Jump Risk: Sudden Discontinuous Moves That Break Continuous Models

Key Takeaways

• Jump risk captures the fat-tailed, discontinuous moves in asset prices that pure Brownian motion assigns near-zero probability to but that happen in real markets every few years.
• Merton's jump-diffusion adds a Poisson jump process to standard diffusion, producing the steep near-term implied volatility skew observable in equity options markets.
• Delta hedging is not jump-neutral: when a stock gaps 15%, the put's intrinsic value jumps to $15M while the long equity hedge moves only $7.5M, leaving a $7.5M net loss on a nominally delta-hedged book.
• Using 99% one-day VaR as the sole tail metric is especially inadequate for jump risk, because VaR ignores precisely the worst 1% of outcomes where jumps are most damaging.

What It Is

In quantitative finance, a jump is a discontinuity in the price path that cannot be reached by continuous movement. Merton introduced jump-diffusion in 1976 to capture the fact that equity and option prices sometimes move much more than a normal model can explain.

Jump risk shows up in two main forms. Diffusive jumps are large moves caused by information arrivals (earnings, rate decisions, geopolitical events). Default jumps are instantaneous losses caused by a credit event that drops a bond or CDS from par to recovery in one step.

The Intuition

Pure Brownian motion assigns vanishingly small probability to a 10 standard deviation move. Real markets produce those moves every few years. The reason is that information does not arrive smoothly and dealers do not always quote tightly. When an announcement hits, the clearing price is set by an auction at the new level, not by trading through every price in between.

A jump process captures that behaviour by adding rare, large shocks on top of the everyday small moves. The risk management consequence is that any measure built on continuous paths (standard VaR, delta hedging, stop orders) underestimates loss from these events.

How It Works

The Merton jump-diffusion model writes the price process as:

dS/S = (mu - lambda k) dt + sigma dW + (J - 1) dN

Where:

• dW is Brownian motion (continuous part)
• dN is a Poisson counter with intensity lambda (jumps per unit time)
• J is the jump multiplier, often lognormal with mean k + 1
• The drift correction (-lambda k dt) keeps the expected return equal to mu

Parameters are estimated from returns using maximum likelihood or from option prices by calibrating to the implied volatility surface. Steep near-term skew and term structure of implied vol contain a lot of information about expected jumps.

In risk measurement, jump risk appears as:

• Volatility smile. Out-of-the-money puts are priced at implied vols well above at-the-money, because the market is pricing the left-tail jump.
• FRTB Default Risk Charge. A Basel capital add-on for jump-to-default exposures held in the trading book.
• Expected shortfall (CVaR) at 97.5 percent. Replaces 99 percent VaR under FRTB specifically because ES better captures jump-driven tails.

Worked Example

An equity options desk is short 10,000 contracts of a one-month at-the-money put on a large-cap stock. The Black-Scholes delta is -0.5, so they appear hedged after buying 500,000 shares.

A jump-diffusion calibration of the same surface reports an expected 15 percent downside jump with intensity 0.25 per year (about one every four years). A simple scenario: jump down 15 percent.

• Put payoff: roughly 15 percent intrinsic value on 10,000 contracts, 100 shares each. Payoff owed to counterparties: 15 percent of 100 stock price times 1 million shares, about 15 million.
• Delta hedge payoff: 500,000 shares long, down 15 percent, losing about 7.5 million.
• Net loss on the short put book: 15 million paid out minus about 7.5 million on the hedge, roughly 7.5 million.

Under a pure diffusive model the expected loss in this window is much smaller because the probability of a 15 percent move is tiny. The jump-diffusion model assigns it meaningful weight, which is why option prices carry a jump premium and why desks buy tail hedges even when pure-diffusion hedging looks complete.

Frequently Asked Questions

Q: What is jump risk in simple terms? Jump risk is the probability that an asset's price will suddenly move much more than its normal daily range suggests, not a gradual drift but a discontinuous leap. It explains why options on even "boring" stocks price more downside protection than a normal distribution would justify.

Q: How does jump risk affect investment decisions? It requires hedging beyond delta. A delta-hedged short-put book looks fully protected against small continuous moves but takes a large loss when the underlying jumps. Jump risk demands gamma and vega management, or explicit tail hedges, not just delta neutrality.

Q: What is a real-world example of jump risk? An options desk is short 10,000 contracts of a large-cap one-month at-the-money put, delta-hedged with 500,000 shares. A 15% overnight gap down produces about a $7.5M loss on the short put beyond what the delta hedge recovered, the gap was large enough to expose gamma and higher-order risk completely unhedged by delta alone.

Q: How can investors account for jump risk in their risk measures? Use expected shortfall at 97.5% instead of 99% VaR, ES averages the tail rather than ignoring it, and is far more sensitive to jump-driven losses. Calibrate jump intensity and size from option implied volatility surfaces, which are forward-looking, rather than from short historical return series.

Q: How is jump risk different from gap risk? Jump risk is a probability model concept, the Poisson-process component of a price model that generates occasional large shocks. Gap risk is the market-structure realisation of that model, what happens when a jump occurs during a market closure and prices reopen at the new level with no intermediate trades. They are two views of the same underlying phenomenon.

Common Mistakes

1. Calibrating only to realised returns. Historical series are often shorter than the jump arrival rate. A ten-year sample may contain zero jumps. Option-implied calibration usually produces more stable estimates because the market prices the expectation forward.

2. Treating delta hedging as complete. Delta captures the first derivative. A jump is a large move, where gamma, vega, and higher-order Greeks all matter. Delta-neutral is not jump-neutral.

3. Using 99 percent one-day VaR as the tail metric. By construction, a 99 percent VaR ignores the worst 1 percent, which is exactly where jumps live. Expected shortfall at 97.5 percent, the FRTB choice, averages the tail and is more sensitive to jumps.

4. Ignoring cross-sectional jumps. In a risk-off event multiple names jump together. A book that looks diversified on standalone jump intensity may still suffer a large loss because the jumps are correlated.

5. Selling tail risk as a yield product. Short-volatility strategies that earn the jump premium in quiet markets bleed out quickly when jumps hit. February 2018 short-volatility ETPs blowing up and March 2020 option market-making losses are recurring examples.

Sources

1. Basel Committee on Banking Supervision. "Minimum capital requirements for market risk (FRTB)." BCBS d457. https://www.bis.org/bcbs/publ/d457.htm
2. CFA Institute. "Quantitative Methods and Risk Management." https://www.cfainstitute.org/membership/professional-development/refresher-readings
3. Federal Reserve. "Finance and Economics Discussion Series on Jump Diffusion." https://www.federalreserve.gov/econres/feds/index.htm
4. Bank for International Settlements. "Working Papers on Asset Price Dynamics." https://www.bis.org/publ/work_papers.htm

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.