Skip to content

README.md

Latest commit

 

History

History
90 lines (62 loc) · 3.5 KB

README.md

File metadata and controls

90 lines (62 loc) · 3.5 KB
Raw

Round 5 - The Final Stretch

Overview

This folder contains the final algorithmic trading system developed for Round 5 of the IMC Prosperity Trading Challenge.

This round represents the final stage of the competition, where the objective is to maximize cumulative PnL across a newly introduced universe of assets while adapting to a complete reset of tradable instruments.

All previous round assets have been removed. The strategy must be fully redefined for a new and significantly broader market structure.

In addition to algorithmic trading, a secondary opportunity is available in the Ignith market, where external informational signals (Ashflow Alpha) may be used to guide discretionary trading decisions.

Market Environment

Algorithmic Trading Universe

A total of 50 new tradable instruments are introduced, grouped into 10 categories of 5 assets each:

  • Galaxy Sounds Recorders
  • Vertical Sleeping Pods
  • Organic Microchips
  • Purification Pebbles
  • Domestic Robots
  • UV-Visors
  • Instant Translators
  • Construction Panels
  • Liquid Breath Oxygen Shakes
  • Protein Snack Packs

Each instrument has a strict position limit of 10 units.

This round significantly increases the dimensionality of the asset selection problem, requiring prioritization and filtering of exploitable structures.

Strategy Overview

The trading system is designed around a selection-first architecture, where the primary challenge is not execution, but identifying which instruments contain persistent inefficiencies.

1. Asset Selection Layer

  • identification of instruments with stable statistical structure
  • elimination of low-signal or noise-dominated assets
  • clustering of similar behavioral patterns within categories

2. Strategy Allocation Layer

  • assignment of appropriate trading logic per asset class
  • mean reversion strategies for range-bound instruments
  • momentum or breakout logic for structured trend assets
  • avoidance of overfitting across highly heterogeneous products

3. Risk and Position Control

  • strict adherence to low position limits (±10)
  • diversification across categories
  • reduced concentration risk due to small inventory constraints

External Market: Ignith

A parallel trading opportunity exists in the Ignith market, consisting of 9 tradable goods.

Trading decisions may incorporate:

  • Ashflow Alpha informational feed
  • market sentiment signals
  • cross-market structural observations

This component is treated separately from the algorithmic system.

Repository Structure

  • round_5/selection/ → asset filtering and product ranking logic
  • round_5/strategies/ → trading logic per selected asset group
  • round_5/utils/ → shared utilities for execution and analysis
  • round_5/notebooks/ → exploratory analysis and feature evaluation
  • round_5/backtesting/ → simulation framework for final evaluation
  • round_5/ignith/ → external market trading logic (Ashflow Alpha-based)

Key Assumptions

  • not all assets contain exploitable structure; selection is critical
  • cross-sectional filtering improves performance more than per-asset optimization
  • strict position limits require high turnover efficiency
  • informational edge in Ignith is separate from algorithmic edge in core market

Notes

This final round shifts the problem from:

  • strategy design → strategy selection
  • execution optimization → information filtering
  • individual asset modeling → portfolio of micro-strategies

The primary challenge is identifying where not to trade.