📈 Trading Bot — PoC × Ramify Hackathon

A quantitative portfolio allocation bot built for the PoC × Ramify Hackathon.

---

📖 Overview

This project implements a quantitative trading algorithm in Python that dynamically allocates capital between two assets and cash. The core idea is a custom Channel Breakout strategy: it measures where each asset's current price sits relative to its recent high/low range, then translates that into a portfolio weight.

The bot was submitted during the PoC × Ramify Hackathon, where participants competed to maximize risk-adjusted returns on a provided historical dataset.

---

🧠 Strategy — Channel Breakout

Signal Generation

For each asset, the algorithm computes its normalized position within a rolling price channel:

$$\text{Position} = \frac{\text{Price} - \text{Lowest}_{n}}{\text{Highest}_{m} - \text{Lowest}_{n}}$$

This gives a value between 0 (at the bottom of the channel) and 1 (at the top), indicating the relative strength of the current price move.

Asset	High window	Low window	Sensitivity
Asset A	30 periods	10 periods	Squared (^2.0) — noise filter
Asset B	13 periods	4 periods	Power 0.7 (^0.7) — aggressive

Allocation Logic

1. Asset B is allocated first (higher priority / higher volatility play).
2. Asset A fills the remaining allocation proportionally to its signal.
3. Whatever is left goes to Cash — the bot stays defensive when signals are weak.

weight_B    = clamp(signal_B × 10)
remaining   = 1 - weight_B
weight_A    = remaining × clamp(signal_A)
weight_Cash = 1 - weight_A - weight_B

The bot holds 100% cash for the first 35 epochs while building a sufficient price history.

---

🗂️ Project Structure

trading_bot/
├── bot_trade.py   # Core strategy: signal calculation & allocation logic
├── main.py        # Entry point: data loading, decision loop, scoring
└── README.md

File	Role
bot_trade.py	make_decision(epoch, priceA, priceB) — the brain of the bot
main.py	CLI runner, CSV parsing, decision validation, scoring display

---

🚀 Getting Started

Prerequisites

• Python 3.8+
• pandas
• matplotlib

Install dependencies:

pip install pandas matplotlib

Note: This project also depends on a scoring/ module provided by the hackathon organizers (not included in this repo). Without it, main.py cannot run, but bot_trade.py — the core strategy — is fully standalone.

Running the Bot

python main.py <path_to_prices.csv>

Display the performance graph:

python main.py <path_to_prices.csv> --show-graph

Expected CSV Format

The CSV file must have two columns (Asset A, Asset B) indexed by epoch:

epoch,Asset A,Asset B
0,100.0,50.0
1,101.2,49.8
...

Decision Validation

Every decision is automatically validated to ensure:

• Keys are exactly {'Asset A', 'Asset B', 'Cash'}
• All values are in [0, 1]
• Weights sum to exactly 1.0

---

📊 How It Works — Example

Epoch 50:
  Asset A price: 105.3  →  channel pos (30/10): 0.72  →  signal: 0.52
  Asset B price: 48.1   →  channel pos (13/4):  0.88  →  signal: 0.91

  weight_B    = min(1.0, 0.91 × 10) = 1.0   → 100% in Asset B (strong breakout)
  remaining   = 0.0
  weight_A    = 0.0
  weight_Cash = 0.0

  Decision: {'Asset A': 0.0, 'Asset B': 1.0, 'Cash': 0.0}

---

👤 Author

Mathis Haba — Second-year student at Epitech

Passionate about algorithms and quantitative finance.

• 📧 mathis1.haba@epitech.eu
• 🐙 @mathish06

---

📄 License

This project is licensed under the MIT License — see the LICENSE file for details.