Pay Attention to Proprietary Data

Financial institutions rely heavily on information sourced from vendors, whether that's market or reference data. Yet their most valuable data sets tend to be those they create themselves. Proprietary or "derived" data gives these institutions an edge by providing the ability to spot mispricing of risk, capture alpha and outperform the market.

Given the importance of proprietary data and analytics, one would think most firms would support its production with the most secure, robust and auditable tools and processes in their armory. At the very least, they should make sure inputs and outputs were vetted by enterprise data management (EDM) teams and subject to the same checks and balances as any externally sourced content.

Unfortunately, that is not always the case. Proprietary data is often handled only by the person who created it, both for custody and governance. Often, this data is locked away in a "black box" or Excel spreadsheet, fed by data from a single vendor that has not been validated—completely outside any EDM process.

Data inputs and outputs that are not validated can corrupt critical business processes with bad data. Running analytics in an environment without auditing leaves the firm exposed to manual errors, as changes to the model are difficult to control or detect. Proprietary models may also often by controlled by a single employee. If that person leaves, the firm will be exposed to risks.

The way to mitigate all these risks sounds simple enough—bring derived data and analytics into a secure, auditable EDM system with process controls. Unfortunately, that is much easier said than done. The problem many firms face is that their existing EDM technologies are not well equipped to deal with the complexities of proprietary pricing and risk analytics.

To fully incorporate proprietary data into EDM processes, firms need to adopt EDM technologies that accommodate all business requirements, understanding business logic and data dependencies, even for complex objects like curves, surfaces and cubes. It also means having APIs to deliver data into a full range of tools—whether those are financial analytics platforms such as Numerix or Fincad, bespoke models developed in mathematical programming languages such as R, Python and F#, or a simple Excel spreadsheet.

Regulatory Drivers

Bringing proprietary pricing and risk analytics into centralized, auditable systems and processes can reduce operational risks, but regulations have become an added incentive to do so, starting with the BCBS 239 risk data aggregation principles. Another Basel Committee effort, the Fundamental Review of the Trading Book, a standard for calculating market risk and capital buffers, will impact all trading operations and provide clear business incentive to improve data management systems, processes and governance, in the form of reduction of the cost of capital.

Some data management challenges specific to FRTB, such as the need to demonstrate risk-factor modellability, may require industry collaboration to source as much ‘real' pricing data as possible to meet its criteria. Other challenges, such as back testing and P&L attribution, will require firms to aggregate and analyze internal datasets typically stored and managed across disparate systems, including risk model outputs, historical market data, trading desk positions and transactions.

And just as FRTB will prompt trading, risk and EDM teams to collaborate more closely over data management systems and processes, IFRS 9 will bring finance teams into the mix—specifically requiring accounting teams to more accurately mirror the hedging practices of market practitioners.

In an industry built on the value of proprietary data and analytics, EDM has not received the attention it deserves. But looking forward, there are more drivers than ever for firms' management teams to prioritize EDM.

In addition to reducing operational and regulatory risk, the business imperative to optimize the cost of trading capital may spur the most action. But meeting future business and regulatory deadlines will require firms to focus on more agile EDM frameworks—ones that are unconstrained in the types and temporality of data they can handle.