IMC Prosperity 4 — Alpha Hunters Padua

Quantitative trading strategies and research for the IMC Prosperity Trading Challenge 2026, a global algorithmic + manual trading competition run by IMC Trading on a series of fictional markets.

Results

Ranking	Placement	Field
Overall	208 / 18,803	all teams
Algorithmic track	184 / 18,803	algo-only score
Italy	2 / 61	Italian teams

Field: 30,000+ participants, 1,549 universities, 117 countries.

Format

Five rounds, each opening a new market with its own instruments, position limits, and mechanics. Every round is scored independently on cumulative PnL; phases 2+ reset the leaderboard. Each round combines:

• an algorithmic track — a single Strategy.py submitted against the IMC datamodel simulator (TradingState in, Order lists out), backtested on provided historical price/trade data;
• a manual track — discrete decision problems scored separately.

This repo holds the submitted strategy plus the research that justified it for each round.

Rounds

Round	Codename	Market	Detail
1	Trading Groundwork	2 spot assets (ASH_COATED_OSMIUM, INTARIAN_PEPPER_ROOT)	README
2	Growing Your Outpost	Same 2 assets + Market Access Fee auction	README
3	Gloves Off	Spot + VELVETFRUIT_EXTRACT options (VEV_4000–6500)	README
4	The More The Merrier	Same universe + counterparty visibility	README
5	The Final Stretch	50 new assets, 10 categories × 5	README

Per-round instruments, mechanics, and strategy rationale live in each round's README. Research subfolders (where present) carry the analysis behind the strategy, with their own READMEs.

Repository structure

.
├── ROUND 1/          # spot market making + signal exploitation
│   ├── research/     # price-structure analysis (per-asset plots)
│   ├── README.md
│   └── Strategy
├── ROUND 2/          # refined R1 strategy + access-fee bidding
│   ├── README.md
│   └── Strategy.py
├── ROUND 3/          # options: IV smile, term structure, gamma scalping
│   ├── research/     # smile/reversion/deep-strike analysis + findings
│   ├── README.md
│   └── Strategy.py
├── ROUND 4/          # R3 universe + counterparty analysis
│   ├── research/
│   ├── README.md
│   └── Strategy.py
├── ROUND 5/          # 50-asset selection: lead-lag + fingerprinting
│   ├── research/     # cluster lead-lag pipeline + category fingerprints
│   ├── README.md
│   └── Strategy.py
├── README.md
└── LICENSE

Tech stack

Python. Strategies target the IMC Prosperity datamodel API (single-file submission, no external deps at runtime). Research uses:

• numpy, pandas — data handling and numerics
• scipy — statistics, optimisation, hierarchical clustering
• statsmodels — VAR, ARMA, ACF/PACF, ADF, GARCH-adjacent diagnostics
• scikit-learn — scaling, Lasso, SVD
• matplotlib — figures

Per-round results

Round	Final PnL
Round 1	96,632 XERICS
Round 2	89,294 XERICS
Round 3	61,613 XERICS
Round 4	219,004 XERICS
Round 5	33,264 XERICS

Team & contributions

This repository contains work developed collaboratively during the challenge. Contributions were distributed across research, implementation, strategy design, manual trading, and review.

Team member	Main contribution
Alessio Dal Pozzolo	Strategy design and experimental implementation, translating theoretical ideas into executable trading decisions.
Enrico Berto	Sole contributor to manual trading, with responsibility for discretionary execution and live strategy review.
Giorgio Cottini	Data analysis and research, signal discovery, market-structure analysis, and parameters optimisation for strategies.
Harish Jawahar	Data analysis and research, strategy development and implementation, and review of trading ideas.

License

See LICENSE.