Fully automated Multi time frame Algo Trading System
Budget: $3,000 – $5,000 USD
Build a 100% fully automated algorithmic trading system. This system will scan for patterns, execute intraday trades across multiple symbols and brokers, and manage risk with minimal human intervention.
The system must include the following fully modular components. Each module will be built for clean unit testing, CI/CD integration, and alignment with the full specification. Each will be implemented and validated against the specifications outlined in two separate documents: the full strategy specification and the flowchart-based architecture breakdown. The project will follow a 9-phase modular structure that includes data ingestion, pattern detection, signal generation, optimization, execution, capital allocation, monitoring, and broker failover support.
Project Scope (9 Modules)
1. Pattern Screener
Detect patterns like 1st Green Day, Parabolic Short, 2nd Red Day, etc.
Use HL, volume, gap days, and resistance filters
Segment by session (premarket, regular, after-hours), float, and market cap
Time-based rules and resistance distance logic per pattern
2. Multi-Timeframe Strategy Engine
Multi-timeframe signals (1m, 4m, 10m, 30m, 2h)
Signal types: Entry, Stoploss, Exit, Reversal, Partial Profit
Indicators: Stochastic, PSAR, double top/bottom, cross up/down
Signal confluence and partial candle logic included
3. Data Infrastructure
Pull 10+ years of 1m and daily OHLCV data ( broker API)
Tag each bar: HL%, volume spike, abnormal behavior, resistance, session
Store in DuckDB for backtest and TimescaleDB for live trading
Handle timezone, DST, and NaNs
4. Backtesting & Optimization
Backtest across thousands of tickers with realistic fill models
Hyperparameter tuning: walk-forward, nested CV, Optuna, sensitivity analysis
Signal evaluation: F1-score, MFE, AE, stoploss precision
Monte Carlo drawdown simulation and parameter pruning
5. Execution & Capital Allocation
Live broker integration:
Borrow short logic with fee filter, retry queue, top HL% candidate scanning
Capital allocation per pattern %, per stock, with dynamic reallocation
Risk controls: max trade loss, daily drawdown, spread filter, fill enforcement
6. Live Infrastructure (VPS + Sync)
Primary + Backup VPS with real-time sync and automated failover
Daily snapshots, config replication, heartbeat monitoring
Transition from backtest (DuckDB) to live (TimescaleDB) pipeline
7. Monitoring & UI
Web dashboard for strategy control, logs, optimization status
Alerts: missed signal, execution error, halt, max loss breach
2FA-secured admin-only panel
8. Security & Logging
Encrypted secrets (Vault, KMS), token rotation
IP whitelisting, GitHub-protected deployment
Full audit logs for signals, overrides, trade executions
9. CI/CD & Testing
GitHub Actions CI/CD with pytest coverage
Cron jobs for backups, signal health, version syncing
Sandbox support and full rollback capability
The system must include the following fully modular components. Each module will be built for clean unit testing, CI/CD integration, and alignment with the full specification. Each will be implemented and validated against the specifications outlined in two separate documents: the full strategy specification and the flowchart-based architecture breakdown. The project will follow a 9-phase modular structure that includes data ingestion, pattern detection, signal generation, optimization, execution, capital allocation, monitoring, and broker failover support.
Project Scope (9 Modules)
1. Pattern Screener
Detect patterns like 1st Green Day, Parabolic Short, 2nd Red Day, etc.
Use HL, volume, gap days, and resistance filters
Segment by session (premarket, regular, after-hours), float, and market cap
Time-based rules and resistance distance logic per pattern
2. Multi-Timeframe Strategy Engine
Multi-timeframe signals (1m, 4m, 10m, 30m, 2h)
Signal types: Entry, Stoploss, Exit, Reversal, Partial Profit
Indicators: Stochastic, PSAR, double top/bottom, cross up/down
Signal confluence and partial candle logic included
3. Data Infrastructure
Pull 10+ years of 1m and daily OHLCV data ( broker API)
Tag each bar: HL%, volume spike, abnormal behavior, resistance, session
Store in DuckDB for backtest and TimescaleDB for live trading
Handle timezone, DST, and NaNs
4. Backtesting & Optimization
Backtest across thousands of tickers with realistic fill models
Hyperparameter tuning: walk-forward, nested CV, Optuna, sensitivity analysis
Signal evaluation: F1-score, MFE, AE, stoploss precision
Monte Carlo drawdown simulation and parameter pruning
5. Execution & Capital Allocation
Live broker integration:
Borrow short logic with fee filter, retry queue, top HL% candidate scanning
Capital allocation per pattern %, per stock, with dynamic reallocation
Risk controls: max trade loss, daily drawdown, spread filter, fill enforcement
6. Live Infrastructure (VPS + Sync)
Primary + Backup VPS with real-time sync and automated failover
Daily snapshots, config replication, heartbeat monitoring
Transition from backtest (DuckDB) to live (TimescaleDB) pipeline
7. Monitoring & UI
Web dashboard for strategy control, logs, optimization status
Alerts: missed signal, execution error, halt, max loss breach
2FA-secured admin-only panel
8. Security & Logging
Encrypted secrets (Vault, KMS), token rotation
IP whitelisting, GitHub-protected deployment
Full audit logs for signals, overrides, trade executions
9. CI/CD & Testing
GitHub Actions CI/CD with pytest coverage
Cron jobs for backups, signal health, version syncing
Sandbox support and full rollback capability