Stress testing is a must-have for traders in Singapore to prepare for extreme market conditions. Traditional risk models like Value at Risk (VaR) often fail during crises because they rely on historical data and static assumptions. Stress testing solves this by simulating “what-if” scenarios – like market crashes or liquidity crises – to expose weaknesses and hidden risks.

Key takeaways:

• Why traditional models fail: Overreliance on historical data, incorrect assumptions, and poor response to shocks.
• How stress testing helps: It simulates extreme scenarios to test your portfolio’s limits.
• Methods to use: Scenario analysis, sensitivity analysis, and reverse stress testing.
• Tools to try: Python libraries, TradingView, or Singapore-based platforms like fintech.supremacy.sg.

Stress testing isn’t just about identifying risks – it’s about taking action before market shocks hit. By integrating it into your systematic trading process, you’ll gain better insights into your portfolio’s resilience and be better prepared for the unexpected.

sbb-itb-466c9b0

Common Problems in Risk-Adjusted Screening Models

Risk-adjusted models are built on assumptions – and those assumptions often fall apart just when they’re needed most. Many traders in Singapore rely on tools like VaR (Value at Risk) and historical backtesting to manage their portfolios. However, these models have three major weaknesses that become glaringly obvious during market shocks. These flaws highlight the importance of stress testing as a way to address these issues.

Overreliance on Historical Data

Historical simulation is limited to what has already happened. If a future crisis takes a different shape than past events, the model may fail to predict it. Most risk models rely on a 3- to 5-year historical window, but even a 5-year period can miss significant market cycles. Take the 2022 rate shock, for example. Between January and October 2022, “60/40” portfolios – designed to balance risk between equities and bonds – suffered because their models underestimated the impact of rapid rate hikes. These models, calibrated during a long period of low interest rates, failed to account for the sudden rise in duration risk, leading to simultaneous losses in both equities and supposedly “safe” Treasuries.

One of the biggest challenges is rising asset correlations during periods of extreme stress. When markets are under pressure, correlations often “go to 1”, meaning that diversification offers little to no protection as assets fall together. Adding to the problem, standard statistical models assume a normal distribution of returns, ignoring “fat tails” – the higher likelihood of extreme losses – which makes them blind to the true frequency of catastrophic events.

Incorrect Assumptions in Risk Projections

Beyond the limitations of historical data, many models suffer from static assumptions. Even small changes to inputs like growth rates, volatility, or correlations can drastically alter forecasts. The trouble is, pricing models like Black-Scholes are designed for stable market conditions. Under extreme stress, these models often deliver unreliable results because their foundational assumptions no longer apply.

“Under extreme stress, these models may produce unreliable outputs – precisely when accurate risk estimates matter most.”

• Ryan O’Connell, CFA, FRM

For instance, a 1-day 95% VaR of S$25,000 suggests there’s a 5% chance – roughly once a month – that losses will exceed that amount. However, VaR doesn’t explain how severe those losses could be. It only sets the threshold for the worst 5% of days, leaving traders blind to whether the breach will be minor or catastrophic.

Poor Response to Market Shocks

Traditional models often fail to keep up during sudden market disruptions. Price discontinuity is a key issue during shocks. Many models assume prices move incrementally, but in reality, prices can jump between levels without intermediate steps, causing Stop Loss orders to fail.

A stark example is the EUR/CHF crash on 15 January 2015. When the Swiss National Bank abruptly removed its currency floor, the pair plunged by roughly 3,500 pips in minutes. Traders with Stop Loss orders at 1.1950 saw executions near 0.9700 – an astonishing 2,200 pips away – because liquidity dried up.

During such shocks, liquidity vanishes as market makers pull their orders to avoid taking on risk. This lack of liquidity amplifies slippage, where actual losses far exceed planned ones. In some cases, realised losses can be three to five times higher than expected. Compounding the issue, many risk models rely on historical data that smooths out price movements, masking the true volatility and gaps that occur during real-time shocks. Stress testing, by revealing these gaps, becomes an essential tool to complement traditional risk models.

Stress Testing Methods for Trading Models

Choosing the right stress testing methods is crucial for uncovering weak spots and assessing how well trading models can withstand extreme market conditions. These approaches go beyond standard backtesting by simulating challenging scenarios and pinpointing vulnerabilities. Three key methods – scenario analysis, sensitivity analysis, and reverse stress testing – each provide distinct perspectives on potential risks.

Scenario Analysis

Scenario analysis examines how portfolios perform during specific historical, hypothetical, or stylised market events, rather than relying solely on statistical probabilities. For instance, during the Long-Term Capital Management (LTCM) crisis in August 1998, the fund’s unchecked exposures led to massive losses and emergency measures. This case highlights the importance of identifying key exposures before running scenarios.

Another example involves a 2026 simulation of a S$500 million defined-benefit pension fund with a 65/35 equity-bond allocation. When subjected to a 2008-style financial crisis, the fund faced a possible S$175 million loss (a 35% drawdown), surpassing its 25% loss threshold. To mitigate risk, the fund reduced its equity holdings to 50%, shifted S$75 million to Treasuries, and purchased S&P 500 put options for tail-risk protection. These cases stress the need to map critical exposures – like equity beta, duration, convexity, and currency risk – before conducting scenario tests.

Insights from scenario analysis often lay the groundwork for sensitivity analysis, which digs deeper into how individual variables influence model outcomes.

Sensitivity Analysis

Sensitivity analysis focuses on how changes in a single input affect the model’s output, keeping all other variables constant. This method helps identify which assumptions have the most significant impact on performance. For example, a 100 basis point rise in Treasury yields can cut the net present value of growth stocks by 15–20%. Similarly, the same rate increase could add millions to a company’s interest expenses, revealing the sensitivity of financial outcomes to interest rate movements.

In recent years, many mid-cap firms have experienced a rise in their weighted average cost of capital, jumping from an average of 7.0% to about 9.5% between 2025 and 2026. Tools like Excel‘s Scenario Manager and Data Tables can visually demonstrate how changes in the top 5 to 10 key variables affect outcomes. Running these tests helps pinpoint the most critical drivers of your model.

Reverse Stress Testing

Reverse stress testing flips the usual question around. Instead of asking, “What happens if this event occurs?” it starts with a failure point and works backward to uncover the sequence of events that would lead to that failure.

This method identifies tipping points, such as a sharp revenue drop, a spike in costs, or rising interest rates that could breach debt covenants or wipe out equity. The idea is to determine how manageable factors could combine into a catastrophic outcome. As the Financial Services Authority explains:

“We are asking firms to identify those events (or confluences of events) that would lead to their business model failing after currently available management actions had been taken.”

• Financial Services Authority

Reverse stress testing also explores how reputational damage (a “name crisis”) might trigger a liquidity run. By linking these risks, firms can define robust risk limits, set adequate capital buffers, and ensure sufficient liquidity reserves. When combined with scenario and sensitivity analyses, reverse stress testing provides a well-rounded strategy for strengthening trading models.

Benefits of Stress Testing for Systematic Trading

Stronger Model Performance

Stress testing plays a key role in strengthening trading models by uncovering risks that standard metrics might miss. For instance, while metrics like Value at Risk (VaR) might indicate a 5% chance of losing more than S$50 million, they often fail to show how much worse losses could get during a crisis. Stress testing addresses this limitation by simulating extreme conditions and pushing variables beyond their usual ranges. This process reveals hidden correlations and concentrated risks that only become apparent in turbulent markets. By shedding light on these vulnerabilities, stress testing helps refine models, boosting their reliability and effectiveness.

Greater Trader Confidence

Well-executed stress tests not only identify weaknesses but also serve as a strategic tool for traders. As POEMS highlights:

“The effectiveness of a stress test lies not only in its ability to uncover weaknesses but also in its role as a strategic compass.”

By running these tests proactively, traders can make informed adjustments to their portfolios and put protective measures in place before market shocks hit. This preparation builds confidence among traders and stakeholders, knowing that portfolios are better equipped to handle severe downturns.

Better Response to Changing Markets

Markets are constantly evolving, and stress testing ensures trading models stay relevant and responsive. With 75% of financial institutions now incorporating Artificial Intelligence into their modelling frameworks, traders can simulate a wide range of adverse scenarios and detect patterns that might escape manual analysis. Modern systems leverage real-time data to integrate stress models, allowing for rapid re-testing whenever market conditions shift. This continuous testing approach ensures that models remain effective and adaptable, even during volatile periods. By providing deeper risk insights, stress testing helps traders move beyond simply identifying vulnerabilities to taking meaningful, proactive actions.

How to Add Stress Testing to Your Trading Process

Steps to Perform Stress Testing

Incorporating stress testing into your trading process can help uncover hidden risks and strengthen your models. Start by performing a standard backtest on historical data to establish baseline metrics like the Sharpe ratio and maximum drawdown. These benchmarks provide a clear picture of your strategy’s “normal” performance.

Next, dive into micro-level stress tests by altering individual variables. For instance, you could double historical volatility, increase transaction costs, or tweak key parameters to see how your model handles slight disruptions.

After that, move to macro-level scenario analysis. Test your strategy against past crises like the 2008 Global Financial Crisis or the 2020 COVID-19 crash. You can also simulate hypothetical events, such as a 35% flash crash. For Singapore-based portfolios, calculate the beta of your holdings – stocks, REITs, and Business Trusts – against the Straits Times Index (STI). This will give you a sense of how your portfolio might behave during a local market downturn.

Monte Carlo simulations are another essential step. By randomising trade sequences thousands of times, you can uncover potential drawdowns that a single historical path might miss. For reliable estimates of extreme outcomes (e.g., 99th percentile), aim for 50,000 to 200,000 simulated paths. Finally, reverse stress testing can help pinpoint vulnerabilities. Start with a failure threshold, like a revenue drop leading to insolvency, and work backward to determine how likely that scenario might be.

Tools for Stress Testing

Traders in Singapore have access to various tools for stress testing. For TradingView users, BacktestBase offers over 1,000 randomised scenarios and provides a “Robustness Grade” ranging from A+ to F. If you prefer working with local data, fintech.supremacy.sg is a Singapore-hosted tool that runs 1,000 Monte Carlo simulations directly in your browser.

For those who need more control, Python libraries like Pandas, NumPy, and Backtrader allow for customisation, including trade-order randomisation and bootstrap resampling. Even Excel can handle basic Monte Carlo analysis using the RAND() function, though it may struggle with large datasets. On the professional side, platforms like StrategyQuant offer Monte Carlo testing with features such as parameter adjustments and slippage variance. They even provide a 14-day trial licence for new users.

Best Practices for Stress Testing

Once you’ve chosen your tools, follow some key practices to make your stress testing more effective. Create scenarios that go beyond historical data. Since financial returns often have “fat tails”, use Student-t distributions or empirical bootstraps instead of assuming normal distributions to better capture extreme outcomes.

Account for market frictions by simulating shocks to liquidity and transaction costs. For example, test your strategy with a 1% commission and 2% slippage to reflect real-world execution challenges.

A real-life example from the Global Financial Crisis highlights the importance of robust stress testing. Singapore investor Lee Chin Wai conducted a Value-at-Risk (VaR) stress test at a 99.5% confidence level, predicting a 24% decline. However, his portfolio actually dropped by 51%, more than double the estimate. This was due to newer asset classes like REITs and Business Trusts lacking historical bear market data, as well as correlations with the STI spiking to nearly 1.0 during the crisis.

Keep your stress scenarios up to date. Revisit them quarterly or after major market events like interest rate hikes or geopolitical disruptions. To detect overfitting, adjust parameters by ±5% and randomly drop 5–20% of trades in your simulations. This helps account for execution errors and ensures your models remain robust.

For more insights on integrating stress testing into your trading process, check out the educational resources offered by Collin Seow Trading Academy.

Conclusion

Stress testing plays a crucial role in uncovering hidden weaknesses, identifying correlation spikes, and addressing concentrated risks. It strengthens the reliability of risk-adjusted models by focusing on areas that standard metrics like Value at Risk (VaR) often overlook during stable periods. As Ryan O’Connell, CFA, FRM, aptly points out, stress testing tackles the critical question: What happens when markets break? – a question that standard models fail to answer.

History has repeatedly shown the importance of stress testing. For instance, during past financial crises, VaR models provided a false sense of security, while well-constructed stress tests – simulating scenarios like sovereign defaults or liquidity freezes – could have exposed fatal risks. A practical example of this was the March 2026 pension fund case, where a single stress test modelled after the 2008 financial crisis led to actionable measures, including reducing equity exposure, purchasing put options, and establishing liquidity reserves.

The key takeaway is to translate stress test results into meaningful actions. Whether it involves adjusting position sizes, hedging against extreme risks, or revisiting assumptions on a regular basis, stress testing should be a cornerstone of strategic decision-making rather than a mere compliance exercise. For traders in Singapore, this means tailoring scenarios to local market conditions, such as STI index correlations and realistic transaction costs, to ensure practical and relevant outcomes.

If you’re ready to enhance the resilience of your trading strategies, consider integrating systematic stress testing into your process. Educational resources like those offered by Collin Seow Trading Academy can help you build models that stand strong when it matters most – during the volatile market conditions that truly challenge your strategy.

FAQs

How is stress testing different from VaR?

Stress testing and Value at Risk (VaR) serve different purposes in risk management. VaR calculates potential losses within a set confidence level, relying on historical data to predict outcomes. On the other hand, stress testing examines how a portfolio would perform under extreme, yet plausible, scenarios that go beyond normal statistical assumptions.

While VaR provides a snapshot of risk under typical market conditions, stress testing highlights vulnerabilities that might not appear in VaR calculations – especially during financial crises. This makes stress testing an essential tool for identifying risks that could otherwise go unnoticed, complementing VaR in a well-rounded risk management strategy.

Which stress test method should I start with?

When working on your financial model, start by conducting a sanity check or scenario analysis. This means testing the formulas and logic with basic calculations under various conditions to ensure the results make sense. For managing risks, scenario analysis is especially helpful. It allows you to assess how the model performs under extreme yet plausible situations, giving you a clearer picture of potential vulnerabilities. These simple methods are a good starting point before moving on to more advanced approaches like extreme scenario simulations or Conditional Value at Risk (CVaR) analysis for deeper stress testing.

How often should I update my stress scenarios?

You should review and update your stress scenarios on a regular basis – ideally at least once a year. It’s also important to revisit them whenever there are major shifts in market conditions or changes to your financial assumptions. This helps keep your models accurate and ensures they remain effective for managing risks.