Written by Peter Lee
Published on: Thursday, October 27, 2016
Click here to read the original article on Euro Money
Krzana aims to give users an edge over rival investors by sucking the rare, important signals that will move markets out of the vast and noisy din of financial news media.
Can you remember where you were on the morning of April 9, 2015? If you work for or own shares in Oil & Gas UK, you probably do. That was the day the company announced the discovery under Horse Hill near Gatwick Airport of the largest onshore recoverable oil reserves in the UK for 30 years.
Sandip Sarda, Krzana
The company had drilled exploratory wells a year earlier. The official confirmation of the discovery began to break on news wires shortly after 8am and was picked up by other broadcasters and major news sources, including the BBC, just before 9am. Shares that had closed the previous night at £1.09 ($1.33) started a rapid climb, doubling in price in morning trading and ending the day up at £4.29.
Oh, to have been a hedge fund or commodity trader that morning plugged into Krzana.
Krzana is a new kind of real-time search engine that sucks unstructured data out of thousands of sources – not just financial markets media such as Bloomberg and Reuters, but specialist trade journals, local papers, general news services, regulatory reports, RSS feeds and, perhaps most important of all, social media – and spots the rare pearls that will be valuable to its users amid all the dross.
Krzana is customizable, allowing hedge funds or investors focused on specific sectors or financial instruments to register areas of interest and suck in all news of relevance without combing through all the sources. It is like doing a Google search and instead of trawling back through historical stored data, throwing that search forward into the future and dragging up new findings almost as soon as they become available.
It so happens that investors focused on oil and energy markets were early adopters of Krzana. It delivered news of what was to come from Oil & Gas UK at 7.09am, before the stock had moved.
Someone, perhaps a site worker, had tweeted something: one of those messages intended for a handful of friends and followers – ‘You won’t believe what’s coming at Gatwick’ – that Krzana’s search and filtering algorithms plucked from the 2.5 quintillion bytes of data we now produce every day, spotted could be significant and sent to customers focused on the sector and the company.
Sandip Sarda, chief executive of Krzana – the name translates from Sanskrit as “pearls” – says: “What we do is reach out into the vast pools of unstructured data now streaming in real time from so many tiny sources, such as blogs and social media. We normalize that data, apply our own advanced proprietary and some freely available natural language algorithms to give it structure and then our users can filter it against their very specific interests.”
A lot of fintech companies are looking at technology for this kind of data capture, and banks are reviewing their substantial budgets for both market data and news data.
Sarda says: “The combination of data feeds, our unique sector-based ontologies and filtering gives us an edge to provide the desired output for our customers.”
The results are impressive and not just from tweets about oil discoveries.
“We were 30 minutes ahead of the financial news wires reporting an oil spill in Louisiana because we picked up local newspaper reports,” says Sarda. “You have to remember that the fortunes of many companies and their stocks and bonds are driven by real world events, such as natural disasters, not just financial market events.”
“So, for example, we take direct feeds from all the earthquake monitoring stations or local newspapers that report legal settlements that may have national market implications – especially in the US.”
The key for many users is being able to reduce the time between hearing news and enacting their response – Sandip Sarda, Krzana
Krzana has found that networks of expatriates in certain countries can be a useful source to monitor, because there is a tendency to gossip, for example about contracts being awarded.
Sarda adds: “We were ahead recently in alerting investors of protesters preparing to shut down five pipelines in Canada, plus we picked-up one of the first alerts of the BASF factory fire in Germany last month. We were ahead of the wires and broadcasters on news of a new CFO at Microsoft, just because we picked up a tweet from someone inside the meeting.”
The idea for the company germinated in the first place when two of the founders were due to meet on London’s Southbank on the day of a fatal helicopter crash. People nearby were uploading pictures and accounts for what seemed like an age before the story was reported in any news media and the colleagues got in touch to make sure both were unhurt.
Some investor mandates and governance requirements restrict fund managers from acting on rumours circulating in social media until a story is confirmed by a major respected new source.
Sarda says: “Even so, Krzana aims to give its users enough time to think first and plan the best hedge or trade to put on the very moment a story is officially confirmed. The key for many users is being able to reduce the time between hearing news and enacting their response.”
This sheer proliferation of data sources is not going to get any simpler for banks, corporations, traders and insurers to cope with. Adding to the bloggers, the tweeters – running at roughly 8,500 per second – the weather and earthquake monitoring stations, IBM estimates that 25 billion devices will be hooked up to the internet of things by 2020.
Having started with a number of hedge funds and commodity traders focused on the energy sector, Krzana is gaining interest from across the financial sector and has more than 55 potential customers trialling its service.
These range from small hedge funds to major companies both in energy and banking, including Morgan Stanley, which sees potential value in what Krzana offers both in new methods for real-time data mining and perhaps also in reducing their reliance and spend on news aggregators such as Bloomberg and Reuters.