๐Ÿ‡บ๐Ÿ‡ธ US ยท ๐Ÿ‡ฎ๐Ÿ‡ณ India ยท 90+ stocks ยท Live data

Analyze Any Stock
With ML in Seconds

Real market data ยท Technical indicators ยท Risk quantification ยท Machine learning anomaly detection ยท NLP news sentiment

Indian stocks need .NS suffix โ€” e.g. UJJIVANSFB.NS ยท ZOMATO.NS ยท IRCTC.NS

What it analyzes

๐Ÿ“ˆ

Technical Indicators

SMA 20/50/200RSI 14Volatility

Moving averages show trend direction. SMA crossovers signal momentum shifts. RSI (0โ€“100) flags overbought (>70) and oversold (<30) conditions before price reverses.

๐ŸŽฏ

Risk & Portfolio Metrics

Sharpe RatioVaR 95/99%Max Drawdown

Quantify risk before you invest. Sharpe ratio measures return per unit of risk. VaR tells you the worst-case daily loss at a given confidence level.

๐Ÿค–

Anomaly Detection

Isolation Forest5% contamination4 features

Machine learning flags unusual trading days. The model scores each day on price return, rolling volatility, volume change, and price range โ€” no labelled data needed.

๐Ÿ“ฐ

News Sentiment (NLP)

VADER NLPโˆ’1 to +1 scoreNo API key

Every recent headline is scored as Bullish, Bearish, or Neutral using VADER โ€” a rule-based NLP model tuned for financial text. Runs entirely on the server.

The ML / Data Science Stack

Isolation Forest

Anomaly Detection ยท scikit-learn

Builds 100 random decision trees and measures how quickly each data point gets isolated. Points that reach leaf nodes in fewer splits are statistical outliers. Uses 4 features per trading day: daily return %, 5-day rolling volatility, volume change %, and highโˆ’low price range. Contamination = 5% means roughly 1 in 20 days is flagged.

Monte Carlo Simulation

Risk Analysis ยท NumPy

Runs 200 independent 30-day price simulations. Each day draws a random return from a normal distribution N(ฮผ, ฯƒ) fitted to historical data. The resulting paths show a probabilistic range of futures โ€” from worst-case (P5) to best-case (P95). The median path (P50) is the most likely outcome based on past behavior.

VADER Sentiment

NLP ยท vaderSentiment

Rule-based NLP with a hand-crafted lexicon of 7,500+ words, each with a sentiment score. Understands CAPS emphasis ("CRASH" scores worse than "crash"), punctuation amplification (!!!), and negation ("not great" flips polarity). Compound score from โˆ’1.0 to +1.0. No GPU, no model download, runs in microseconds per headline.