What Algo Trading Actually Is

Algorithmic trading decodes to something you already do: pre-written rules executing without a human pressing the button at decision time. That's it. The mystique dissolves once you see it as a ladder you've already started climbing:

Rung 1 — Standing orders (you're here): GTT, brackets, baskets (Chapter 6) — single rules, resident at the broker, executing while you sleep. Congratulations: a stop-loss order is an algorithm with one instruction.

Rung 2 — Alert-driven semi-automation: platform conditions fire (Chapter 6's sentries), a human confirms, pre-built orders execute. The judgment stays human; the watching and the mechanics are machine. For most retail traders, this rung is the sweet spot — most of automation's discipline benefits, none of its unattended risks.

Rung 3 — API-driven automation: your own logic places orders programmatically. An API (Application Programming Interface) is simply a doorway a broker opens so software — instead of thumbs — can do what the app does: fetch prices and positions, place and modify orders. Indian brokers offer these openly (Zerodha's Kite Connect is the best-known; most major brokers have equivalents), typically as authenticated web endpoints with official libraries in Python and other languages. The concept in one line of pseudo-logic:

`` every minute: price = get_latest_price("NIFTY_FUT") if price > my_trigger and no_open_position: place_order(buy, quantity=my_size, type=LIMIT, price=price) place_order(sell_stop, at=my_stop_level) # exit born with entry ``

Readable, yes — but notice what the four lines quietly demand: session handling (API logins expire daily in India — your code must re-authenticate), rate limits (brokers cap requests per second; exceed and you're blocked at the worst moment), error handling for every call (what if the order rejects — Chapter 1's checkpoints apply to code too), and state tracking (what does "no_open_position" mean after a partial fill?). The strategy logic is usually the easy 20%; the operational plumbing is the hard 80% — a preview of Chapter 11, which is entirely about that 80%.

Rung 4 — Institutional HFT: co-located servers inside exchange premises, microsecond latency, market-making at scale (your Market Structure school, Chapters 7–8). Named here for one reason: so you stop competing with it. No retail API strategy wins a speed race — home internet to broker to exchange is thousands of times slower than co-location. Retail automation's edge is never speed; it's discipline at scale — rules executing without fear, boredom, or FOMO, across more instruments and hours than any human can honestly watch (the Behavioural Finance school's mast, given hands).

The SEBI framework — India's rules of the road (know the shape, verify the current details): India regulates retail algo trading through a broker-accountability model that has been progressively formalized (SEBI's framework for retail participation, with provisions effective from 2025): brokers are responsible for the algos their clients run; API-driven orders are tagged and identifiable; strategies beyond basic thresholds route through registration/approval via the broker; and unregulated third-party "algo platforms" promising returns have been explicitly targeted — the era of anonymous Telegram bots wiring into your account is legally closed, by design. Two practical consequences: (1) automate through your broker's official API and processes, never through credential-sharing with third parties (Chapter 13 will call this what it is — handing strangers your account); (2) the regulatory details evolve — before building Rung 3, read your own broker's current algo/API policy page, because it is the operative document for your account. (Content team note: this section needs a scheduled regulatory review alongside the link checks.)

The honest decision framework — should you climb? Automation amplifies whatever exists: a positive-expectancy, precisely-written strategy gains consistency; a vague or negative-expectancy one loses money faster and more reliably than any human could. The qualifying exam is Module 3, verbatim: if your rules aren't stranger-enforceable (Behavioural Finance school's test — a computer is the ultimate stranger), they aren't automatable; if they haven't survived the three-filter pipeline (Chapter 9), automation just industrializes the fantasy. And plenty of excellent traders rationally stop at Rung 2 forever: if your edge involves judgment (structure reading, regime feel), the right division of labor is machines watching and executing mechanics, human making the calls — automation of the periphery, not the decision.