Bloomberg has combined its cloud-based storage and management for written enterprise communications with Orange Trading Solutions’ voice recording expertise. The combination enables the integration of voice data into Bloomberg Vault to produce electronic records in near real-time, and it will work with any existing trading floor recording system.
“This collaboration allows clients to use their existing voice recording infrastructure, there’s no re-inventing the wheel,” said Harald Collet, global business manager for Bloomberg’s Vault. “We are leveraging data repositories that are out there and using that for analytics.”
The Vault stores written messages such as Bloomberg messages, Microsoft Exchange, Office 365, Notes, instant messages, social media and mobile messaging for about 540 clients who have stuffed 66 billion records, several petabytes, into the Vault.
Now the FSA in England and the CFTC in the U.S. want more voice recordings; for more than a year the FSA has required that any company mobile phone used in trading be recorded.
With the Orange Business alliance, Bloomberg can link voice recordings with written messages for comprehensive archiving and indexing and eventually analytics -- analysis of voice recordings still requires people to listen to the speakers although that may be changing.
Collet said Bloomberg is responding to regulatory demands with the Orange alliance.
“They are a large player in voice recording -- they are in 1,300 trading floors -- so they provide access to all that recorded voice data and an ability to move it into the Vault for analytics and followup for indexing and archiving. Then a compliance officer can run a single search and produce all the consolidated records related to a regulatory request.”
The explosion of data, even in mid-size firms requires automation, Collet added.
“In the the old world of regulatory investigations and e-discovery, you used to retrieve records and review them one at a time. The advent of even mid-size firms producing multiple terabytes of data makes it is impossible to take that approach.
“We need newer approaches. In the unstructured data world, especially text, the big trend is predictive coding or predictive analytics that identify high-risk communications, flag them for followup and use machine learning to train a data set.”
Voice records can provide a lot of information even without listening to the recordings, he added.
“You can map lines to a particular person and see directional flows from one number to another, the frequency of communication and the length of calls. You can use voice indexing technology employing various approaches. Some focus on phonetic search or various other voice indexing technologies whose accuracy rate depends on the quality of the line recording, the language spoken and the dialect. One firm in Switzerland has 16 different dialects on their trading floors.”
You want as little human review as possible because it is expensive, he added.
“We are trying to automate it so a compliance officer can drill down and find all the text communications and see the [related] voice communications.” The company doesn’t offer voice indexing yet, he added.
The move to improve voice recording and indexing received a big boost from the CFTC’s move to require more record-keeping in the derivatives market, including pre- and post-trade reporting.
“Record-keeping is a major component [of CFTC regulation] and since the swaps market is very large and trades were often settled on the phone, recording has a very direct impact on a lot of these firms. They are identifying their requirements, looking at their particular role in the swaps market and many are putting in voice recording solutions. Our announcement is a direct response to that.”
The U.S. has led in electronic record-keeping while the UK has been ahead in requirements for voice recording, especially recording mobiles, Collet added.
Other global markets are moving to increase their requirements, he added.
The financial service industry history of mergers and acquisitions makes compliance particularly challenging because it has led to a mix of technology platforms at many firms.
“All of this data is typically fragmented, with consolidation and acquisitions,” said Collet. “The amount of data is challenging and few firms have a solid handle on unstructured information. With regulators adding voice to the mix, it is causing real headaches for clients.”
