Alternative data can provide an opportunity to improve an investment's performance. Thematic alternative data represents company or security 'intangibles': a theme for a company may be related to a product, like smartphones for Apple, or online retail for Amazon, or they can be related to a strategy like 'metaverse' for Meta and Microsoft. Turning thematic alternative data into a strategy or an investment product that can be tested for performance presents many challenges. For example, most of today's thematic ETFs are based on discretionary processes, instead of data or algorithms, and are challenged to provide convincing answers to why a company is included, how is it weighted, why it is selected over another participating in the theme. Last week, Noonum finished our first technical paper describing a backtest that demonstrates a successful approach to using alternative thematic data for creating thematic products.
To test our hypothesis, we did not create a hypothetical portfolio that typically would rely on several additional factors for universe selection and security weighting. Our goal was to achieve a result that was more general than using thematic data within a broader set of factors; we wanted to find a result that could speak directly to the valube of thematic data. To test whether our thematic data could successful align an investment toward a thematic strategy, we defined a broad strategy around energy and power themes, selected a broad market ETF matching the S&P 500, and identified a commodity index that represents the global price of oil. If our thematic data captures material relationships to the strategy (energy and power), we expect to see superior correlation to the oil index with our tilting than the original market index and other approaches to tilting. We used this approach because it goes beyond the processes to create a product that may perform well over some specific period of time, this approach addresses the thematic investing strategy: with a thematic strategy in hand, thematic alternative data is used to systematically align an existing set of securities, e.g. within a portfolio or direct index product representing a broad market ETF, to the strategy.
Using the defined investment strategy, over a multi-year period, with monthly rebalancing, we tilted the market ETF using Noonum's thematic data. We compared our approach to the prevailing keyword statistical techniques to show that Noonum's thematic data consistently tilted the market ETF more successfully toward the strategy. Additionally, we found that the prevailing technology of text mining is unable to achieve a similar correlation - it appears to introduce a high degree of randomness, implying that co-mention or co-occurrence based keyword approaches for thematic strategies are not capturing meaningful material relationships. We are currently in the process of sharing the report with key customers and prospects, and in the coming weeks will be offering the report to others to help them understand the value of alternative thematic data.