Two and a half years ago, Germán Soto Sanchez backed his own belief and moved from a major US bank to a technology provider making fast progress with its cloud-based voice communications and data platform for institutional traders. Radar caught up with him to find out what is rocking the voice capture world.
During his time at JPMorgan Chase, Germán Soto Sanchez led the firm’s investment in Cloud9, a cloud-based communications and data platform for financial firms and institutional traders. He was so taken with what he found at Cloud9 that he made the move to become the company’s president. Watching the voice trading market rapidly transform over the last few years, he has followed three major themes that are now playing out in favor of Cloud9’s focus on next-generation voice and data capture.
“Cost is always the first item that firms look to reduce,” he said. “The capital expenditure around the trader desktop is significant, especially around trader voice. Trader voice is the third most expensive trading floor line item, only surpassed by the costs associated with people and data. Bank costs can be in the tens of millions with a traditional legacy voice trading turret system. A turret can cost $8,000 or more and will last five to seven years, but then there’s recording equipment, private lines, etc., thus leading to soaring infrastructure expenditures.”
“New regulations (such as MiFID II and Dodd-Frank) comprise the second theme. Historically, voice has been viewed negatively in the communications surveillance arena, as it has been very hard to monitor,” Soto said. Specifically, “a lot of surveillance and compliance activity in trader voice has been and continues to be manual, making it very expensive and difficult to do with accuracy and precision. This is exacerbated by the 30 to 40-year-old technology at the heart of the current legacy voice trading turret infrastructure. Think about the need to know who your counterparty is, to timestamp accurately, to record all inquiries – it’s very hard to achieve this with the legacy systems since they were not built for this and are unable to collect the necessary data.
“For example, one major global bank has a team of people manually recording trade inquiries. There needs to be an electronic trail, transcription, an automated timestamp, or some kind of path to structured data. The lack of these resources is why the industry has been so ripe for innovation.”
The third theme is, somewhat unsurprisingly, the advent of big data.
“Data is king. If you have proprietary data, you need to exploit it to generate alpha for the business. You need a better sense of who your people are talking to and what conversations are happening. You might get ten quotes for a security in which you’re interested and only use one, but all that other data still has value.” This can make the sales and trading staff much more effective, Soto said.
Today, with Cloud9, the quality and fidelity of the voice records and related metadata is “amazing”. “You have unique data that was previously unavailable but can now be retrieved instantly. You have recordings that can be transcribed with far greater accuracy than with legacy turret systems. As a result, voice is no longer a nightmare to manage; you can electronify it, mine it, and integrate it with your CRM systems. This is why Cloud9 was purpose-built for the next generation needs of institutional traders.”
Cloud9 is currently witnessing heavy traction across the industry. “We have more than 5,000 users on our platform generating over 1.1m trader conversations per day. We have a big presence in commodities and fixed income, and we are growing across other asset classes.”
However, it takes time to digest such a disruptive technology, Soto warns. Although Cloud9 will remove a huge chunk of the existing infrastructure cost, it is a big change that can make some people uncomfortable. Just a few years ago, we had banks on the record stating an avoidance to the cloud, so we’ve come a long way. Cloud9 adoption is becoming a very strategic product decision as we are viewed as a critical piece of the workflow that can add value; no longer just a phone system.”
“As a result, heads of trading and senior management are getting involved in decisions that previously only involved technology staff.”
“Analytics, biometry, transcription, keyword spotting, it’s all a reality now,” Soto said. “Turret transcription accuracy used to be about 50 percent on average, 60 percent at best.” “But we can hit 90 percent due to the voice fidelity on our system,” he said. “We also collect much more metadata around the conversation so there is a rich context that can be mined. We can create a biometric model to authenticate the identity of the caller and access, security and logins can all be controlled much more easily. These are the benefits of a software approach to a traditional hardware-based business.”
In its roadmap, Cloud9 is investing more in the development of its voice technology – a core trader communications and data platform with data generation and delivery capabilities. “We must be innovative as to how we generate and deliver data,” said Soto. “This proprietary data will drive next-generation analytics, which is something our customers are keen to have. We are working on unique data fields about trader voice conversations and are investing in interoperability and APIs to facilitate the delivery of data in real-time to our customers and partners.”
The importance of voice in the markets is certainly not diminishing, Soto said. “While voice trading has decreased, the most profitable trades and certain asset classes are still dominated by voice. While traders still want to speak to their counterparties for complex trades, the back office, research teams, private bankers and any other regulated voice users are all seeking premium tech support as well. With the digitization of voice, we envision numerous new use cases on the trading floor.”