AI-Auto Trading Signal System Development
Budget: $750 – $1,500 SGD
Project Brief: AI-Based Auto Trading Signal System
Goal
Build an AI system that automatically generates buy/sell signals for trading instruments (Gold, BTC, and others) and adapts to live market conditions for intraday trading.
Core Features Needed
1. Data Ingestion
Live price feeds (via APIs like Binance, Yahoo Finance, Alpha Vantage)
OHLCV data (Open, High, Low, Close, Volume)
News/sentiment feed (optional but powerful)
2. AI Signal Engine
ML models (LSTM or XGBoost) trained on historical price data
Auto-generates Buy / Sell / Hold signals
No hardcoded strategy — model learns patterns from data itself
3. Adaptive Market Condition Detection
Automatically detects regime: Trending / Ranging / Volatile
Switches model behavior based on detected condition
Re-trains or fine-tunes on recent data periodically
4. Intraday Focus
Operates on 1min / 5min / 15min timeframes
Signals generated in real time during market hours
5. Output / Dashboard
Simple UI showing current signal, confidence score, market regime
Alert via Telegram or email when signal triggers
Tech Stack Suggestion
Python (core)
scikit-learn / TensorFlow / PyTorch (ML)
CCXT library (crypto exchange connectivity)
Streamlit (quick dashboard)
PostgreSQL (data storage)
Important Notes for Coder
No predefined strategy — system must learn from data
Must handle both crypto (24/7) and commodity markets (Gold)
Risk management module needed: stop-loss, position sizing suggestions
Backtesting module essential before going live
Goal
Build an AI system that automatically generates buy/sell signals for trading instruments (Gold, BTC, and others) and adapts to live market conditions for intraday trading.
Core Features Needed
1. Data Ingestion
Live price feeds (via APIs like Binance, Yahoo Finance, Alpha Vantage)
OHLCV data (Open, High, Low, Close, Volume)
News/sentiment feed (optional but powerful)
2. AI Signal Engine
ML models (LSTM or XGBoost) trained on historical price data
Auto-generates Buy / Sell / Hold signals
No hardcoded strategy — model learns patterns from data itself
3. Adaptive Market Condition Detection
Automatically detects regime: Trending / Ranging / Volatile
Switches model behavior based on detected condition
Re-trains or fine-tunes on recent data periodically
4. Intraday Focus
Operates on 1min / 5min / 15min timeframes
Signals generated in real time during market hours
5. Output / Dashboard
Simple UI showing current signal, confidence score, market regime
Alert via Telegram or email when signal triggers
Tech Stack Suggestion
Python (core)
scikit-learn / TensorFlow / PyTorch (ML)
CCXT library (crypto exchange connectivity)
Streamlit (quick dashboard)
PostgreSQL (data storage)
Important Notes for Coder
No predefined strategy — system must learn from data
Must handle both crypto (24/7) and commodity markets (Gold)
Risk management module needed: stop-loss, position sizing suggestions
Backtesting module essential before going live