Why Satellite Data Could Change Economic Monitoring
Iron ore is essential in the global economy, primarily used to produce steel. Typically, iron ore demand is tracked with reports that often arrive too late to show current market trends. This proof of concept shows that by using satellites to monitor steel plants, it is possible to get timely and actionable insights into iron ore demand and its impact on prices.
Hypotheses Behind the Approach:
- Steel Plants as Major Demand Drivers: Large steel plants that produce steel using iron ore (rather than recycled scrap) are the primary source of iron ore demand, directly impacting its market price. Monitoring these facilities can, therefore, reveal critical demand patterns.
- Stable Iron Ore Supply: Since iron ore is abundant, its supply is relatively stable. Therefore, price changes are likely to be driven by shifts in demand, which often reflect the production levels of the steel industry.
- Satellite Data as a Reliable Indicator: By continuously observing thermal emissions from steel plants, satellite data can capture production activity shifts in near real-time. These shifts in production may provide early indicators of changes in iron ore demand and its impact on price trends.
Main Findings
Rare but Meaningful Signals: The model detected just four major trade signals over three years, highlighting the slow but steady production cycle of the steel industry. This low frequency shows that significant shifts in iron ore demand don’t happen often but are highly relevant.
Real-World Accuracy: A notable sell signal in mid-2021 matched a major drop in iron ore prices, reflecting an actual slowdown in steel production. This shows the model’s ability to pick up on important economic changes.
Better Returns: When compared to the standard buy-and-hold approach of the S&P 500, the strategy based on iron ore demand showed higher cumulative returns, suggesting that satellite data can reveal important patterns in commodity demand.
Limitations and Further Research
While these results are promising, it is important to consider potential limitations. The observed returns may be partly due to an extraordinary price increase in iron ore during the study period, which could have amplified the model’s performance. Additionally, this analysis relies on backtesting, meaning it examines historical data without assessing risks in a real-world application. To fully validate this approach, it would be necessary to conduct further studies over extended timeframes and varying market conditions.
Conclusion: Satellite Data as a Tool for Economic and Market Insights
This study demonstrates the potential of using satellite data to monitor global steel plant activity and anticipate changes in iron ore prices. By linking real-world production activities to market prices, LaGrand has shown that neural networks can learn price formation mechanisms along the supply chain. This approach provides a faster, more objective way to analyze economic trends compared to traditional reports and opens up new possibilities for tracking the market dynamics of essential commodities.