DEV Community

Sanket Jagtap
Sanket Jagtap

Posted on • Originally published at sanket-jagtap.in on

Inside AlgoETF: An Algorithmic ETF Trading Platform

#angular #node #trading #websockets

AlgoETF is an algorithmic trading platform for Indian ETFs. It runs rule-based strategies against live market data, places orders through connected brokers, and lets you research everything with a built-in backtester before risking real money.

The core strategy

At its heart is a simple, explainable idea: buy the dip, sell the rise , gated by momentum. The engine watches each ETF and acts when price moves a configured percentage, but only when the RSI(14) filter agrees — so it avoids buying into a falling knife or selling into strength.

  • Buy on dip and sell on rise thresholds, per ETF.
  • RSI oversold / overbought gate to filter bad entries.
  • Cycle targets so each position has a defined exit.

Real-time architecture

The frontend is Angular; the backend is Node.js + Express. Live prices and order updates stream over WebSockets , Redis handles pub/sub and hot state, and MySQL is the system of record for strategies, cycles and orders. Broker APIs sit behind an adapter layer so new brokers can be added without touching the strategy code.

The backtester

You can replay historical 1-minute data through the exact same strategy logic — research only, no orders placed. The best part is the A/B improvement rules : toggle a stop-loss, cap averaging, an EMA trend filter, a volume filter or a trailing exit, and compare the result against the base strategy.

Parameters: stock, date range, buy dip %, sell rise %
Gate: RSI period, oversold <, overbought >
Risk: qty/order, start capital, brokerage %, slippage %
A/B rules: stop loss, cap averaging, EMA filter, trailing exit
Enter fullscreen mode Exit fullscreen mode

Safety first

Trading code is unforgiving, so the platform separates live , paper and backtest modes explicitly. Paper trading mirrors live behaviour with no money at stake, and every order is logged with a clear success / pending / failed status.

Lessons learned

  • Make strategies boring and explainable — you have to trust them with money.
  • Backtesting and live execution must share the same code path, or your results lie.
  • An audit trail of every order is not optional in fintech.

Top comments (0)