Are we getting the trades we want? How fast should we get? Understanding exchange networking is essential to success in technology-driven trading. However, network dynamics tend to be complex, and the data we get provides only a part of the picture. We need a model that can fill in the gaps in the data, help us understand these dynamics, and predict the results of the changes we introduce.

There are multiple approaches to tackle this problem, depending on the model complexity and the amount of data/knowledge to incorporate; and Python ecosystem provides a wide arsenal of tools to explore these paths. In this presentation I will show examples of probabilistic modelling, discrete-event simulations and supervised learning, using modules such as scikit-learn, pymc3, and simpy. I will discuss the trade-offs and challenges of each approach.

Basic knowledge of computer networks, probability and PyData stack may be beneficial, but no special knowledge is required to follow the presentation.