All-the-Time Tail-Risk Management
Build for the extreme periods hidden by ordinary averages
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 96%
Start from the premise that investors experience every market regime, not merely the statistically typical one. Identify the variables that can move discontinuously—such as home prices, rates, liquidity, or correlated defaults—and forecast a broad distribution of outcomes. Include severe scenarios grounded in history even when recent data makes them appear unlikely. Then test whether collateral coverage, repayment, financing, and operational liquidity survive those paths. The goal is not to predict the exact crisis or constantly bet on catastrophe; it is to avoid structures that work only when ordinary averages persist. Protection must be arranged before stress becomes obvious because markets resist buying an umbrella until it is already raining.
Origin
Dobson described this as a core Amherst risk-management culture shaped by experiencing repeated supposedly rare shocks during his career.
Core principles
- 01Typical conditions do not represent the full lived investment experience.
- 02Rare shocks occur more often than labels such as hundred-year event imply.
- 03A range of outcomes is more useful than a single prediction.
- 04Leverage that survives ordinary volatility may fail during a tail event.
How to run it
- 1
Find the Fragile Variables
Identify prices, rates, correlations, liquidity assumptions, and behaviors that could change sharply.
Pro tip Focus on assumptions that support leverage or solvency.
Watch out Stable recent data can make a fragile variable look harmless.
- 2
Build a Range of Outcomes
Create ordinary, adverse, and extreme scenarios using historical events and stochastic paths.
Pro tip Include combinations of shocks rather than testing each variable alone.
Watch out Do not confuse a scenario with a point prediction.
- 3
Trace Transmission
Follow each shock through collateral values, borrower decisions, cash flows, financing, and reported losses.
Pro tip Model timing lags as well as ultimate loss severity.
Watch out Delayed accounting recognition can disguise an existing impairment.
- 4
Test Survivability
Determine whether the portfolio retains sufficient capital and liquidity in every plausible path.
Pro tip Set explicit limits for losses, margin calls, and refinancing needs.
Watch out An attractive average return cannot compensate for insolvency in one plausible path.
- 5
Prepare Before Stress
Reduce leverage, diversify, or buy protection while markets remain liquid and complacent.
Pro tip Review protection costs against modeled tail losses.
Watch out Waiting for confirmation usually makes hedging expensive or unavailable.
In the wild
An 80% loan-to-value mortgage appears protected by the borrower's 20% equity. A scenario in which the home falls from 100 to 65, however, leaves the loan underwater and changes the borrower's repayment incentives. Modeling that path reveals risk that an average annual appreciation assumption misses.
→ The lender sees why apparently conservative leverage still requires tail-loss pricing.
Common mistakes
Living inside the average
Designing for typical annual moves ignores the occasional regime that determines long-term survival.
Predicting one exact future
A single forecast hides uncertainty and encourages exposures that cannot tolerate alternative outcomes.
Searching for crises everywhere
After finding one major dislocation, investors can overfit and mistake every anomaly for another catastrophe.
Is it for you?
Best for
It is best for leveraged investors and lenders exposed to prices, rates, liquidity, or correlated defaults.
Not ideal for
It is not ideal as a precise short-term forecasting tool or as justification for assuming every extreme scenario will occur.
From the transcript
“we think that tiller risk is way more probable than everyone else does.”
“we don't get to live most of the time. We get to live all the time”
“there's a range of things. Can I be okay with a broadery of outcomes versus just betting on that one thing?”
From the episode
Sean Dobson on the US Real Estate Industry
Sean Dobson