Operational and Model Risk
Operational risk comes from failed processes, people and systems; model risk from wrong or misused models. Learn real examples and the key controls.
Some of the largest trading losses in history had nothing to do with markets moving the wrong way in ordinary fashion. They came from a software deployment gone wrong, a rogue trader hiding losses, a fat finger order or a model that everyone trusted but nobody truly understood. Operational risk is the risk of loss from failed internal processes, people, systems or external events. Model risk is the risk of loss from models that are wrong, misapplied or misunderstood. Both are harder to measure than market risk, and both can be catastrophic.
Operational risk categories#
| Category | Examples |
|---|---|
| People | Errors, fraud, rogue trading, key person dependence |
| Processes | Missing controls, poor reconciliation, settlement failures |
| Systems | Software bugs, outages, cyber attacks, data errors |
| External events | Power failures, natural disasters, pandemics, vendor failures |
Banking regulation, through the Basel framework, requires banks to hold capital for operational risk.
Famous operational failures#
| Event | What went wrong | Lesson |
|---|---|---|
| Barings Bank, 1995 | Nick Leeson hid losses of about £827 million in a secret account; the bank collapsed | The Fall of Barings Bank |
| Société Générale, 2008 | Jérôme Kerviel built unauthorised positions; unwinding them cost about €4.9 billion | Lessons From Market Failures |
| Knight Capital, 2012 | A faulty deployment sent erroneous orders, losing about $440 million in 45 minutes | Risk Controls and Kill Switches |
| Fat finger errors | Many cases of mistyped order sizes or prices causing large moves | Alerts, Error Handling and Reconnection |
The common thread is weak controls: separation of duties, reconciliation, limits and monitoring.
Model risk#
| Source | Example |
|---|---|
| Wrong assumptions | Assuming normal returns when tails are fat. See Fat Tails |
| Estimation error | Unstable correlations or volatilities. See Covariance and Correlation |
| Overfitting | A backtest that captures noise. See Overfitting and Curve Fitting |
| Implementation errors | Bugs in pricing or risk code |
| Misuse | Applying a model outside the conditions it was built for |
| Stale models | Relationships that have changed. See Structural Breaks and Regime Changes |
Controls for operational risk#
- Separation of duties: traders do not control settlement, accounting or risk reporting.
- Daily reconciliation of positions and cash with independent sources. See Trade Accounting and Reconciliation.
- Pre trade limits and kill switches. See Risk Controls and Kill Switches.
- Change management for software, with testing and rollback.
- Audit trails of all actions. See Logging, Audit Trails and Incident Response.
- Business continuity and disaster recovery plans. See Failover, Backups and Disaster Recovery.
- Mandatory leave for staff, which can expose hidden activity.
Controls for model risk#
US banking regulators' guidance known as SR 11 7 (2011) set out widely followed practices:
- Independent validation of models before use and periodically.
- Documentation of assumptions, limitations and intended use.
- Ongoing monitoring comparing model outputs with outcomes, such as VaR backtesting.
- Model inventory with owners and approval status.
- Conservative use where uncertainty is high, with overrides documented.
For individual traders#
Individuals face the same risks at small scale: an order typed with an extra zero, a bot with a bug, a spreadsheet error in position sizing, or trusting a backtest too much. Double checking order tickets, using platform limits, testing code and reviewing assumptions are cheap protections. See Pre-Trade Checklist.
Frequently asked questions#
What is operational risk in trading?#
The risk of losses from failed processes, human error or misconduct, system failures or external events, rather than from market moves.
What is model risk?#
The risk of losses from models that contain errors, rest on wrong assumptions or are used outside their intended purpose.
How do firms reduce operational risk?#
Through separation of duties, reconciliation, limits, kill switches, change management, audit trails and continuity planning.
Next, learn about risks to the whole financial system in Systemic Risk.
3 quick questions on this lesson. Get them all right to finish it.
Turn on JavaScript to take the quiz.
Mentioned in
- Liquidity RiskPortfolio and Performance
- Statistical ArbitrageStrategies and Styles
- Bridge and Smart Contract RiskCrypto
- Poisson and Exponential DistributionsMath and Statistics
- Neural Networks and Deep LearningMachine Learning
- Online LearningMachine Learning