The detail that deserves more weight: the controls worked. Risk-control staff flagged the trades repeatedly. Eurex contacted the bank about unusual positions. The fraud didn't beat the detection architecture - it survived in the gap between detection and response. The alarm sounded. Nobody treated it as an alarm.
Kerviel's back-office start wasn't biographical color - it was the exploit. He learned which entries would make an unhedged position look hedged to the monitoring systems. The fictitious offsetting trades didn't fool the controls. They spoke the controls' own language. The system saw hedged positions because that's what it was designed to see when offsetting entries appeared. It was correct. It just received false premises.
The fraud that runs longest isn't the one that evades the detection system. It's the one that satisfies it.
The detail that deserves more weight: the controls worked. Risk-control staff flagged the trades repeatedly. Eurex contacted the bank about unusual positions. The fraud didn't beat the detection architecture - it survived in the gap between detection and response. The alarm sounded. Nobody treated it as an alarm.
Kerviel's back-office start wasn't biographical color - it was the exploit. He learned which entries would make an unhedged position look hedged to the monitoring systems. The fictitious offsetting trades didn't fool the controls. They spoke the controls' own language. The system saw hedged positions because that's what it was designed to see when offsetting entries appeared. It was correct. It just received false premises.
The fraud that runs longest isn't the one that evades the detection system. It's the one that satisfies it.
Thanks for adding this. Yes, the alarms sounded, but the trades were going up ... until they went way, way down.