durpie-git/imc-prosperity-4

This repository is a detailed writeup from a five-person team that placed 4th globally and 1st in Europe, out of nearly 19,000 teams, in IMC Prosperity 4, a 2026 international quantitative trading competition for university students. The competition ran over five rounds, and teams built automated trading algorithms to maximize profit in simulated markets, alongside separate manual challenges based on probability and strategic decision making. The team earned prize money for their result. Rather than relying on the slow official website tester, the team forked and extended an open source backtester and visualizer from a previous year of the competition, letting them run local simulations quickly instead of waiting on the platform queue. On top of that they built a custom dashboard showing price charts, trading volume, profit and loss broken down by product, and position tracking over time, which became one of their main research tools. They also used Jupyter notebooks as a scratchpad for exploratory analysis, testing signals, and tuning parameters, keeping fast experimentation separate from the more careful validation work done in the dashboard. The write-up walks through the strategies used across topics like market making, statistical arbitrage, microstructure analysis, derivatives pricing, signal extraction, and optimization, covering 64 traded products in total, with 50 active by the final round. Beyond describing what worked, the document explains how the team approached research, validated ideas, avoided overfitting a strategy to past data, and managed a large space of possible approaches under tight round deadlines of 48 to 72 hours. The repository is aimed at people interested in quantitative trading, algorithmic strategy design, or how a top competition team organizes its research tools and workflow. It is written in Python and includes the team's backtesting infrastructure, dashboard code, and notebooks alongside the narrative explanation of their approach.