4’000
GPD FPR 0.4 Pct	634	484	1’251	2’248	3’450
GPD FPR Opt 0.4 Pct	81	91	207	361	489
Mass attack no detection	892	903	2’029	3’635	4’826
Mass attack FPR 0.4 Pct	732	743	1’690	2’983	3’920
Mass attack FPR Opt 0.4 Pct	514	523	1’199	2’100	2’799

Total represents the risk figures for all three distributions generalised Pareto (GPD), Beta and mass attack