AlgoTradingAI
Start Free
Docs/Indicators & Models

Indicators & Models

AlgoTradingAI uses a multi-layer signal generation pipeline combining technical indicators, pattern detection, proprietary AI models, and time-series forecasting.

Technical Indicators

The following indicators are computed on each candle update for watched symbols:

  • SMA (Simple Moving Average): Configurable periods (default 20, 50). Used for trend direction and crossover signals.
  • EMA (Exponential Moving Average): Faster-reacting trend filter. Used in conjunction with SMA for signal confirmation.
  • RSI (Relative Strength Index): 14-period momentum oscillator. Overbought (>70) and oversold (<30) zones used for reversal signals.
  • ATR (Average True Range): 14-period volatility measure. Used for dynamic stop-loss calculation (ATR-multiple mode) and position sizing.
  • VWAP (Volume Weighted Average Price): Intraday benchmark. Signals consider price position relative to VWAP.
  • Bollinger Bands: 20-period, 2 standard deviations. Used for volatility breakout and mean-reversion signals.
  • MACD: 12/26/9 configuration. Histogram direction and signal line crossovers used for momentum confirmation.
  • OBV (On-Balance Volume): Cumulative volume flow. Divergences between OBV and price can signal trend weakness.

Pattern Detection

The strategy layer detects the following candlestick and price patterns:

  • Breakout: Price closing above previous day high (bullish) or below previous day low (bearish) with volume confirmation.
  • Hammer / Inverted Hammer: Single-candle reversal patterns at support/resistance zones.
  • Engulfing: Two-candle reversal pattern where the second candle fully engulfs the first.
  • Inside Bar: Consolidation pattern used for breakout anticipation.

11-Module Consensus System

The AI Advisor uses an 11-module consensus system where each analysis module independently scores a trade candidate. The modules cover:

  • Trend & Direction: Moving average crossovers, ADX strength, Supertrend direction
  • Momentum & Mean-Reversion: RSI zones, stochastic crossovers, CCI, Williams %R
  • Volatility & Range: Bollinger Bands, Keltner Channels, ATR context
  • Volume & Flow: OBV trend, Volume Delta, Chaikin Money Flow, volume profile proximity
  • Pattern Recognition: Candlestick and chart pattern detection
  • Market Structure: Wyckoff accumulation/distribution phase and event detection
  • Price Forecast: Multi-model fused forecast direction across multiple horizons
  • Strategy Signals: Rule-based strategy signals with optional ML approval gating

Each module produces a score from -1.0 (strong sell) to +1.0 (strong buy). The consensus is a weighted average with configurable weights per module.

Proprietary ML Quality Filter

Every candidate trade passes through a proprietary ML scoring engine that predicts trade quality:

  • Features: Trend (SMA/EMA slopes), momentum (RSI, MACD), volatility (ATR, BB width), volume/OI context, pattern metadata, time-of-day features.
  • Training: Per-symbol model trained via the Profile Tuning page. Uses walk-forward validation.
  • Threshold: Default 0.6 (configurable). Only candidates scoring above threshold are surfaced.
  • Output: Probability score (0.0-1.0) and feature importance rankings.

Multi-Horizon Price Forecasting

A proprietary multi-model forecast engine is used for price direction forecasting:

  • 8 Horizons: 2 hours, End of Day, 1 day, 2 days, DTE-Short, 1 week, 1 month, DTE-Long.
  • Fused Forecast: Multiple proprietary AI models are fused to combine statistical pattern recognition with deep learning time-series modeling.
  • Confidence Bands: Each horizon includes prediction intervals (10th, 50th, 90th percentiles).
  • Input: Historical OHLCV candles (minimum 60 data points).
  • Dual Forecast: For derivatives, both the option and its underlying are forecasted independently.

Morning Scan Pipeline

A scheduled job runs each morning before market open:

  • Fetches overnight data and pre-market indicators for all watched symbols.
  • Runs the full signal generation pipeline (patterns + indicators + ML filter).
  • Generates candidate trades ready for review when the market opens.
  • Results available on the Trade Signals page with a "Morning Scan" tag.

News Sentiment Analysis

The AI Advisor pipeline includes NLP sentiment analysis tuned for Indian financial markets:

  • Engine: Proprietary NLP sentiment engine fine-tuned for Indian financial text.
  • Input: News articles retrieved via Google News RSS search.
  • Output: Per-article sentiment (positive, negative, neutral) with confidence score.
  • Integration: Sentiment scores are included in the AI advisory prompt for context-aware recommendations.

Next Steps