FRM Part II · FRM Exam Part II · Case Study: Financial Crime and Fraud
A bank's fraud analytics team tests an alert model on 10,000 transactions, of which 200 are truly fraudulent. The model flags 160 of the fraudulent transactions and also flags 490 legitimate ones. What proportion of all flagged transactions are actually fraudulent (precision)?
Precision is about 24.6%. The model flags 650 transactions in total, 160 true frauds plus 490 false alarms, so 160 divided by 650 gives 24.6%. The 80% figure is recall, which measures the share of actual frauds caught, not the quality of alerts.
- A80.0%
- B24.6%Correct
- C32.7%
- D1.6%
Explanation
Total flagged = 160 + 490 = 650. Precision = 160/650 = 24.6%. The 80% figure is recall (160/200), which uses the wrong base. 32.7% is 160/490, ignoring true positives in the denominator.
Did you get it right without looking?
One question tells you little. A timed set on Case Study: Financial Crime and Fraud shows your real accuracy, how long you take and where you lose marks.
More Case Study: Financial Crime and Fraud questions
- A bank's transaction monitoring system generated 2,000 alerts in a month. Investigators closed 1,960 as false positives and filed suspicious…
- A bank detects that a customer deposits cash just under the reporting threshold across many branches, wires the funds to shell companies abr…
- A bank's compliance officer is reviewing its customer due diligence program. A new corporate client is a holding company whose shares are ow…
- A bank has 40 critical business services. After a payments fraud incident forced a systems shutdown, management sets an impact tolerance for…
- A bank applies a risk-based approach to AML. A client is a politically exposed person (PEP) who is a senior official in a high-corruption-ri…
- A compliance officer notices a customer making many cash deposits of just under the reporting threshold across several branches in one week,…