AlgoTrader

A high-frequency multi-strategy algorithmic trading system built on Alpaca Markets. Runs a fully event-driven WebSocket architecture for live/paper trading and a bar-by-bar simulation engine for backtesting.

Strategies

Four independent intraday strategy layers run concurrently on a universe of large-cap equities:

Layer	Strategy	Description
L1	VWAP Mean Reversion	Fades VWAP deviations with RSI confirmation
L2	Opening Range Breakout	Trades breakouts from the 09:30–10:00 opening range
L3	RSI Reversal Scalp	Short-duration oversold bounces on 7-period RSI
L4	Volume Surge Momentum	Enters on abnormal volume with strong bar positioning

A market regime filter (SPY 200-SMA + VIXY proxy) scales position sizes and halts trading in high-volatility environments.

Architecture

WebSocket bars → BarStore → BarDispatcher → Strategy Layers → OrderManager → Alpaca
                                                            ↘ RiskManager (PDT, circuit breaker, stop loss)
                                                            ↘ NotificationQueue → Discord

• StreamEngine — async WebSocket consumer, routes 1-min bars to all strategy layers
• BarDispatcher — fan-out coordinator, enforces max positions and layer arbitration
• APScheduler — pre-market warmup (08:00), market open reset (09:30), ORB finalize (10:00), EOD close (15:30), daily report (16:30)
• BacktestEngine — bar-by-bar historical replay with slippage modeling, equity curve, and trade log output

Risk Management

• Per-trade dollar risk cap ($150) with ATR-based stop placement
• Max daily loss ($1,500) and drawdown ($5,000) circuit breakers
• PDT guard — limits day trades to 3 per week
• Regime-based position size scaling (50% reduction above VIXY threshold, full halt above hard threshold)
• Trailing stop activation at 1.0R, trailing at 40% of move
• EOD forced close at 15:30 ET

Quick Start

pip install -r requirements.txt
cp .env.example .env  # add ALPACA_API_KEY, ALPACA_SECRET_KEY, DISCORD_WEBHOOK_URL

# Paper trading (default)
python main.py --paper

# Dry run — full signal generation, no real orders
python main.py --dry-run

# Historical backtest (365 days)
python main.py --test

# Backtest from a specific date
python main.py --test --start 2024-01-01

# Generate dashboard report only
python main.py --report-only

Configuration

All parameters are in config.yml — strategy thresholds, risk limits, scheduler times, and universe filters. No code changes needed to tune behavior.

Key defaults:

• Starting capital: $100,000
• Max concurrent positions: 15
• Universe: large-cap equities (price ≥ $15, ADV ≥ $50M)
• Target trade cadence: ≥300 trades/year

Stack

• Python 3.11+ — asyncio, APScheduler
• Alpaca Markets — brokerage API + WebSocket data feed
• Supabase / SQLite — trade and portfolio state persistence
• Discord — real-time trade alerts and daily summaries
• Docker — docker-compose up for containerized deployment