London’s Krzana storms into big data league with energy traders and hedge fund clients
Krzana provides data-hungry companies like hedge funds with relevant signals drawn from over 10,000 sources.
London-based real-time search specialists Krzana are disrupting the marketplace for financial data by signing up energy focused hedge funds and equities watchers. Krzana provides data-hungry companies, such as hedge funds looking to trade quickly on relevant signals (drawn from over 10,000 sources), with a simple to use dashboard that filters news, social media, RSS feeds, blogs and pictures.
The platform, which has been built to deal with half a billion tweets a day, has an elegant and agile system of filters (eg. powerful enough to process real-time data streams) so users can decided what is relevant to them and arrive back a point where less is literally more.
The company’s recent success is also being driven by its pricing, which is also designed to be disruptive: licences can start from as little as £200 per user per month.
A powerful illustration of how Krzana can help traders find alpha concerns a recent fire that broke out at Motiva Enterprises’ oil refinery in Louisiana. Before Bloomberg, Reuters or Dow Jones had the news, an energy trading hedge fund client of Krzana had accessed the first tweets about the fire coming from a local Baton Rouge-based newspaper.
In the intervening, crucial ten minutes or so the hedge fund made a tidy profit. The margin for refined fuel products, known as the diesel crack spread, widened 7%. Meanwhile, expectations of reduced demand from Louisiana for oil in need of refining sent the cost of the local Gulf Coast Medium Sour benchmark tumbling 25 to 35 cents a barrel.
Sandip Sarda, CEO of Krzana, told IBTimes UK: “Krzana picked up the tweet before any of the majors. Our engine consists of multiple data feeds that flow into our custom filters, which we call ontologies, and they are then processed before being displayed. So for example, our energy module Krzana Energy Terminal, is very energy-sector specific. It can include earthquakes, seismic data, weather events, pipelines, transmissions, anything to do with energy-related structures, the businesses and companies involved in the sector.
“Typically news breaks at a local point, and the ability of the user to get that information before it comes on to the main news feeds is absolutely critical. That’s what gives them the edge.”
Geoffrey Todd, chief marketing officer at Krzana, explained that the two core modules it delivers today are equities and energy. Todd said: “We train the engine specifically around those sectors. Our focus today is on financial markets, but we are also looking at others; there are lots of siren calls in other risk environments and also security – and of course from a media perspective, it’s all news.”
Krzana’s technology team is leading the shift out of an era dominated by detailed market data and broad, sweeping fundamentals data into an era of highly-granular, ordered data, generated by machine learning and artificial intelligence.
One way to deal with the noise that social media generates, and the risk of users reacting to false positives, is allowing Krzana to provide “story aggregation”. “If one tweet comes in and then another 100 come in, it will show up as a cluster, as opposed to each just being an errant tweet in its own right,” said Sarda.
Krzana says it takes five minutes to configure its filters and then the cloud-based platform is up and running. It’s an inexpensive “no brainer”. Todd said that as more users trial it, he has noticed a tendency to switch off the Twitter feeds in order to compare and contrast data outputs.
Todd said: “Don’t get me wrong – we love Twitter’s raw data power. But you can easily set up two near identical search channels with the same sets of qualifiers, words and domains you are looking for, with Twitter active in one and not in the other.
“Then there is also the ability to exclude terms. If somebody is trading gold, say as the Olympics were in progress, we could exclude the word ‘Rio’. We still want ‘gold’ but as we see the results emerging, we can exclude irrelevant ancillary terms in a very agile way.
“So we can refine that term from the Twitter stream, and perhaps keep running it separately. We find Twitter is also a good place to pick up pictures for example. It’s the agility of the search tool that counts.”
In addition to filtering data on an external basis, there is also a growing opportunity to partner with the big systems’ integrators who are well placed to engage on large scale BPR projects. “There’s the social feed, there’s other external data. But then how do I also monitor the chatter in my email system? Well you can only do that by being inside your firewall,” said Todd.
“Then I might want to go back and look into my archive. Perhaps I want to compare and contrast something; what’s being said about the history of a product sign-off, what do people know and what’s out in the market place? Organising and ordering multiple big data and real-time sources is a growing challenge – there is a significant integration opportunity for embedding this kind of engine.”