monteCore — Quant Trading Backtesting + Monte Carlo Engine

Overview

A full-stack quantitative trading platform that backtests moving average strategies on historical equity data, evaluates risk-adjusted performance, and runs progressive Monte Carlo simulations to model uncertainty.

Stack: Python (FastAPI) · Next.js 16 · TypeScript · Recharts · yfinance · NumPy/Pandas

---

Features

1. Backtesting Engine

• Load historical OHLCV data via yfinance (auto-fetched and cached to CSV)
• Moving average crossover strategy with configurable short/long windows
• Simulates positions, daily returns, and portfolio equity curve from $10,000
• Transaction costs: commission + slippage deducted on every position change

2. Performance Metrics

• Total return
• Sharpe ratio (annualized, 252 trading days)
• Maximum drawdown
• Full equity curve + daily returns

3. Monte Carlo Simulation

• Vectorized bootstrap resampling of historical strategy returns
• Progressive simulation: 100 → 500 → 1,000 → 5,000 paths
• VaR 95% and VaR 99% — worst-case portfolio value at each confidence level
• Scenario cards: Optimistic (90th pct), Median (50th pct), Conservative (10th pct)

4. Web Dashboard (Next.js + FastAPI)

• Voltrex-inspired dark UI with purple radial gradient background
• Navbar with live portfolio value and ticker
• Stats bar: Portfolio Value · Total Return · Sharpe · Max Drawdown · VaR 95% · VaR 99% · Simulations Run
• Equity Curve / Returns chart — tabbed, orange line chart
• Monte Carlo chart — two tabs:
  • Paths: 150 faded simulation lines + bold median (pure SVG)
  • Distribution: histogram of final returns (teal = positive, red = negative)
• Right panel: Deposit-style controls for ticker, strategy, windows, commission, slippage, simulation count
• Progressive simulation UI: stage dots pulse 100 → 500k → 1k → 5k as paths fill in live

---

Project Structure

quant-project/
│
├── backend/                       # Python — deploy to Railway / Render / Fly.io
│   ├── main.py                    # FastAPI server — /backtest and /simulate
│   ├── main_standalone.py         # Standalone CLI with matplotlib output
│   ├── requirements.txt           # pip dependencies
│   ├── Procfile                   # Start command for Railway/Heroku
│   ├── .env.example               # ALLOWED_ORIGINS env var template
│   ├── data/
│   │   └── data_loader.py         # yfinance fetch + CSV cache (auto-downloads)
│   ├── strategies/
│   │   └── moving_average.py      # SMA crossover signal generation
│   ├── engine/
│   │   └── backtester.py          # Position simulation + commission/slippage
│   ├── metrics/
│   │   └── performance.py         # Sharpe, drawdown, total return
│   └── monte_carlo/
│       └── simulator.py           # Vectorized bootstrap MC (numpy matrix ops)
│
├── frontend/                      # Next.js — deploy to Vercel
│   ├── .env.local.example         # NEXT_PUBLIC_API_URL template
│   ├── app/
│   │   ├── page.tsx               # Main dashboard — progressive simulation logic
│   │   ├── layout.tsx             # Root layout with Inter font
│   │   └── globals.css            # Full dark theme (Voltrex-inspired)
│   ├── components/
│   │   ├── Sidebar.tsx            # Navbar
│   │   ├── MetricsBar.tsx         # Stats bar (7 metrics)
│   │   ├── ControlPanel.tsx       # Right panel (run/reset + all params)
│   │   ├── ChartCard.tsx          # Equity/returns line chart
│   │   ├── MonteCarloChart.tsx    # Multi-path SVG chart + histogram
│   │   └── MonteCarloPanel.tsx    # Scenario cards
│   └── lib/
│       └── api.ts                 # runBacktest() + runSimulate() with AbortSignal
│
├── .gitignore
└── README.md

---

API Endpoints

POST /backtest

Full backtest + initial 100-sim MC run.

Request:
{
  "symbol": "AAPL",
  "short_window": 20,
  "long_window": 50,
  "n_simulations": 100,
  "commission": 0.001,
  "slippage": 0.0005
}

Response:
{
  "sharpe": 0.84,
  "drawdown": -0.153,
  "total_return": 0.312,
  "equity_curve": [...],
  "dates": [...],
  "var_95": 9240.50,
  "var_99": 8870.10,
  "monte_carlo": [...],
  "paths": [[...], ...],
  "final_values": [...],
  "simulation_count": 100
}

POST /simulate

MC-only progressive update — same request shape, returns only MC fields (no equity_curve/dates). Used for stages 500, 1000, 5000.

---

How It Works

Step 1 — Load Data

yfinance fetches OHLCV data. CSVs are cached in data/ and auto-downloaded on first request for any ticker.

Step 2 — Generate Signals

SMA_short > SMA_long  →  BUY  (position = 1)
SMA_short ≤ SMA_long  →  SELL (position = 0)

Step 3 — Backtest

Simulate positions with lagged signals. Deduct commission + slippage on each trade entry/exit. Compute equity curve from $10,000 initial capital.

Step 4 — Evaluate

Compute Sharpe ratio, max drawdown, total return.

Step 5 — Monte Carlo

Bootstrap-resample daily strategy returns into 5,000 simulated 252-day paths (vectorized numpy). Compute VaR, percentile scenarios, and final value distribution.

Step 6 — Progressive UI

On button click: stage 1 runs /backtest (100 sims, instant results), then stages 2–4 call /simulate at 500/1000/5000 sims — the chart fills in live as each stage completes.

---

Local Development

# Backend
cd backend
pip install -r requirements.txt
uvicorn main:app --reload
# → http://localhost:8000

# Frontend (separate terminal)
cd frontend
npm install
cp .env.local.example .env.local   # uses http://localhost:8000 by default
npm run dev
# → http://localhost:3000

Deployment

Backend → Railway / Render / Fly.io

1. Point the platform at the backend/ directory
2. Set env var: ALLOWED_ORIGINS=https://your-app.vercel.app
3. Start command is already in Procfile: uvicorn main:app --host 0.0.0.0 --port $PORT

Frontend → Vercel

1. Set root directory to frontend/
2. Set env var: NEXT_PUBLIC_API_URL=https://your-backend.railway.app
3. Deploy — Vercel auto-detects Next.js

---

Supported Tickers

SPY · AAPL · MSFT · NVDA · TSLA — any ticker supported by yfinance can be added.

---

Key Concepts

Sharpe Ratio — annualized return per unit of risk. Higher = better risk-adjusted performance.

Max Drawdown — largest peak-to-trough loss. Measures worst-case historical loss.

Value at Risk (VaR) — the portfolio value not expected to be breached with 95% / 99% confidence, derived from the Monte Carlo final value distribution.

Bootstrap Monte Carlo — resamples historical daily returns (with replacement) to simulate thousands of possible future equity paths without assuming a return distribution.

Transaction Costs — commission (e.g. 0.1%) + slippage (e.g. 0.05%) are deducted on every position change, giving a more realistic backtest.

---

Future Improvements

• Multiple strategies (RSI, Bollinger Bands, mean reversion)
• Portfolio optimization (multi-asset allocation)
• Real-time paper trading mode
• Walk-forward / out-of-sample validation

---

Notes

This project is for educational purposes only and does not constitute financial advice.