The commodities market is ripe with data, from the traditional supply and demand metrics reported by government agencies on a regular (though delayed) basis, to real-time data, such as port inventory that can be computed using satellite imagery. Finding ways to use this largely non-price (or non-financial) data in rules-based, investable indices can present attractive opportunities.
The scale and scope of non-financial data available in commodities markets is unique. Yet processing and modeling that data to produce investable signals with attractive risk/return profiles on a large scale has proven difficult. However, a new generation of commodities indices utilizing supervised machine learning to compute large non-linear datasets to generate investable signals may be on the horizon.
This approach could also offer a solution to incorporating ESG data into commodities indices. For example, S&P Global Platts, a sister company to S&P Dow Jones Indices, announced its intention to develop AI-driven physical carbon credit price assessment indices in February 2021. The new price assessments will leverage environmental AI expertise provided by Viridios Capital, which has been trained on over 20,000 data points, representing transactions from across the range of carbon projects around the world.
