Stress Testing Portfolios in the Age of AI

Executive Summary

Current market conditions are characterised by elevated concentration and heightened sensitivity to changes in discount rates. These features are most visible in parts of the equity market linked to structural growth themes, including AI, but they are not unique to them. Stress testing is a complement to standard risk metrics. Scenario analysis shows that portfolio resilience depends less on recent volatility and more on behaviour under combined macro, policy, and geopolitical shocks.

Why stress testing matters in the current environment

Traditional risk measures, such as volatility or drawdown, are backward-looking and implicitly assume stable correlations. Stress testing addresses this limitation by explicitly modelling how adverse but plausible shocks affect the key drivers of returns: expected cash flows, discount rates and risk premia.

The framework deliberately combines historical stress tests with forward-looking scenarios. This distinction is increasingly important in a market shaped by new sources of concentration, capital intensity, and policy uncertainty. Hypothetical stress testing complements history by defining shocks to cash flow growth, equity risk premia, real rates, inflation expectations, and term premia, and mapping them consistently to asset-class outcomes.

Stress scenarios and transmission channels

Among the forward-looking scenarios, one explicitly considers a correction linked to overinvestment and disappointed expectations in a high-growth segment of the equity market. While often associated with AI, the underlying mechanism is broader: downward revisions to expected cash flows combined with a rapid rise in equity risk premia. The resulting adjustment propagates across equity markets through wealth effects and tighter funding conditions.

More severe outcomes arise when growth disappointments interact with geopolitical or policy shocks. In a fragmented global economy, higher trade barriers and reduced cooperation lead to permanently higher discount rates and lower growth expectations. In this environment, globally integrated and capital-intensive business models are particularly exposed, and diversification across regions or asset classes provides limited protection. This scenario produces the largest aggregate drawdowns.

Across most adverse scenarios, fixed income offers only partial and highly conditional diversification. While policy easing mitigates losses in some cases, other regimes see equities, bonds, and real assets declining together.

Concentration and tail risk

A consistent finding across scenarios is that equities account for the majority of portfolio losses, reflecting low starting risk premia and high concentration. This is not a valuation argument alone, nor a short-term one. Stress testing highlights the interaction between concentration, discount rates, and macro shocks as the dominant source of tail risk.

Overlapping shocks, such as growth disappointment coinciding with geopolitical fragmentation or inflationary pressure, result in materially larger drawdowns than any single stress considered in isolation.

Historical perspective

Long historical simulations of diversified equity–bond portfolios show that deep drawdowns can persist over multi-year horizons, even when long-term returns are attractive. Periods of strong structural growth have not eliminated this pattern. Stress testing reframes past recoveries as conditional outcomes, dependent on policy responses, funding availability, and investor confidence.

Expected shortfall analysis reinforces this point by showing the non-linear nature of tail losses and their sensitivity to currency and policy assumptions. These effects become more pronounced in globally exposed portfolios.

Implications for portfolio construction

First, exposures linked to dominant growth themes should be assessed as clustered factors rather than independent positions. Correlations converge sharply in adverse regimes. Second, diversification across asset classes is regime-dependent and cannot be assumed to hold in all stress environments. Third, portfolio resilience should be evaluated against combined macro and policy shocks rather than standalone scenarios.

The objective is not to avoid structural growth exposure, but to size it consistently with drawdown tolerance and liquidity requirements under stress.

Role of stress testing in allocation decisions

Stress testing is not a forecasting tool. It is a governance framework that informs position sizing, concentration limits, and liquidity planning. In an environment of elevated concentration and policy uncertainty, it provides a disciplined way to distinguish between volatility that can be absorbed and drawdowns that would impair long-term capital compounding.

—

Federico Polese

This newsletter is the intellectual property of 20Quant Srl.

Get free Turbulence Index alerts: turbulence-index-alerts-index-alerts