Trading and risk management during system stress events

How stress regimes change market behaviour

Stress regimes are not merely louder versions of normal markets. They are structurally different.

Distributions expand and become uneven. Correlations that normally offer diversification weaken as weather, outages, and flows synchronize. Liquidity diminishes unevenly, often vanishing first during the hours and at the locations where it is most crucial.

This is why relying on historical averages can be risky during stressful periods. Models based on long-term behaviour assume a stability that may no longer be present. What truly matters is not the average, but the shape of the distribution's tails.

A useful mental shift for traders is this: during periods of stress, risk is no longer smooth. It arrives in jumps.

Position sizing when tail risk dominates

Sizing mistakes often cause desks to struggle during stressful times, highlighting how important it is to get the measurements just right.

When tail risk takes centre stage, exposure tends to act in a more amplified way. Losses can grow more quickly than gains as situations worsen, especially during busy hours and close to deadlines. Because of this, sticking to usual position sizes might not be suitable, even if you're quite confident about your position.

Rather than thinking in terms of how right a view is, traders need to think in terms of how wrong it could be.

In practice, stress-aware sizing usually reflects a combination of:

• Scaling exposure down as uncertainty widens

• Lower positions during periods of liquidity risk

• Avoiding concentration in short-dated peak products.

Smaller positions in stressed markets do not represent defensive trading. Instead, they recognise that execution risk and asymmetric downside, rather than directional errors, are often the main constraints.

Liquidity collapse and execution challenges

Liquidity is usually the first thing traders underestimate during periods of system stress.

Bid/ask spreads suddenly widen, making displayed depth less reliable, and prices jump through levels that seemed tradeable just minutes before. As delivery approaches, these issues worsen: obligations to balance take precedence over optionality, and timing risk rises significantly.

This is often where losses happen - not due to incorrect market predictions, but because positions needed adjustment under pressure.

Instead of treating execution as an afterthought, stress-aware desks explicitly plan for the following:

• Situations where liquidity vanishes without warning

• The cost of being forced to trade near delivery

• The possibility that “working an order” is no longer viable.

Balancing markets often reveal these fractures initially, making imbalance behaviour a crucial real-time stress indicator.

Scenario-based stress testing

Stress regimes cannot be managed using point forecasts.

Scenario thinking changes the question from “what do we expect?” to “what happens if this goes wrong?” The main value is not in predicting the scenario, but in understanding how the portfolio behaves if it occurs.

Effective scenarios show how stress typically develops; through combinations rather than isolated shocks. Weather-related stress aligns with outages. Dependency on imports becomes problematic, especially when neighbouring systems are also under pressure.

Instead of keeping a lengthy list of abstract scenarios, many teams concentrate on a few key ones:

• Severe weather driving demand and renewable stress

• Clusters of unplanned outages reduce operational flexibility

• Sudden interconnector loss during peak periods.

The crucial step is converting these scenarios into action: setting exposure caps, escalation thresholds, and predefined responses. If a scenario fails to alter behavior, it is not considered a risk management tool.

Post-event learning and governance

The true impact of a stress event becomes evident only after it has concluded.

Without a structured review process, desks often recall outcomes but overlook their causes. Effective post-event learning identifies whether issues arose from forecast uncertainty, outage shocks, flow constraints, or execution errors - and it adjusts systems based on those insights.

This is where governance becomes crucial. Informal lessons tend to fade rapidly, especially when team members change. Dashboards, limits, and triggers need to be updated intentionally, rather than relying on them to "evolve" on their own.

Well-run desks typically institutionalise learning through:

• Ensuring that we clearly recognise who made decisions and which drivers are involved helps us keep things transparent and trustworthy

• Updating stress indicators and scenario assumptions

• Review and refine the limits and escalation rules where they didn't meet expectations

• Embedding lessons into standard trading processes.

This closes the loop back to the reliability and system stress frameworks established earlier in the series (internal link opportunity).

Conclusion

System stress events are unavoidable in today's power markets. The key difference between desks lies in their level of preparedness, not in their exposure to stress.

Effective risk management in stressful situations isn't about eliminating volatility. Instead, it involves detecting regime shifts early, adjusting exposure to manage asymmetric risks, acting decisively when liquidity is limited, and systematically learning from extreme events.

In markets where a few hours can determine an entire year, maintaining discipline is essential. It provides the advantage.