CogniTrade

A QHacks 2026 Project

AI-Powered Trading Psychology Platform

Identify, understand, and overcome psychological trading biases with machine learning

🎯 Overview

CogniTrade is a comprehensive trading psychology platform that combines paper trading simulation with AI-powered behavioral analysis. It uses machine learning models to detect common psychological biases in your trading patterns, helping you become a more disciplined and profitable trader.

Why CogniTrade?

Most traders don't fail because of bad strategies—they fail because of psychological biases. CogniTrade helps you:

📊 Track your trades with a realistic paper trading simulation
🧠 Detect psychological biases using ML models trained on trading behavior
💬 Get AI-powered insights from an expert trading coach
📈 Improve your discipline with educational content and quizzes

✨ Features

📊 Paper Trading Simulator

Full-featured paper trading with real market data powered by the Alpaca Markets API:

Market & Limit Orders — Execute buy/sell orders with market or limit pricing
Stop-Loss & Take-Profit — Set automatic exit levels for risk management
Watchlist Management — Track your favorite stocks with live price updates
Position Tracking — Monitor P&L, average cost basis, and portfolio allocation
Order History — Complete log of all trades with timestamps and notes
Voice Memos — Record voice notes for trades using speech-to-text (Gradium API)
Interactive Charts — Real-time candlestick charts with multiple timeframes
Persistent Data — All data saved to localStorage for consistency across sessions

🧠 AI Behavioral Analysis

Upload your trade logs (or use your in-app trades) for comprehensive psychological analysis:

Machine Learning Models

Model	Detection Target	Algorithm
Overtrading Detector	Excessive trading frequency	XGBoost Classifier
Revenge Trading Detector	Post-loss aggressive trading	Custom ML Pipeline
Loss Aversion Detector	Holding losers / cutting winners	Pattern Analysis
Journal Bias Detector	NLP analysis of trade journals	RoBERTa Multi-Label

Five Psychological Biases Tracked

FOMO — Chasing trades out of fear of missing out
Loss Aversion — Cutting winners early, holding losers too long
Revenge Trading — Trading emotionally after losses to "win it back"
Overtrading — Taking excessive low-quality trades
Gambler's Fallacy — Believing in false patterns and "due" wins

🤖 AI Trading Coach (Expert Agent)

An intelligent assistant powered by LangGraph + Backboard AI that:

Analyzes your complete trading history with SQL queries
Provides personalized feedback on your trading patterns
Answers follow-up questions about specific trades
Offers actionable advice for improving discipline
Streams responses with real-time progress updates

📚 Educational Content

Behavior Summary — Quick-reference guide to the five trading biases
Pre-Trade Checklist — Questions to ask before every trade
Interactive Quiz — Test your bias recognition with real training examples
Daily Intentions — Mindset prompts for disciplined trading

🏗️ Architecture

CogniTrade/
├── src/                     # React Frontend (Vite + TypeScript)
│   ├── pages/               # Route pages
│   │   ├── Index.tsx        # Dashboard with portfolio overview
│   │   ├── TradePage.tsx    # Trading interface with charts
│   │   ├── SearchPage.tsx   # Stock search and watchlist
│   │   ├── AnalysisPage.tsx # ML bias analysis + AI chat
│   │   ├── SummaryPage.tsx  # Educational content + quiz
│   │   ├── SettingsPage.tsx # Account settings
│   │   └── LogsPage.tsx     # Trade history logs
│   ├── components/          # Reusable UI components
│   │   ├── AlpacaMarketChart.tsx  # Interactive stock charts
│   │   ├── TradePanel.tsx         # Buy/sell order form
│   │   ├── AgentChat.tsx          # AI expert chat interface
│   │   ├── LogUpload.tsx          # Trade log upload + analysis
│   │   ├── BehaviorAnalysis.tsx   # Bias score cards
│   │   └── ui/                    # Radix UI primitives
│   └── lib/
│       └── tradingStore.ts  # Zustand state management
│
├── backend/                 # Python Flask Backend
│   ├── app.py               # Flask API routes
│   ├── agent.py             # LangGraph AI agent with DuckDB
│   ├── models/              # ML model inference
│   │   ├── overtrading_model/
│   │   ├── revenge_trading_model/
│   │   └── loss_aversion_trading_model/
│   ├── journal_model_training_script/
│   │   └── train_bias_detector.py  # RoBERTa training
│   └── requirements.txt
│
└── public/
    └── cognitrade_logo2.png

🚀 Getting Started

Prerequisites

Node.js 18+ and npm
Python 3.10+
API Keys (see below)

1. Clone the Repository

git clone https://github.com/yourusername/CogniTrade.git
cd CogniTrade

2. Configure Environment Variables

Copy the example environment file and add your API keys:

cp .env.example .env

Edit .env with your credentials:

# Alpaca Markets (for real-time stock data)
VITE_ALPACA_API_KEY=your_alpaca_key
VITE_ALPACA_SECRET_KEY=your_alpaca_secret
VITE_ALPACA_BASE_URL=https://paper-api.alpaca.markets

# Backend API URL (default for local dev)
VITE_API_URL=http://localhost:5001

# Gradium (for voice transcription)
GRADIUM_API_KEY=your_gradium_key
GRADIUM_REGION=us

# Backboard AI (for the AI trading expert)
BACKBOARD_API_KEY=your_backboard_key

# OpenAI (backup for AI features)
OPENAI_API_KEY=your_openai_key

3. Install Frontend Dependencies

npm install

4. Install Backend Dependencies

cd backend
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt
cd ..

5. Start the Development Servers

Terminal 1 — Frontend:

npm run dev

Terminal 2 — Backend:

cd backend
source venv/bin/activate
python app.py

The app will be available at http://localhost:5173 or http://localhost:8080

📡 API Endpoints

Backend API (Flask)

Endpoint	Method	Description
`/`	GET	Health check
`/health`	GET	Service status
`/transcribe`	POST	Single-shot audio transcription
`/transcribe/stream`	WebSocket	Real-time streaming transcription
`/analyze_journal`	POST	Analyze journal text for biases (RoBERTa)
`/analyze_trades`	POST	Analyze trade CSV for all three biases
`/agent/analyze`	POST	Full AI analysis with ML + expert report (SSE)
`/agent/chat`	POST	Follow-up chat with the AI expert (SSE)

CSV Upload Format

For the /analyze_trades and /agent/analyze endpoints, upload a CSV with these columns:

Column	Type	Description
`timestamp`	datetime	Trade execution time
`asset`	string	Stock symbol (e.g., AAPL)
`side`	string	"buy" or "sell"
`quantity`	number	Number of shares
`entry_price`	number	Entry price per share
`exit_price`	number	Exit price per share
`profit_loss`	number	P&L for the trade
`balance`	number	Account balance after trade

🔧 Tech Stack

Frontend

Technology	Purpose
React 19	UI framework
TypeScript	Type safety
Vite	Build tool & dev server
TailwindCSS	Utility-first styling
Radix UI	Accessible component primitives
Zustand	State management
Motion	Animations
Recharts	Data visualization
Lightweight Charts	Stock candlestick charts
React Query	Server state management

Backend

Technology	Purpose
Flask	Web framework
Flask-CORS	Cross-origin requests
Flask-Sock	WebSocket support
Pandas	Data manipulation
scikit-learn	ML utilities
XGBoost	Overtrading model
PyTorch	Deep learning runtime
Transformers	RoBERTa bias detector
LangGraph	AI agent orchestration
Backboard SDK	AI thread management
DuckDB	In-memory SQL for trade analysis

🧪 Model Training

Overtrading Model

The overtrading detector uses XGBoost trained on windowed trading features:

Trade frequency per window
Average position size deviation
Time between trades
Win/loss streaks

Journal Bias Detector

The RoBERTa-based NLP model is trained on labeled trading journal entries:

cd backend/journal_model_training_script
python train_bias_detector.py

Training data is in backend/rational_training/train.json.

📸 Screenshots

Dashboard	Trading	Analysis
Portfolio overview with P&L	Real-time charts + order form	ML bias scores + AI chat

📄 License

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

🙏 Acknowledgments

Alpaca Markets for real-time market data
Backboard AI for AI agent infrastructure
Gradium for speech-to-text
Radix UI for accessible components
TradingView Lightweight Charts for stock charts

Trade smarter, not harder.

Made with ❤️ for disciplined traders everywhere.

Name		Name	Last commit message	Last commit date
Latest commit History 37 Commits
backend		backend
public		public
src		src
.env.example		.env.example
.gitignore		.gitignore
README.md		README.md
SETUP.md		SETUP.md
eslint.config.js		eslint.config.js
index.html		index.html
package-lock.json		package-lock.json
package.json		package.json
postcss.config.js		postcss.config.js
tailwind.config.js		tailwind.config.js
tsconfig.app.json		tsconfig.app.json
tsconfig.json		tsconfig.json
tsconfig.node.json		tsconfig.node.json
vite.config.ts		vite.config.ts

Folders and files

Latest commit

History

Repository files navigation

CogniTrade

🎯 Overview

Why CogniTrade?

✨ Features

📊 Paper Trading Simulator

🧠 AI Behavioral Analysis

Machine Learning Models

Five Psychological Biases Tracked

🤖 AI Trading Coach (Expert Agent)

📚 Educational Content

🏗️ Architecture

🚀 Getting Started

Prerequisites

1. Clone the Repository

2. Configure Environment Variables

3. Install Frontend Dependencies

4. Install Backend Dependencies

5. Start the Development Servers

📡 API Endpoints

Backend API (Flask)

CSV Upload Format

🔧 Tech Stack

Frontend

Backend

🧪 Model Training

Overtrading Model

Journal Bias Detector

📸 Screenshots

📄 License

🙏 Acknowledgments

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages