Measuring and Using Trading Algorithms Effectively
Hosted by Columbia IEOR/Waterloo
Abstract
In order to build effective trading algorithms, you need to effectively measure trading algorithms. In this talk we will talk about what factors we look at when measuring trading engine performance in tuning our algorithms. Specifically we will discuss several common benchmarks and discuss what each of these focus their lens upon, and what these measurements are blind to. We will focus on the precision of these measurements and where these sources of noise and uncertainty come from. We will show a lower bound on the amount of noise expected in these measures so you can determine how precisely one can expect to be able to measure trading performance for a given amount of flow. This has material implications on the feasibility and applicability of quantitative best-execution measures for many users. Finally we will show how use these methods in our engine tuning process.
Bio
Heath Windcliff, Managing Director, is the head of the Quantitative Research team at MS which is responsible for equity algorithmic trading research and development. He actively works on the design of the optimization tools, the limit order worker and the venue selection models used in our products. He is also responsible for the PostTrade analytics framework that MS uses to design and tune the algorithmic offering for the use cases and needs of users and clients. Heath has a PhD in Computer Science focused on numerical methodologies from the University of Waterloo in 2003 following a Masters and Bachelors in Applied Mathematics.