Project Description

Overview

This project implements a simple automated trading system that combines:

• Market making
  
• Sentiment-based news trading
  
• Risk-controlled execution
  

The system runs in a continuous loop and reacts to both market data and social-media feeds.

---

1. Market Data Handling

Mechanisms Used

• Polling live market data from the exchange
• Tracking mid-price history
• Tick-size-aware rounding for quotes

Ideas Implemented

• Quote around the current mid price
• Control quote width using a spread factor
• Stay within position limits

---

2. Market Making Engine

Mechanisms Used

• Inserting bid/ask limit orders
• Canceling stale orders
• Adjusting quotes when mid price changes

Ideas Implemented

• Keep a two-sided market at all times
• Maintain balanced inventory
• Avoid directional exposure

---

3. News-Driven Sentiment Trading

Mechanisms Used

• Polling real-time social-media feeds
• Running a transformer-based sentiment model
• Mapping sentiment to BUY / SELL / NONE
• Filtering decisions by a confidence threshold

Ideas Implemented

• Positive sentiment → short-term BUY
• Negative sentiment → short-term SELL
• Use IOC orders for immediate execution
• Neutralize existing positions before reacting to news

---

4. Linked Instruments

Mechanisms Used

• A mapping from news identifiers to affected instruments
• Applying reactions across all linked assets

Ideas Implemented

• Each news event can impact multiple markets
• Delete old orders before executing new sentiment trades
• Ensure consistent behavior across all correlated instruments

---

5. Position & Risk Management

Mechanisms Used

• Hard position limits per instrument
• Automatic neutralization when limits are exceeded
• Protective cleanup before new trades

Ideas Implemented

• Never accumulate excessive exposure
• Keep risk small during news reactions and quoting
• Prioritize flattening positions before new decisions

Built With

• python