Knowledge Graphs Will Disrupt Finance, Redeeming Quantitative Trading & Quant-Based Hedge Funds

In recent years, quantitative trading and quant-based hedge funds have largely failed to live up to their promise of identifying true alpha and exploiting market inefficiencies, with only few, headline-grabbing exceptions. However, the adoption of knowledge graphs may change this gap in performance. A knowledge graph is a structure for representing knowledge in a way that mimics leading theories of human cognition, in a computationally tenable way. These structures can provide a new way of thinking about data and modeling that can lead to the identification and exploitation of market inefficiencies.

In the early 2010s, many of my friends and colleagues went into finance as quantitative analysts, lured by the promise of being the next Jim Simmons and remaking Renaissance Technologies. However, they were all ultimately disappointed. As the years went on, conference after conference showed waning interest in quantitative funds, and General Partners (GPs) seemed to run away from heavy quant strategies altogether. Small funds would show promise, build a track record, and make their thesis robust, but would ultimately plateau in performance around the $75mm AUM mark.

The failure of quantitative firms can be attributed to several factors. As a topical simplification, high-frequency trading (HFT) and the publication of Michael Lewis’ “Flash Boys” revealed that many quant funds were disconnected from intrinsic asset value and were simply playing on inefficiencies in silicon chips, not market dynamics. The gap between quantitative and fundamental analysis reached a zenith. Several of my colleagues from academia assumed that they could pair HFT with alternative data to identify real alpha, but ultimately all but a few of those combinations proved fruitless.

While the profile of quant fund performance skews positively (mean>median), anecdotally it appears that Limited Partners (LPs) have expressed increased interest in diversifying through quant, while experienced GPs have distanced themselves. Many firms have tried to hybridize quant and fundamental approaches, but this has led to real issues.

So, why have a small number of firms realized such outsized and consistent returns? By looking at examples of their approach, we can see that they thought about data and modeling differently. Renaissance didn’t just focus on models, they undoubtedly considered the structure of data. For example, they once considered a thesis based on weather systems and coffee bean production. Renaissance wasn’t just a quantitative modeling company, they were an alternative data ontology integration company. They integrated their data into a holistic, likely graph-based ontology, which allowed them to continue to find alpha.

Now, the primary tools to support this approach have been democratized. Graph databases, which are widely used in the industrial world, are now disrupting sectors by creating webs of networked information. Funds that will win with quant-heavy approaches will need to adopt these tools quickly or risk fossilization.

Knowledge graphs pose an existential threat to many market inefficiencies that have laid dormant for decades. They can show unexpected connections between assets and even asset classes, leading to a more deterministic view of the market. As the financial industry continues to evolve, the adoption of knowledge graphs will be crucial for firms looking to achieve true alpha and exploit market inefficiencies.

Follow us on Medium.com for more insight into the future of Enterprise AI