Interviewees

Sercan Yıldız — Director of Analytics Research, Axioma by SimCorp

Melissa Brown — Head of Investment Decision Research, Axioma by SimCorp

How should systematic investors think about designing risk models that are robust across regimes, not just optimised for the past?

Melissa Brown:

First and foremost, a risk model should be used to ensure that portfolio bets are commensurate with the risk they entail. Sometimes even small tweaks can bring risk back in line with return expectations. Risk should also be re-evaluated on a regular basis, as underlying volatilities and correlations can change even in the absence of large market moves.

Even when a primary model is fundamentally oriented, using an additional model structure — such as a statistical model — can help reveal hidden risks that may otherwise be missed. Forecasting volatility over different horizons can also provide insight into how risk may evolve. For example, a short-horizon risk forecast is often a leading indicator relative to longer-horizon estimates and can help managers deploy their turnover budget more effectively.

While risk models are effective at identifying today’s risks, stress testing helps investors understand what may happen tomorrow. There are many forms of stress testing, from replaying historical scenarios to applying relationships between variables observed in other periods — sometimes referred to as transitive stress tests. These approaches can help assess the impact of market moves that may not be directly represented in the investment process, such as changes in short-term interest rates affecting an equity portfolio.

Although these considerations are especially relevant for systematic portfolios, they apply broadly to any portfolio where the manager is actively balancing risk and return.

When traditional factors stop behaving as expected, what practical indicators help distinguish a temporary dislocation from a genuine regime shift?

Melissa Brown:

Fundamental equity risk models are typically constructed using factors — such as industries, countries, currencies, and investment styles — that have demonstrated an ability to predict volatility over long periods of time and across different market environments. In theory, these models should work across regimes.

However, there are periods — often sudden and without warning — when the relationship between risk exposures and subsequent returns breaks down. It’s also important to distinguish between different types of factors. Some factors, such as Value or Momentum, are associated with long-term risk premia, while others, such as Leverage, represent sources of risk without an expectation of long-term excess return.

What investors often describe as “factor breakdowns” typically refer to compensated risk factors moving against their long-term expectations. For example, while smaller stocks are often expected to outperform larger stocks over the long term, the period from 2023 to 2025 saw the opposite, with large-cap stocks significantly outperforming smaller names.

This does not necessarily imply that the risk model itself has failed, as risk models are designed to treat risk symmetrically. That said, we have observed — anecdotally rather than with statistical precision — that when style factors such as Momentum reverse sharply and generate outsized returns in the opposite direction of expectations, it can signal shifting underlying market dynamics and potentially a change in market direction.

Because risk models are designed to deliver accurate risk estimates over long horizons, these episodes are unlikely to prompt immediate changes to model construction. From a portfolio risk perspective, however, combining a statistical model with a fundamental one may help surface risks “bubbling under the surface” that warrant further investigation and may indicate an emerging regime shift.

As investors diversify across asset classes while maintaining capital efficiency, where do you see the biggest trade-offs between risk control, return objectives, and implementation complexity?

Sercan Yıldız:

The most significant trade-offs arise when balancing efficiency and flexibility within a portfolio construction architecture that allows investors to express macroeconomic or alpha views effectively.

Investors are increasingly moving beyond traditional strategic asset allocation approaches, which can constrain both returns and diversification, toward true multi-asset frameworks that integrate risk, return, and liquidity considerations across public and private markets. The growing adoption of the Total Portfolio Approach among asset owners is one example of this shift.

Implementing such frameworks typically requires a more unified data infrastructure that consolidates positions, exposures, and cash flows into a single analytical view. While this adds complexity, it enables consistent risk and performance attribution, scenario analysis, and liquidity forecasting across the entire portfolio.

Robust portfolio construction depends on this level of integration. Optimisation frameworks can only generate coherent cross-asset allocations when they are fed harmonised data, aligned factor models, and synchronised pricing and rebalancing frequencies. Without integrated analytics, portfolio construction remains fragmented, and critical interactions — such as factor correlations, liquidity shocks, and multi-asset drawdowns — are difficult to incorporate into the decision-making process.

How is multi-asset portfolio construction evolving in response to higher macro uncertainty?

Sercan Yıldız:

Rising macro uncertainty is accelerating the shift toward regime-aware, factor-driven portfolio design. Unified cross-asset frameworks that capture exposures to equities, interest rates, credit, inflation, currencies, and commodities allow investors to anchor dynamic allocation and risk budgeting decisions in macro sensitivities.

Portfolio managers are increasingly relying on stress testing and scenario analysis to understand how portfolios may behave under different macroeconomic outcomes and to prepare for potential shocks. Systematic frameworks that integrate macro-linked factor models with stress testing can provide a stable foundation for multi-asset portfolio construction, improving portfolio robustness when traditional asset-class relationships break down.

If you had to give one practical piece of advice to asset owners reassessing their risk frameworks today, what should they prioritise over the next 6–12 months?

Sercan Yıldız:

Asset owners should prioritise moving away from asset-class silos toward a unified, factor-based view of risk and performance. While this transformation takes time, beginning the process now is critical.

In the near term, I would recommend integrating stress testing more deeply into investment decision-making — beyond regulatory or compliance requirements — and adopting multiple risk forecasts built using different methodologies and spanning short-, medium-, and long-term horizons. Short-horizon models are better suited to capturing fast-moving dynamics and event-driven volatility, while longer-horizon models reflect structural trends and strategic risks. Combining these perspectives provides a more comprehensive foundation for allocation and rebalancing decisions.

Melissa Brown:

Effective risk management should not only help investors understand the risks embedded in their portfolios, but also serve as a catalyst for action. Risk management should not be treated as a rubber-stamping exercise, but as an integral component of managing efficient and effective portfolios. Taking risk without a clear expectation of return simply consumes turnover and trading costs.

One useful way to think about this is through a simple four-step framework: Ask, Adapt, Act, Ask Again. Ask what could derail the portfolio. Adapt views on alpha and risk to reflect uncertainty, particularly known unknowns. Act by trading to either mitigate risks or take advantage of opportunities. Then Ask again: does the portfolio still reflect current views, or have market conditions changed enough to warrant another reassessment?