How Algo Trading Works in Indian Markets (NSE & Beyond)
Understand how algo trading works in Indian markets: signals, broker APIs, order types, sessions, risk checks, and what happens between your strategy and the exchange.
6 August 2026 · 13 min read
The big picture in one paragraph
In Indian markets, algo trading works like this: your strategy evaluates live (or candle) data → a signal is generated → risk checks run → an order request goes to your broker’s API → the broker routes to the exchange → fills and statuses stream back → your platform updates positions and may manage exits (SL/target/trailing/time square-off).
You remain the account holder. The platform is software sitting between your rules and your broker — not a substitute broker.
Market sessions that algos must respect
Pre-open, regular continuous session, and closing / special sessions behave differently. Strategies that fire market orders in the wrong window get rejects or ugly fills.
Intraday algos usually need an explicit square-off time before the broker’s auto square-off — never assume “the platform will remember.”
Expiry days and event windows (RBI, Budget, results) change volatility and slippage. Good algos either reduce size or pause.
Signal generation layer
Indicator rules: RSI, MACD, SuperTrend, moving averages, Bollinger, VWAP, ADX, and combinations.
Price action rules: breakouts, opening range, candle patterns encoded as conditions.
External signals: TradingView alerts or other charting tools via webhook bridges into platforms like AlgoCrab.
Key design choice: evaluate on candle close (more stable) vs every tick (faster, noisier).
Risk and order management layer
Position sizing: fixed quantity, risk-per-trade %, or volatility-based sizing.
Stops and targets: absolute points, percentage, ATR multiples, or trailing logic.
Portfolio guards: max open positions, max daily loss, max trades per day.
Without this layer, “how algo trading works” becomes “how accounts blow up automatically.”
Broker API and authentication
Brokers expose REST/WebSocket APIs after you enable API access and complete security steps (API key/secret, TOTP, OAuth — varies by broker).
Some brokers require static IP whitelisting for algo / API orders. Plan for this early.
Tokens expire. Production setups refresh sessions deliberately — laptop sleep is not an infrastructure strategy.
What the exchange actually sees
The exchange sees orders from your broker member ID — not “AlgoCrab” as a broker. Your order must pass exchange validations: lot size, freeze quantity, margin, circuit limits, product type (MIS/NRML/CNC), and timing.
Rejects are normal during learning. Read the reject reason; fix the rule; do not spam retries.
Monitoring and failure modes
Network blips, auth expiry, rate limits, and partial fills all happen in live markets.
A robust setup shows clear logs, alerts you on disconnects, and lets you flatten manually.
Paper trading hides some of these — always do a tiny live rehearsal.
Where AlgoCrab fits in the pipeline
AlgoCrab hosts strategy templates and AlgoCode, consumes live data via connected brokers, applies your configured risk rules, and sends orders through those brokers.
Backtesting lets you rehearse the signal logic before the API layer matters.
Bridge tools let external charting platforms feed the same execution pipeline.
FAQs
Q: Is algo trading only for HFT? A: No. Most retail algos are rule-based intraday or swing systems, not microsecond HFT.
Q: Do I need co-location? A: Retail traders typically do not. Focus on correct rules and reliable connectivity first.
Q: Can one algo trade multiple brokers? A: On multi-broker platforms like AlgoCrab, yes — within each broker’s permissions.
Q: Why did my algo not enter? A: Often filters, time windows, margin, or data gaps — check logs before rewriting indicators.
Learn by doing
Connect a broker on AlgoCrab, run a single-condition strategy with hard stops, and watch the full signal → order → fill loop. That experience teaches more than any abstract diagram.
Disclaimer: Educational only — not investment advice.
