Help & Support

Our Trend Prediction system uses a multi-ensemble machine learning approach combining:

• LSTM Networks (Long Short-Term Memory) for time series analysis
• Random Forest for pattern recognition
• Gradient Boosting for enhanced accuracy
• Sentiment Analysis from news and social media

Each model is trained on 5+ years of historical data and weighted based on past performance. The ensemble approach reduces overfitting and improves reliability.

Monte Carlo simulation is a statistical technique that runs thousands of random scenarios to predict portfolio outcomes.

How we use it:

• Simulates 50,000+ possible market scenarios
• Uses historical volatility and returns
• Accounts for correlations between assets
• Provides probability distributions of returns

This helps you understand risk ranges rather than single predictions.

Alphy AI is powered by Google Gemini Pro with a custom financial layer:

• NLP Engine: Understands context and intent
• Financial Knowledge Base: Pre-trained on 1000+ financial documents
• Real-time Data Integration: Connects to live market APIs
• Chart Generation: Automatically creates visualizations

It can analyze stocks, explain concepts, and provide actionable insights in conversational language.

We aggregate data from institutional-grade providers:

• Market Data: Twelve Data, FinnHub
• Insider Trading: SEC EDGAR (Form 4 filings)
• M&A Data: SEC filings (S-4, 8-K, DEFM14A)
• News & Sentiment: NewsAPI - Bloomberg, Financial Times, CNBC...
• Economic Data: FRED (Federal Reserve)

All data is validated, cleaned, and updated in real-time.

The Insider Flow Tracker monitors SEC Form 4 filings in real-time:

• Detects C-suite transactions (CEO, CFO, directors)
• Calculates Cluster Score (unusual buying/selling activity)
• Tracks transaction value and timing
• Flags anomalies (e.g., 10x normal volume)

Why it matters: Insiders often buy before positive news (and sell before negative news).

Our M&A Predictor identifies companies likely to be acquired:

• Analyzes SEC filings: S-4, 8-K, proxy statements
• Fundamental filters: Undervalued stocks, high cash, low debt
• Industry trends: Sector consolidation patterns
• Ownership changes: Activist investors, stake increases

It assigns a Probability Score (0-100%) for acquisition likelihood.

Confidence scores reflect model certainty based on:

• Historical Accuracy: Backtested performance (last 3 years)
• Data Quality: Completeness and freshness
• Volatility: Lower volatility = higher confidence
• Model Agreement: Consensus across ensemble models

Example: 85% confidence means the model was correct 85% of the time in similar past scenarios.

• Firebase Authentication: Industry-standard OAuth 2.0
• End-to-End Encryption: All data in transit (TLS 1.3)
• Firestore Security Rules: User-isolated data access
• No Third-Party Sharing: Your data is never sold
• GDPR Compliant: Right to export/delete data