Market Making and the IMC Competition
15th in the US out of 12,600 teams — and what trading under latency teaches you.
In April 2025 I placed 15th in the United States out of 12,600 teams in IMC's Prosperity 3 trading competition — the top 0.12%.
What the competition actually is
You write agents that trade in a simulated market against other people's agents, under real constraints: latency, position limits, and a The Limit Order Book that responds to what you do. Passive strategies get picked off; aggressive ones bleed on the spread.
What I built
Fair value estimation. Before you can quote, you need a belief about what something is worth. Get this wrong and every subsequent decision compounds the error.
Dynamic bid shading. Quote too tight and you're adversely selected by informed flow — you only get filled when you're wrong. Quote too wide and you never trade. The shade has to move with your inventory and with how confident you are.
Signal-driven quoting. Mean-reversion patterns and synthetic mispricing between related instruments, used to skew quotes rather than to take directional bets.
The lesson that transferred
Market making is Optimal Execution with the sign flipped. An execution agent has inventory it must shed and asks how to shed it cheaply. A market maker takes on inventory it didn't ask for and must decide what price makes that worthwhile. Both are inventory-risk problems, and both live or die on how well you model Market Impact and adverse selection.
Both are also, fundamentally, problems where the distribution of outcomes matters more than the average — a competition rewards the strategy that survives, not the one with the best expected value on a good day.