Trounceflow is a friendly data company: we get data and make it human and machine-friendly. We're just getting started, so we're niche, focusing on the kind of data (and tools) about investor behaviour that helps investment professionals make money trading emerging-market (EM) rates & FX.
Some of this data is the kind a macroeconomist would recognise; capital flows and IIP's; data that tells you about the properties of the investments themselves, but we like to work in real-time, not "last quarter". But most of our data is about the properties of the investors, not the investments. A lot of our data is based on what mutual funds are doing, and again, we like to work with almost real-time data.
The best data type is positioning data. It tells you who is long and who is short. But often we have to make do with the second best data: flow data, which is like a 'first-differential' of positioning. Frankly, you need more than just data to get an edge, and helpfully our founder Mike Trounce has "stood on the shoulders of giants" developing analytical frameworks for data-driven monetisation of investor behaviour in his past life as a strategist at JPMorgan amongst other places.
The service - and the investment edge - comes from a mix of forensic science (anyone remember Quincy?), detective work (anyone remember Columbo? - "Just one more thing...") and a healthy dollop of confirmation bias. So, it's positioning - or more often than not, flow - data and it's led by Mike, hence the name Trounceflow.
Who buys it? Hedge funds, real-money, the sell-side and the IFI's. Not everyone wants it - it's a bit like Marmite, you tend to either love it or hate it - and a lot of people who hate it tell us that flows are either a coincident or a lagging indicator of prices. You know what? We agree! (except when we don't).
Anyway, as a startup we are very grateful to our customers that do like it enough to pay for it, and we try and treat them pretty well, by giving them full access to our codebase, our methods, our team - everything, really, including building bespoke stuff like spiders, parsers, and something called a "tool for the local, semi-parametric detection of regime changes in a bivariate time series setting in which regime changes can be defined with respect to a given bivariate mapping (eg. correlation, tracking error) and a kernel weighting parameter that controls the fidelity of the estimator to local changes". Yowch. Bottom line: we like the 'radical transparency' thing and if you have an idea that's in our wheelhouse, we'll have a crack at it.
Data Entry Specialist
Jr. Full-Stack Developer
Teresina, Piauí, Brazil
Junior Backend Developer
Mike works directly with our emerging-market focused customers to give them an edge.
Data Team Leader
Saint Petersburg, Russia
Software QA specialist
Data & Analytics
Trounceflow is powered by Python and Django, providing us with both the flexibility and robustness to collect and display our unique data. Our needs for data processing evolve in critical parts of the application. In order to achieve superior performance, we gradually move away from standard ORM-based data processors toward raw SQL implementation.
Our spiders continuously check data sources and trigger parsers which autonomously extract information to update our database and charts.
The Twelve Factor App methodology inspired our architecture.
We use Heroku to manage and deploy our application code and Postgres databases to manage logs, queues, batch jobs and asynchronous workers.
Amazon S3 lets us store market data without worrying about running out of space.
We've also adopted Slack for the superb team communication it allows, JIRA to manage our constant flurry of projects.
Company number: 10367410
Level 1, Devonshire House,
One Mayfair Place
London W1J 8AJ
Call us at: (44) 7887 546080