How do you validate your climate risk scores/assessments?

Helios’ climate risk scores are validated through trusted data sources, rigorous back-testing, real-world event alignment, expert oversight, and continuous feedback, making them one of the most reliable tools for predicting agricultural risk and its impact on markets.

Helios’ climate risk scores start with climate data from the most trusted global sources—ECMWF, NOAA, NASA, DWD, JMA, BOM, and others—combined with maritime, snow, and mesonet datasets. Before analysis, all data is cleaned, normalized, and adjusted for factors like reporting time zones and station reliability. This ensures our starting point is as accurate as possible.

Our anomaly detection and clustering models—including WA%R (Weighted Agricultural Percent of Risk) and seasonal risk ratings—are validated against decades of historical weather and production outcomes. We test model outputs against:

-Historical climate patterns to ensure signal alignment with known events.

-Real-world agricultural impacts, such as yield drops, price spikes, or supply disruptions that occurred after weather anomalies.

This back-testing process confirms that our risk scores are predictive, not just descriptive.

Helios actively works with market experts, climate scientists, agronomists, and industry advisors to validate our assessments. These experts—ranging from former global procurement leaders at Coca-Cola to sustainability heads at Starbucks and ALDI—review risk outputs and flag discrepancies. This human-in-the-loop process ensures our models aren’t just statistically sound, but market-relevant.

We compare climate risk scores against commodity price movements, trade flows, and supply disruptions. For example, if WA%R spikes for cocoa-producing regions, we assess whether futures and cash prices react in the predicted window. This alignment between forecasted risk and actual market behavior reinforces score accuracy.

Client interactions—especially via Horizon, our AI analyst—play a critical role. When users thumb up/down a climate risk assessment, our team reviews those cases to understand the reasoning and improve the algorithms. This feedback loop ensures our scoring system evolves with market dynamics and user needs.