Skip to content

FRM Exam Part II · Case Study: Model Risk and Model Validation

London Whale and LTCM Model Risk Case Studies

Updated 11 October 2026 · Fact-checked

The London Whale (JPMorgan CIO, 2012) and LTCM (1998) are model risk failures. At JPMorgan, a new VaR model understated risk and was poorly validated. At LTCM, models assumed stable correlations and liquidity. Lessons: independent validation, effective challenge, stress testing, limits, and not trusting a single model.

Understand Model Risk Case Studies: London Whale and LTCM

Model risk is the loss that comes from a model being wrong or used wrongly. It has two sources: errors in the model itself (design, data, implementation) and misuse (applying it outside its intended purpose). Case studies show how both play out in real firms.

London Whale (JPMorgan Chief Investment Office, 2012). The CIO held a very large, complex position in credit derivatives, mainly credit index tranches and CDS indices. Losses grew into billions of USD. In early 2012 the CIO's VaR was breaching its limit. A new VaR model was introduced that sharply lowered reported VaR, which let the book appear within limits. The new model was reportedly built and implemented in spreadsheets with manual steps and contained operational errors. One reported flaw was in how volatility was calculated, which understated risk. Validation of the model was weak and rushed. Reported VaR also did not capture the risk of the position well, and other measures (stress results, limits) were not given enough weight. Traders were also reported to have marked positions in a way that hid losses.

LTCM (Long-Term Capital Management, 1998). LTCM was a hedge fund running convergence and relative-value trades, with very high leverage. Its models were calibrated on limited history and assumed correlations between positions were low and stable and that positions could be liquidated at normal prices. After Russia's default in August 1998 there was a flight to quality. Spreads widened together, correlations rose, and liquidity vanished. Losses were far beyond what VaR suggested. A Federal Reserve-organised private-sector rescue followed.

Common lessons. Models need independent validation and effective challenge. A VaR change should be tested against the old model and backtested before it replaces it. VaR is not enough: use stress tests, scenario analysis and liquidity measures. Correlation and liquidity assumptions fail in crises. Leverage magnifies model error. Governance, culture and limits matter as much as the maths.

Key formulas to remember

Model risk sources
Model risk = model error (design, data, implementation) + model misuse (use outside purpose)
Use this split to classify any case failure.
Effective challenge
Effective challenge = competent, independent, influential review with authority to require change
The core control in model governance (SR 11-7 language).
Loss vs VaR check
Exception if actual loss > VaR at the stated confidence level
At 99% one-day VaR, expect about 1% of days to be exceptions.
Leverage effect
Return on equity ≈ asset return × (Assets ÷ Equity) − funding cost effect
High leverage turns a small asset loss into a large equity loss, as at LTCM.

How to solve Model Risk Case Studies: London Whale and LTCM questions

Use this method for any question on a model failure case.

  1. 1Identify the firm and the model or risk measure involved (CIO VaR model, LTCM VaR and correlation assumptions).
  2. 2Classify the failure: model error (design, data, coding) or misuse (wrong purpose, ignored limits).
  3. 3Find the broken assumption: volatility calculation, correlation, liquidity, stable history.
  4. 4Name the missing control: validation, effective challenge, backtesting, stress testing, limits, escalation.
  5. 5Link to governance: independence of validators, board and management oversight, risk culture.
  6. 6Match the lesson to the option: prefer answers about process and challenge over claims that VaR should be abandoned.
  7. 7Check against traps: do not mix up the two cases or invent details not in the case.

Quickest way: Case-to-lesson matching

When to use it: When you have under a minute per question and the stem names a case or describes a failure.

  1. Spot the clue: a VaR model change that cut VaR means London Whale. High leverage, converging spreads and a 1998 crisis means LTCM.
  2. Ask: error or misuse? Spreadsheet or coding flaw means error; ignoring stress means misuse.
  3. Pick the control that would have caught it: independent validation for Whale, stress testing and liquidity for LTCM.
  4. Eliminate options that say a model alone is enough or that VaR is useless.

Common mistakes in Model Risk Case Studies: London Whale and LTCM

  • Mixing up the two cases, for example saying LTCM changed its VaR model to fit limits.

    Both are famous model failures and details blur.

    Fix: Tie one tag to each: London Whale means VaR model change and weak validation; LTCM means leverage, correlation and liquidity assumptions.

  • Saying the failure shows VaR is useless.

    Case stories sound like attacks on VaR.

    Fix: The lesson is that VaR must be validated, backtested and supplemented with stress tests, not dropped.

  • Blaming only the model and ignoring governance.

    Students focus on the maths.

    Fix: Add the governance failure: weak validation, poor challenge, pressure to stay within limits and weak escalation.

  • Treating a lower VaR after a model change as proof of a better model.

    Lower numbers look like improvement.

    Fix: A model change needs parallel runs, comparison to the old model and backtesting. A drop in VaR should raise questions.

  • Forgetting liquidity and correlation in the LTCM case.

    Students remember only leverage.

    Fix: Include all three: leverage, correlations rising in stress, and inability to exit positions at model prices.

Worked examples

Example 1

A bank's trading unit is close to its VaR limit. Management approves a new VaR model that cuts reported VaR by about 40%. Validation is brief and done by a team that reports to the unit head. Which lesson from the London Whale case is most relevant? (A) VaR should be replaced by expected loss only; (B) Model changes need independent validation and comparison with the old model; (C) Lower VaR shows risk was overestimated; (D) Limits should be removed.

Show the solution
  1. Identify the pattern: a model change that lowers VaR when limits bind, with weak validation. This matches London Whale.
  2. Spot the governance flaw: validators are not independent of the unit.
  3. Test each option. A is wrong because VaR is not to be dropped. C assumes the drop is correct without evidence. D removes a control.
  4. B names the right controls: independent validation and parallel comparison.

Answer: (B)

Example 2

LTCM's models showed low portfolio VaR in 1998, yet it lost most of its capital. Which set of factors best explains the gap? (A) Low leverage and stable correlations; (B) High leverage, correlations rising in stress and illiquid positions; (C) Spreadsheet coding errors only; (D) Too much diversification across unrelated assets.

Show the solution
  1. Recall LTCM: relative-value trades with very high leverage.
  2. Recall the 1998 shock: after Russia's default, flight to quality widened spreads together, so correlations rose.
  3. Liquidity vanished, so positions could not be closed at model prices.
  4. A contradicts the facts. C describes the Whale. D is wrong because the diversification benefit disappeared in stress.

Answer: (B)

Exam tips

  • Know one-line summaries of each case and the specific model flaw: VaR model change and validation weakness for the Whale; leverage, correlation and liquidity for LTCM.
  • Questions often ask for the control that was missing. Answer with independent validation, effective challenge, stress testing or escalation.
  • Do not memorise exact loss amounts. Focus on causes and lessons, which are what MCQs test.
  • Link cases to SR 11-7 ideas: model error versus misuse, validation, governance and effective challenge.
  • Reject extreme options such as abandoning VaR or relying on a single model.

Practice questions from Case Study: Model Risk and Model Validation

Model Risk Case Studies: London Whale and LTCM: frequently asked questions

What went wrong in the London Whale VaR model?

JPMorgan's CIO adopted a new VaR model that sharply lowered reported VaR while its limit was being breached. The model was reported to contain implementation and calculation errors, including in volatility, and validation was weak. Risk was understated.

What is the main lesson of LTCM for model risk?

Models built on limited history can fail when correlations and liquidity change in a crisis. High leverage then magnified the losses. Stress testing, liquidity analysis and leverage limits are needed alongside VaR.

Do I need to remember exact loss figures for the FRM exam?

No. Exam questions focus on causes, the broken assumptions and governance lessons. Know the reasoning rather than precise numbers.

How are these cases linked to model validation?

Both show what happens when validation and challenge are weak or ignored. Good practice includes independent validation, backtesting, benchmarking against other models, and stress testing.