A comprehensive web dashboard for visualizing OHLCV (Open, High, Low, Close, Volume) stock prediction data with Monte Carlo analysis powered by transformer models.

GitHub Pages: https://jedarden.github.io/transformers-predictions

• Instant search: Type any ticker symbol for immediate results
• Auto-complete: Results appear as you type (300ms debounce)
• Smart validation: Shows "[Ticker] not found" for invalid symbols
• No button required: Search happens automatically

• 🟢 Green candlesticks: Bullish days (close ≥ open)
• 🔴 Red candlesticks: Bearish days (close < open)
• High-Low wicks: Complete daily trading range visualization
• Historical context: 50 days of market data

• Monte Carlo analysis: 10 simulation paths with uncertainty ranges
• Confidence bands: Statistical prediction intervals (10th-90th, 25th-75th percentiles)
• Actual vs predicted: Real market data overlay for validation
• Interactive controls: Toggle different chart elements

• Accuracy metrics: MAPE, RMSE, MAE for all OHLCV components
• Performance scores: Overall model rating (0-100 scale)
• Quality indicators: Excellent, Good, Fair, Poor classifications
• Technical analysis: Support/resistance levels, volatility patterns

Try these examples:

• AAPL - Apple Inc. (Excellent prediction quality - 85.1 score)
• TSLA - Tesla Inc. (High-volatility analysis)
• MSFT - Microsoft Corporation (Large-cap stability)
• GOOGL - Alphabet Inc. (Tech sector analysis)
• ZEPP - Zepp Health Corporation (Fair prediction quality - 41.7 score)

• Pure JavaScript: No frameworks, fast loading
• Chart.js: Professional financial chart rendering
• CSS3: Modern responsive design with gradients
• HTML5: Semantic structure optimized for accessibility

• Static hosting: No server required, pure client-side
• Direct JSON loading: Fetches prediction files directly
• CDN libraries: Chart.js and date adapters from CDN
• Responsive design: Mobile-friendly layout

Each ticker includes:

• 50 days historical OHLCV data
• 10 Monte Carlo simulation paths (5 days each)
• Statistical summaries with percentiles and confidence intervals
• Performance metrics with accuracy scores
• Technical indicators (support/resistance, volatility)

• Type-to-search: Results appear as you type ticker symbols
• Instant feedback: Immediate validation and error handling
• Smooth animations: Loading states and transitions
• Touch-friendly: Optimized for mobile and tablet use

• Zoom & Pan: Full chart navigation capabilities
• Rich tooltips: Complete OHLCV data on hover
• Toggle controls: Show/hide prediction bands, simulations, actual data
• Professional styling: Financial-grade visualization

• Mobile-first: Designed for all screen sizes
• Adaptive layout: Cards and grids adjust to screen width
• Touch interactions: Mobile-optimized controls
• Fast loading: Optimized assets and minimal dependencies

# Clone the repository
git clone https://github.com/jedarden/transformers-predictions.git
cd transformers-predictions

# Serve locally (optional - can open index.html directly)
python -m http.server 8000
# or
npx serve .

# Open browser
open http://localhost:8000

To add additional ticker data:

1. Copy JSON files to the data/ directory
2. Ensure files follow the naming convention: {TICKER}_ohlcv_prediction.json
3. Files will be automatically available for search

The dashboard expects JSON files with this structure:

{
  "ticker_info": {
    "symbol": "AAPL",
    "model_type": "complete-ohlcv-monte-carlo",
    "chart_compatibility": "candlestick_ready"
  },
  "chart_data": {
    "historical_candlesticks": [...],
    "actual_candlesticks": [...],
    "monte_carlo_simulations": [...],
    "predicted_ohlcv_summary": {...}
  },
  "data": {
    "prediction_metrics": {...},
    "ohlcv_analysis": {...}
  }
}

1. Push changes to the main branch
2. GitHub Actions automatically deploys to GitHub Pages
3. Live site updates within 2-3 minutes

1. Fork or clone this repository
2. Enable GitHub Pages in repository settings
3. Select "GitHub Actions" as the source
4. Push to main branch to trigger deployment

• Fast loading: Minimal dependencies, optimized assets
• Instant search: Real-time ticker lookup with debouncing
• Smooth interactions: Hardware-accelerated animations
• Mobile optimized: Touch-friendly interface design

• Candlestick patterns: Traditional financial chart visualization
• Volume analysis: Trading volume patterns and statistics
• Volatility indicators: Market volatility and risk metrics
• Technical levels: Support and resistance price levels
• Prediction accuracy: Model performance validation with actual data

• Live Dashboard: https://jedarden.github.io/transformers-predictions
• Source Code: https://github.com/jedarden/transformers-predictions
• Issues: https://github.com/jedarden/transformers-predictions/issues

Built with ❤️ for financial data visualization and stock market prediction analysis.