What Is a Worked Example of Tail Risk?

What tail risk means (definition first)

Tail risk is the risk of unusually extreme outcomes—typically losses—that occur in the “tail” (far end) of a probability distribution. In practice, “tail” means that most outcomes may look ordinary, but a small probability event produces a much larger impact than the average case.

For risk management, the key point is that tail risk is not just “high volatility.” Volatility measures how much values move around; tail risk focuses on how bad the worst plausible outcomes can be, given some probability model.

How the worked example works (mechanism and assumptions)

Below is a worked, scenario-style example. It is not based on live market data.

Assumptions (stated explicitly):

1. There are two outcome types for a simple account exposure over a fixed horizon: a “normal” outcome and a “tail” outcome.
2. The probability of the tail outcome in this horizon is 1% (0.01). The normal outcome probability is 99%.
3. In the normal outcome, the net result is a -2% account loss (loss of 2% of starting equity).
4. In the tail outcome, the net result is a -30% account loss (loss of 30% of starting equity).
5. We ignore interest, financing costs, taxes, and changes in account size during the horizon for simplicity.
6. We assume the percentages apply directly to the same starting equity base.

Compute expected loss (not a prediction—just arithmetic):

• Expected loss = (0.99 × -2%) + (0.01 × -30%)
• Expected loss = -1.98% + -0.30% = -2.28%

Show “tail dominance” of the loss contribution:

• Contribution from tail outcome = 0.01 × (-30%) = -0.30%
• Contribution from normal outcome = 0.99 × (-2%) = -1.98%

In this setup, the tail event contributes 0.30 percentage points of expected loss out of 2.28 percentage points. That may seem modest—but the tail event is also the main driver of the worst-case magnitude (a -30% drawdown), which is often what tail risk frameworks are trying to limit or plan for.

A stricter “tail focus” metric often looks at extreme quantiles or worst-case bands (for example, the loss level associated with the 99th percentile of losses). With quantile-based views, the tail event defines the tail level directly.

Worked scenario example for multiple tail events (what can change)

Now add a second assumption change to illustrate sensitivity.

Revised assumptions:

1. Tail probability doubles from 1% to 2%.
2. Tail loss remains -30%.
3. Normal loss remains -2%.

Expected loss = (0.98 × -2%) + (0.02 × -30%) Expected loss = -1.96% + -0.60% = -2.56%

Tail contribution becomes 0.02 × (-30%) = -0.60%, which is now a larger share of expected loss than before.

Material limitation/failure mode: This simple example assumes the probabilities and loss magnitudes stay stable and are independent of the actions taken and of market structure. In real conditions, extreme moves can also be accompanied by widening transaction costs, reduced liquidity, and execution slippage, meaning the realized loss could be worse than the assumed -30%.

Limitations, risks, and how to verify independently

Limitations of the example:

• Probabilities (like 1% or 2%) are assumptions, not measurements. Real-world tails can be heavier or lighter depending on the market regime.
• The loss outcomes are simplified to fixed percentages. Real account losses depend on price paths, position sizing, leverage, and the exact payoff of the exposure.
• Historical relationships do not guarantee future behavior. A distribution that produced certain extremes in the past may produce different extremes later.

Risks to watch conceptually:

• Model risk: If your probability model underestimates tail frequency or tail severity, tail risk may be understated.
• Liquidity and execution risk: In fast markets, actual fill quality can worsen outcomes.
• Correlation risk: Multiple risks can become positively correlated during stress, turning several “moderate” exposures into one combined tail loss.

How to verify facts (without relying on this example):

• To verify the definition, compare how “tail risk” is described in risk management materials (it should refer to extreme outcomes in the distribution tail).
• To verify arithmetic, recompute expected loss and tail contributions using the stated assumptions.
• To verify applicability, replace the assumed probabilities and losses with your own documented assumptions (for example, from your own scenario analysis) and observe how sensitive results are to those inputs.