- View all blogs
- Hivemind beats four other finalists to win the prestigious FIA Innovator of the Year Award for 2019
Hivemind beats four other finalists to win the prestigious FIA Innovator of the Year Award for 2019
Chrissie Cormack Wood •
At the tail end of last month, Hivemind’s Head of Data Science, Christian Gilson, travelled to Chicago to take part in the annual showcase for fintech start-ups organised by the FIA. The selection committee—made up of technology experts from banks, trading firms and venture capital investors—selected 20 companies from the original 45 that applied to take part in the event. From these, a shortlist of five were invited to present a competitive pitch to a panel of independent judges.
Commenting on Hivemind’s winning pitch, the judges said they chose Hivemind because “the software platform has many potential applications for firms seeking to gain insights from the vast amount of unstructured data in the world today”.
You can read the transcript of Christian’s winning presentation below…
“Hivemind’s software provides a really pragmatic and flexible solution to data preparation and it was originally developed within the data science team that I was part of for five years at Winton Group. Winton is a $30 billion quant investment manager based in London, with most of its AUM in Futures.
Hivemind was spun out of Winton in the summer of last year and in February of this year, we raised our Seed Round funding getting investment again from Winton, as well as from Barclays and Fidelity International.
Hivemind was designed for data-driven companies and businesses that want to add more data into their decision-making. Companies learn about the world analytically, using the tools of data science, and for that, data needs to be structured and searchable and clean. But that's not data's natural state.
Knowledge is very often shared with the world in ad hoc, unstructured formats. It happens internally in your companies, in emails, in memos and presentations and it happens externally to your business in regulatory files, annual reports, websites, tweets, and news feeds. And transforming all of that unstructured content into useful tabular, structured data, that's fit for analysis is hard.
It's hard because there's a lot of it. It's hard because it's often ambiguous and unfocused. It's hard because really it was designed for human consumption, not that of machines.
And because companies can't transform this data into data sets for analysis, they can't take advantage of the information assets within the business—and this has the knock-on effect of companies using the exact same data as their competitors.
If companies are structuring their data, it’s often their senior staff who spend a lot of their time digging into dense documents, extracting a piece of pertinent information, and then sticking it into an Excel spreadsheet.
Hivemind's solution to this problem is centered around the principle of collective intelligence, that of man and machine. Machines can do a lot of the heavy lifting, but to solve the problem well requires a human in the loop. Humans are flexible, they have sophisticated intelligence, they have domain expertise and they inherently understand context which means they can make great decisions when faced with uncertainty and ambiguity in those types of unstructured sources.
In a nutshell, Hivemind provides a really practical framework to implement collective intelligence for data preparation. It helps clients solve really complicated data problems by breaking them down into an assembly line of micro-tasks, where those tasks represent conceptually very simple questions. Those tasks are then distributed to humans, or machines, as appropriate.
To put all of this into a business context, here are some of the ways our clients have used the Hivemind software platform over the past year:
It’s been used to monitor speculative position limits, margin rates, and trading hours from exchange websites.
It's been used to map shipping companies to the berths and terminals they operate out of, from really dense shipping company documents. And then it's been used to map those terminals to commodities, from port and terminal websites.
It's been used to clean open-high-low-close futures prices from newspaper archives. And to extract bond prices to then synthesise them into Bond futures.
It's been used to collect agricultural supply and demand estimates from USDA filings.
I’d like to close by summarising the four main benefits that I’ve discussed today that Hivemind brings to its clients.
Firstly, the ability to extract pertinent information from unstructured sources that previously would have been inaccessible.
Secondly, the ability to build your own, bespoke, proprietary data sets that nobody else has, thereby providing a competitive advantage.
Thirdly, the ability to provide transparency to your data, with a full audit trail.
And lastly, the ability to significantly reduce the amount of time spent by senior staff on trivial data work.