The Fallacy of Numbers AI and Unstructured Data: Extracting Signals for Financial Investors

The fallacy of numbers

Traditional financial analysis based on key performance indicators (KPIs) such as EBIDTA, ROA, CFROA, EBIT, ROIC among others can provide a tentative snapshot of how markets and stocks are likely to behave within any given context or timeframe, but they are providing little guidance in a world shaped by concurrent vulnerabilities, which are more and more event-driven.

Whereas most of the analysts are basing their evaluations on a panoply of “official” numbers published in corporate balance sheets, corporate statements and SEC filings, we tend to move beyond what it is already in the known, to search for the “un-knowns un-knowns” that are emerging as the key-drivers of financial performance.

When news about M&A events, spin-offs, deleveraging, change of ownership, or management are published in the press or covered by Bloomberg Television or CBNC, it is likely that the expectations about the future value of any given equity stock or bond are already factored into their price. And thus, it is already too late for investors to adapt their investment strategies or to adjust their portfolio allocation.

While the general sentiment is that financial markets are frequently hit, more than ever, by “sudden surprises” or unanticipated shocks, leaving investors in the dark with respect to the evolution of financial trends, we aim at mapping and analyzing the vast amount of unstructured data that are unlikely to be captured by “numbers” and they will be likely factored into the analysis, when it is already too late for investors to adjust their risk-return ratio and portfolio allocation. More and more, investors are feeling increasingly uncomfortable due to the ever-growing wave of “surprises” that are likely to catch them off guard.

The “fallacy” of numbers is the critical weakness affecting financial investors and it cannot be tamed simply by carrying out research and analysis by adding more resources in a linear way. While investors are increasingly concerned about their “real” capability to extract signals from an increasingly unpredictable financial environment, we see an opportunity in the mapping and analysis of a growing body of unstructured data that contribute to the formation of market prices, while exerting an in-depth, irreversible influence on the evolution of equity and bond markets and the sustainability of sovereign debt.

Monitoring “official” numbers and market prices is not enough to anticipate market rotations or “hidden” surprises, leading to price corrections across the board. Whereas looking at “numbers” is not sufficient to make order out of chaos, it is helpful to look at the growing body of unstructured data to detect patterns that numbers are likely to capture at a much later stage. Where “others” see surprises, we see “patterns” that tend to repeat their evolution over time. And we extract “signals” from these emerging patterns to provide guidance in terms of how “numbers” will be affected, while setting in motions a sequence of events that takes the shape of M&A activity, consolidation, restructuring, divestitures and change of ownership and management.

To achieve a comprehensive understanding of how unstructured data impact on financial performance we focus our computational resources in modeling how different triggers of change can alter the dynamics of financial markets across a wide spectrum of asset classes, and we aim at identifying the magnitude and scope of different types of undetected non- financial risks that are likely to impact on market prices. While publicly available financial data represent a starting point to develop a comprehensive overview of market dynamics, we focus our research work on how to tame the growing body of seemingly unrelated sources of unstructured data that can trigger irreversible patterns, capable to inject more uncertainty and volatility into an already fragile world economy.

Unstructured data represent an elusive target in terms of analysis: they are spread across an ever-growing variety of sources and data bases and they introduce new challenges in terms of data quality, homogeneity and integration. But, all in all they are increasingly becoming the key source of undetected disruptions that are laying underneath the current trends and they are likely to be captured by numbers when it is already too late for investors to adapt to the new reality. 

This is the reason why it is increasingly critical to incorporate into financial analysis the research for patterns emerging across different sources of unstructured data, in the form of different classes of information and data points across multiple dimensions of the world economy that are capable to trigger changes in the market dynamics and the formation of market prices.

Thus, our research work is focused on the challenge of mapping and monitoring the way unstructured data impact on key financial performance indicators, by creating tools, metrics and knowledge repositories that are designed to “track-down” how seemingly unrelated data points are building up pressure on the current market dynamics capable to produce an enduring and irreversible change in the overall value of multiple asset classes.

Identifying patterns across unstructured data is a challenging endeavor per se, given the magnitude and the scope of the research endeavor, but it is of fundamental importance because patterns tend to repeat themselves: detecting seemingly unrelated triggers ahead of time allows for creating the buffer of time needed to adjust to a sudden change in market dynamics, when financial analysis fails to provide guidance.

Unstructured data is elusive to traditional financial analysis due to a number of fundamental flaws: first and foremost, unstructured data are spread across a growing variety of sources, that lack consistency, homogeneity and represent a challenge in terms of “data integrity”. The growing body of “false positives” and “fake news” is not just a challenge affecting main stream media, but it represents a long-term struggle that is here to stay and it will have an enduring impact on investors’ capability to detect the market dynamics. Secondly, unstructured data are generated by a growing number of sources, that require an ongoing, methodical effort to keep track of the unanticipated triggers that can set in motion a sequence of adjustments in the pricing of stocks and bonds.

Thirdly, mapping unstructured data requires an industrious effort at building vast repositories of knowledge which cover long time frames, to adjust for the inherent lack of data homogeneity that represents a challenge per se in terms of analytical efficacy.

Tracking-down multiple data sources that come in different formats and disparate quality requires a robust, resilient data-collection architecture, designed to maximize the capability to dissect vast amounts of text, audio, video, analog data, images, files, web pages, and social media content to detect patterns that can trigger a predefined sequence of events in terms of specific market dynamics. The exponential growth of unstructured data affecting financial markets represent a fundamental challenge that requires an unprecedented overhaul of how research is carried out in terms of scope (what to search and what to look for), tools (how to carry out research into a non-linear world, by venturing out into semantic search, leveraging the advancements in AI), models (where to look for interdependencies that can be captured by developing cognitive maps of how the economy works) and metrics (which are the KPIs that can be extracted from a vast body of seemingly unrelated data that can trigger a specific pattern of change, influencing market dynamics).

Data scientists are of the opinion that, on average, 90% of big data is unstructured data: analysts and investors may opt for not being bothered by the impact of this growing wave of seemingly unrelated and disconnected data sets, but in reality, the effect on the formation of financial expectations and market prices is already generating shocks and “surprises” and this trend will not change anytime, soon.

Unstructured data are increasingly becoming a key source of financial guidance providing investors with intelligence insights on how undetected sources of risks due to seemingly unrelated data sets can impact the evolution of market dynamics and financial resilience.

In order to tap the insights extracted from unstructured data and to map out and identify the emerging sources of vulnerability, there are just a few fundamental questions that need to be addressed:

• what are we missing when moving beyond the analysis of pure financial data? 
• which are the key sources that need to be monitored?
• why is traditional competitive analysis inadequate to capture the dynamics of change and how new analytical frameworks and tools can be leveraged to extract signals from noise? 
• how to transform unstructured data into financial data, capable to trigger adjustments in asset allocation and portfolio management?
• how leading investors can build a core capability to detect or anticipate these unforeseen trends? 
• how to identify repetitive patterns while mapping the sources, the competitive drivers, the key-actors that play a role in challenging the balance of power within and across any given asset class? 
• how can you detect the key triggers that will set in motion an irreversible transition affecting the price of any target investment or equity stock, threatening its financial performance and/or curtailing its financial resilience?

Kaufmann & Partners is leveraging a proprietary, state-of-the-art intelligence analytics platform, and it is scaling it up with massive computational power to provide more impactful competitive analysis and financial decision making.

Whereas most the analysts are working on a limited set of information, the goal of Kaufmann & Partners is to map the vast majority of unstructured data that are not captured by financial analysis and to provide a superior Intelligence.

The purpose of our analysis is to focus on the key areas, that are likely to shape the dynamics of the evaluated companies, by monitoring the critical KPIs (key performance indicators) and the triggers of change.

About Kaufmann & Partners

Kaufmann & Partners is an investor and corporate advisory headquartered in London and Madrid that provides strategic advice and investment insights with our team of Senior Partners, on international political, economic, technology, sustainability and security matters.

Kaufmann & Partners advises leading investment firms, world-class family offices and innovative corporations through a diverse range of industries aiming at enabling them to navigate the unprecedented uncertainties stemming from today’s unstable political, business, technological and regulatory environment.

Selectively, Kaufmann & Partners provides funding and its support to highly disruptive start-up companies in the Artificial Intelligence and FIN-TECH market space.

Our Top Management Team and Advisors have built a world-class capability to address the low probability events shaping the world economy, that are the source of an unprecedented level of volatility across markets.

The overarching goal of Kaufmann & Partners is to identify and measure non-financial risks as a major source of technology disruptions, unexpected fragilities, sustainability and geopolitical challenges by translating them into consistent and robust financial metrics and indicators.

By leveraging big data analytics and semantic search, Kaufmann & Partners has built a state of the art capability to track down the key drivers of industry evolution affecting how investors look at different asset classes.