First published in Tabb Forum, February 2021

Today, data and analytics play an increasingly important role in the workflows of all capital market participants as they seek to uncover hidden patterns that can be used to predict the future market moves. Yet utilising vast data sets for alpha generation is both labour and capital intensive as firms lack in-house resources to manipulate large data sets and face prohibitive cost of acquiring and analysing that data.

However, the rise in processing power has enabled more firms to start looking towards granular Level 3 data and use statistical techniques to uncover hidden patterns in the data that can be used to predict the future. The more granular the data, the more powerful the prediction. Whereas historically, granular order by order (Level 3) data was the preserve of the most sophisticated hedge funds, it is now more widely available to all market participants thanks in part to efforts of exchanges and trading venues in supplying clean data, albeit in a raw unprocessed format and data distributors delivering that data to end clients.

Industry use of Level 3 data to improve alpha generation

The buy-side community is responding to the availability and predictability of this powerful new data set by investing heavily in resources, capabilities and partnerships to drive their never-ending quest for alpha.

In our latest survey we set out to explore how market participants use Level 3 data and analytics in their daily trading decisions and found that high quality data has become a commonly utilised commodity by most market participants who manage their AUM quantitatively.

We found that:

  • 74% of respondents use Level 3 data in their research program to improve alpha generation and mitigate risk.
  • 64% of respondents said that at least 50% of their investment in new data and analytics capabilities will be from buying-in these capabilities.
  • 42% of respondents use Level 3 data for strategy backtesting, e.g. feature generation, alpha generation and model validation.
  • And of the respondents not using Level 3 data in their research programs – nearly 75% said the main reason was its inaccessibility.

What we also found is that 41% of respondents said they will increase their budget allocations significantly for third-party data as a key element of their quantitative research.

And I think it should come as no surprise that many investment firms are looking to partner with third parties to outsource data and analytics generation. Companies that understand the benefit of partnering with a firm that can provide such data and analytics will be in the best place to leverage a more efficient quantitative trading division and generate more repeatable and scalable alpha.

Market expert Dr. Elliot Banks, who directs product development at BMLL, said: “The fact that 74% of respondents are using Level 3 data is a clear indication of the value of this data for research purposes. Level 3 is the only data set that gives you enough information to reliably predict future market states using statistical techniques. Given where the market is going, it is unsurprising that so many firms are starting to use this level of data.”

The next 12-18 months will be critical for all firms looking to embrace data and analytics to help drive increased risk-adjusted returns, as they face increased pressures and competition from cost-effective alternatives. It is clear that for this to happen only the Cloud can offer the critical storage and compute scalability to secure and speed up the analytical process and diversify the data stream. Over 80% of respondents said they were likely to or are already embracing Cloud for their data and analytics generation and processing over the next 12-18 months.

Data analytics is changing

The investment management industry is, at heart, and always has been, a data processing industry. But the ways in which that data is now collected and analysed is being transformed.

Our survey shows the extent to which AI is now used to find more patterns and relationships between asset prices and data from other alternative data sources. Its end result is that portfolio managers and researchers are more effective and better equipped to deliver predictive alphas, faster and at a lower price point than before.

Speed, in the old sense, for asset management is over. No longer is it enough for investors to be slick in teasing trends and investible insights from traditional information sources of trading figures, market shares and economic updates.

Today’s success relies on a different kind of speed (and scalability) and a different way of doing things. And neither length of experience, brand prestige or even the quality of client relationships will insulate them from this. The new speed means demonstrating value to existing and future clients by keeping pace with the digital opportunity.

But adapting to data-driven decision-making brings challenges. Our survey highlighted many of these. Among them is the glaring need for data science talent, new business processes, data governance and the organisational muscle to capture real value from analytics. Predictability only comes from a deep understanding of how the market behaves. The quest for alternative sources of alpha generation ‘twas ever thus…this fact has not changed – the route to it has.

The research, undertaken in Q4 2020, surveyed 100 Heads of Data, Chief Data Officers and Data Scientists at buy-side firms with $10-50bn in AUM, in EMEA and North America, to find out how capital markets participants use predictive data and analytics in their daily trading decisions.

The full report is available here: Buy-side usage of Level 3 data analytics for algorithmic performance”

BMLL’s data and analytics are used by leading institutions across the capital markets ecosystem for alpha generation, to understand how markets behave, back-test trading strategies and to make more informed trading decisions. BMLL’s Data Feeds are pre-computed from the most granular, Level 3 order book data and provide market participants with actionable insight needed to outperform the competition. The BMLL Data Feeds cover venue analytics, trading costs, alternative pricing, pricing analytics and trading analytics.