GSA Capital Technology: “Just a matter of time”

Abstract:

When you write trading algorithms that react to events in real-time, it's natural to want to simulate how they would behave for historic sequences of events. You could isolate the algos in various ways to enable this testing, but what if you build your entire codebase to be temporal and backtestable?

This talk will cover some of the challenges and benefits of making it possible to 'replay' production code for any period of time, along with the interesting consequences it has on the way software is developed overall. As part of this, I'll touch on some different types of temporal data that we encounter, tactical ways of managing those datasets, and how the line between simulations and software tests becomes increasingly blurred.

About us:

GSA is a collaboration of approximately 170 people. For over a decade we have worked together creating and applying innovative technologies to solve problems in the investment management industry. Over 60% of our staff work in research or on developing technology necessary to enable and monetise our research. We are an eclectic mix of mathematicians, scientists, programmers, economists and a great deal more. We live by the scientific method in building quantitative models that analyse and predict real world behaviour. We implement algorithms based on these models to systematically invest in markets. Our approach has won us many accolades in a sector more traditionally associated with pin-striped suits than harnessing the predictive power of big data.

We welcome students of all academic levels from the Computer Science, Engineering, Mathematics, Statistics and any other closely related disciplines are welcome to attend. The event will start with a short presentation followed by informal chats with representatives from GSA over nibbles and drinks.