The six stages
- Hypothesis. One sentence: who is on the other side of your trade, and why do they lose? “Short covering rallies fade by the close” is testable. “Buy when RSI is low” is not a hypothesis.
- Data. Licensed, clean, one level finer than your decisions. See how much history you really need.
- Backtest. Include every cost: brokerage, exchange fees, taxes and slippage. Most edges disappear here, and that is the point.
- Paper trade. Run live, with live data, without money. This catches the timing and data problems a backtest cannot.
- Small live. The smallest size your broker allows, for long enough to see a bad week.
- Scale and monitor. Increase size only while live results stay inside the backtest’s range.
Stocks vs crypto vs forex
| Indian stocks & F&O | Crypto | Forex | |
|---|---|---|---|
| Session | Fixed hours, closed weekends | 24×7 | 24×5 |
| Open interest | Yes, on F&O | On derivatives venues | Not centrally reported |
| Order book | Exchange depth | Per-exchange, fragmented | Dealer quotes, fragmented |
| Gap risk | Overnight and weekend | Low (always open) | Weekend gaps |
| Main data concern | Licensing, corporate actions | Venue quality, wash trading | No single “true” price |
The biggest practical difference is time. A strategy that works in a 6¼-hour session has to be rethought for a market that never closes: when do you size down, and who is watching at 3 a.m.?
Why forex is the deepest pool
At around $7.5 trillion a day in the 2022 BIS survey, foreign exchange is the largest market in the world by turnover. Depth is excellent in the major pairs — but because trading is spread across dealers rather than one exchange, there is no single consolidated tape. Order-flow strategies that depend on seeing every trade are much harder in spot forex than on an exchange.
Risk rules before code
- Maximum loss per day that switches the system off, written before the first line of strategy code.
- Position limits per symbol and in total.
- A kill switch that closes everything, which you have tested.
- Data-staleness checks. If prices stop updating, the system must stop trading — not keep trading on the last known price.
Going live safely
Discretionary traders can take a middle path: let software do the reading and keep the decision human. The Mirra scanners do exactly this — the Order Flow Scanner runs the same engine across 100+ symbols and alerts you when something qualifies, while you place the trade.
If you are building fully automated systems, start with a reliable feed. Pix APIs provide live and historical data over REST and WebSocket, and the same data powers every Mirra indicator.
Frequently asked
Do I need to know programming to start algo trading?
For fully automated systems, yes. For rule-based alerts and scanners, platforms like NinjaTrader let you start without writing code.
How long should I paper trade?
Long enough to include a volatile week and a quiet one. For most intraday systems that is at least a few weeks.
Is crypto easier to automate than stocks?
Access is easier, since exchanges offer APIs directly. Data quality and venue risk are harder.
- Bank for International Settlements, Triennial Central Bank Survey of foreign exchange turnover, 2013–2022. Please verify figures against bis.org before quoting them.
- SEBI circulars on algorithmic trading by retail investors.
Educational content only — not investment advice. Trading in securities, derivatives, crypto and forex involves substantial risk of loss.