The Critical Role of Data in Trading & Risk Management

Market data might not sound glamorous, but its impact is critical. When data fails, everything else follows, from pricing models to trading decisions and risk management.  

To take a closer look into this critical area, Numerix recently hosted a webinar titled “Make Data Sexy Again: Hard-Earned Insights from a Veteran Quant,” featuring Dr. Ola Hammarlid, a distinguished quant with decades of experience. This session offered a deep dive into the importance of data management for quants, traders and risk managers in financial institutions. Below are the key takeaways from this engaging session, capturing Dr. Hammarlid’s expertise and actionable insights.

Why Data Matters More Than Ever 

Dr. Hammarlid opened the webinar session with a playful analogy, "Data is like that ugly cousin quants need to invite to their projects but often overlook." 

While models tend to take center stage in quantitative finance, he emphasized that incorrect or insufficient data is often the silent culprit behind failed projects and miscalculated risks. He went on to highlight that robust data management is not only about maintaining accuracy, but also about mitigating the cascading impacts of errors throughout an organization’s systems.

The Scary Truth About Error Propagation 

In an institution’s complex technology ecosystem, data typically flows through multiple components—whether within the same system or across others—ultimately producing reports and outputs for traders, management, risk control, valuation, and external stakeholders. Errors introduced early in this process have the most significant impact, as they propagate through the ecosystem and become increasingly difficult and costly to detect and correct later. 

Typically, issues are only noticed downstream, requiring time-consuming debugging that involves working backwards through multiple, often poorly connected systems. But the later an error is identified, the greater the financial and operational impact will be. Since early errors amplify exponentially, addressing data quality and calculation integrity at the start of the workflow is critical to minimizing risk and maximizing efficiency. 

How Data Can Stall Vital Projects

During the presentation, Dr. Hammarlid painted a vivid scenario where insufficient or inconsistent data jeopardized an XVA pricing project. Calculating XVAs accurately requires comprehensive and consistent data across the entire portfolio, including market-implied volatilities, credit risk inputs like default probabilities, and other counterparty-specific data. Challenges arise from fragmented data sources—such as various trading systems with different models and conventions—and from the need to extract both raw and derived data. Because XVAs are portfolio-level metrics, the data inputs must be well-modeled, timely, and robust, making the process significantly more sensitive and complex than standalone pricing. Watch the webinar to hear Dr. Hammarlid speak firsthand on the nuances of data with respect to XVAs.

"Almost 70–80% of effort in financial projects is spent on data preparation, cleansing, and ensuring consistency. It's rarely about modeling alone," Dr. Hammarlid commented. Examples of common data challenges include:

• Missing market data, such as volatilities or default probabilities
• Mismatched data formats or data models across systems
• Data fragmentation or inconsistency from multiple trading platforms

These challenges underscore the critical importance of robust data management processes in ensuring accurate and effective XVA pricing.

The Role of Data Cleansing 

Clean data is essential for reliable risk management and valuation. Dr. Hammarlid advocated for an iterative approach to data cleansing, involving close collaboration across departments, including IT, traders, and risk professionals. 

He shared a few “golden nuggets” for data cleansing:

• Involve the Experts: IT teams manage processes, but traders, risk managers, and quants should oversee data accuracy.
• Continuous Improvement: Data quality is a moving target that demands ongoing iteration to meet evolving business and regulatory needs.
• Outcomes of Clean Data: Maintaining well-organized data leads to faster time to market, improved risk and operational control, and reduced resource strain. It also enables firms to capture more sophisticated business opportunities and stronger trading performance.

Proxies and Derived Data 

The use of proxy models and derived data, such as credit default swap approximations, is considered both a necessity and a risk in financial modeling. One key concern is the amplification of risks, as proxies tend to fail during unanticipated market shocks, leading to significant profit-and-loss (P&L) volatility. 

To mitigate such issues, it is crucial that proxy modeling decisions align as closely as possible to trading practices. Dr. Hammarlid emphasized that this alignment ensures consistency in assumptions between traders and risk management teams, promoting better overall decision-making and risk mitigation.

Complexity of Data/Model Relationships 

Data cleansing is both model- and usage-dependent, with the required level of accuracy varying based on the specific application. For instance, algorithmic trading demands high-frequency, precision-tuned data, while long-term investment decisions may tolerate broader approximations. 

In areas like curves and surfaces, smoothness is often prioritized, yet it frequently comes at the expense of robustness. Dr. Hammarlid explored the example of yield curves, explaining that using spline curves for bootstrapping may create smooth curves but can also create “tightrope behaviors” where underlying curve instruments (e.g. FRAs and swaps) overlap, introducing questionable Greeks and unnecessary trading risks. 

Striking the right balance is critical, especially when considering whether a curve is “good enough” or should be re-priced with a different model. Ultimately, the context determines which data characteristics are essential and which trade-offs are acceptable.

Data as a Strategic Asset 

While data management may not seem glamorous on the surface, Dr. Hammarlid’s talk underscored its critical impact on trading and risk management. By treating data as a dynamic and strategic asset, firms can achieve:

• Stronger Risk Controls: Data consistency enables more accurate risk and exposure calculations.
• Faster Time-to-Market: Clean and well-connected data reduces project delays and accelerates insights for decision making.
• Higher Profit Margins: Robust models powered by reliable data improve trading outcomes and P&L performance. 

Dr. Hammarlid described this as the true “sexy” part of data—the tangible benefits it brings, including better end of day (EOD) evaluation, correct hedging and higher profitability. 

Learn more: Did you miss the session? You can watch the on-demand webinar here.