By Brad Foster, Bloomberg’s Global Head of Enterprise Data Content and Daniel Caporaletti, Derivatives Platform Product Manager

More than a decade after the financial crisis, a new threat, the spread of COVID-19, is challenging governments, banks, companies and the global markets as firms execute their business continuity plans and look for new ways to solve problems remotely.

There has been a significant impact to Global Markets characterized by significant intraday volatility, extreme price moves and a lack of liquidity across most asset classes. Unlike the last financial crisis banks are well positioned to capitalize on this increased volatility and, more importantly, service their client funding needs given significant bank regulation they are well capitalized and funded.

As policymakers including central banks and governments respond to this crisis with the many tools available to them, there has also been a significant impact to commodity and FX markets.

The significantly higher levels of volatility have created challenges for long-only funds and for hedges, while at the same time significant trading opportunities for market participants including the growing community of quantitative-driven algorithmic traders.

To adapt and thrive in challenging markets, quants need a vast and increasing amount of real time, high quality, reliable information-rich datasets. Derivatives data provides significant value to clients as levels and changes in option prices can give timely insights into the overall market sentiment and potential movements of the underlying markets and prices.

Trading and risk managing derivatives data can present many challenges. For example, when trading and analyzing equity markets, it is often not possible to have the price of listed options feed directly into trading algorithims. With the proliferation of more strikes and expiries in listed option contracts, capturing the option chains for a single name or index now requires the ability the handle thousands of instruments.

Quantitative-based models require extensive historical time series data — the statistical significance of the calibration increases with the length of the historical set. With the rolling of listed option expiries, each option series has a limited price history, and obtaining constant maturity time series is often impracticable.

More generally, option prices, which have a strong dependency on spot or futures prices, do not allow easy comparison between different underlying products, single name equities, indexes, ETFs and across different asset classes.

Implied volatilities, on the other hand, represent the market cost for optionality in a more comprehensive and transparent way, allowing clients to more easily ingest and use the data in their models and allow for easier comparison across multiple underlyings. Also, interpolated implied volatility pillars with constant maturity and constant moneyness can have the long historical time series which have significant value for algorithmic and quant model training.

For these reasons, implied volatilities are more frequently used by quants as a fundamental input into trading algorithms, with an important growth of consumption into intraday blackbox applications.

For liquid and observable option markets, firms can calculate implied volatilities using listed options prices as main imput and then calibrate volatility surfaces from them. Libraries can be developed in-house or sourced from a vendor. Alternatively, firms can use a third-party software system with technology to consume options market data and derive their own volatilities.

The cost for coding and implementing in-house libraries can be substantial, as there are a number of processes involved in producing a reliable volatility surface including: obtaining and filtering market data, ensuring syncronicity of data, implying forwards, calculating implied dividends, applying a methodology to smooth the volatilities, and taking care of their interpolation and extrapolation.

In addition, the surface fitting must be performed on historical option data as well, and historical time series have to be significant enough to be used for training algorithms.

Purchasing and integrating vendor pricing libraries, systems and infrastructure can involve significant costs — often before any training on the trading algorithms is possible. Accordingly, this often requires a commitment from clients with no certainty around whether the data supports a hypothesis or whether the proposed strategy is able to generate value for clients. Listed options market data needs to be sourced by subscribing to a market data provider, or by going directly to exchanges for direct market access, resulting in additional costs.

When compared to calculating the surfaces in-house, subscribing to a feed of implied volatilities can be a more convenient option from a total cost of ownership point of view: it involves minimal fixed costs, offers faster initialization of the algo training, better scalability and an easier expansion across markets and asset classes.

The vendor selection criteria typically starts with an estimation of how well the provided volatility surfaces replicate the market (accuracy), and also need to include an assessment robustness, stability, frequency of calculations (end-of-day, intraday, how many snaps per day), the length and quality of the historical time series, the fixed and variable costs.

Faster implementati can speed up the algorithm setup processes and empower quantitative trading firms to take timely advantage of trading opportunities.