Trading & Securities

T+1 means a trade executed today (T) is settled tomorrow (T+1). Technically: On trade day, NSE sends trade files to NSCCL. NSCCL performs multilateral netting — for each security, calculates net delivery obligation per broker (sum of all client buys minus sells). On T+1 morning: 1) Securities settlement — brokers with delivery obligation instruct NSDL/CDSL to debit securities from selling clients' demat; depository transfers to buying clients' demat by afternoon. 2) Funds settlement — net funds obligations settled via clearing banks through RBI RTGS, funds credited to selling brokers, debited from buying brokers. India achieved T+1 in Jan 2023 for all securities — world's first major market to do so. Key challenge: compressed timeline required real-time reconciliation between broker, depository, and clearing corp.

SPAN (Standard Portfolio Analysis of Risk) is the margin methodology developed by CME and adopted globally for derivatives. It calculates the maximum one-day loss of a portfolio across 16 risk scenarios (price up/down × volatility up/down combinations). How it works: 1) Define risk array — 16 scenarios for each contract. 2) Calculate worst-case loss per scenario for each position. 3) Sum across portfolio considering correlations. 4) SPAN margin = worst single-day loss scenario. In India: NSCCL calculates SPAN daily; brokers must collect at least this much from clients. Exposure margin (additional 2–5% of position value) collected on top. Example: If Nifty Futures SPAN margin is ₹1.2L per lot, buying 1 lot blocks ₹1.2L + exposure margin from client account.

A matching engine maintains an order book per instrument — a sorted list of buy orders (bid side, highest price first) and sell orders (ask side, lowest price first). Matching algorithm: 1) Price-time priority — best price matched first; among same-price orders, earlier timestamp wins. 2) When a new buy order arrives at price ≥ best ask, a trade is generated. 3) Partial fills possible if counter order quantity is smaller. Performance requirements: NSE NEAT processes millions of orders per second, sub-millisecond latency. Implementation: C++ with lock-free data structures, FIFO queues per price level, pre-allocated memory pools. Special scenarios: market orders (execute at best available), limit orders (execute only at specified price), stop-loss (trigger only when price crosses threshold).

A Depository (NSDL/CDSL) is like a bank for securities — holds all shares electronically. A Depository Participant (DP) is the intermediary between investor and depository — banks (HDFC, SBI), brokers (Zerodha, Groww) are DPs. Every investor has a BO (Beneficiary Owner) account at a DP. Demat transfer flow on selling: 1) Investor sells shares via broker. 2) After trade, NSCCL sends delivery obligation to clearing member (broker). 3) Broker instructs DP to debit shares from client's demat (DIS — Delivery Instruction Slip, now digital). 4) DP debits client's BO account, credits clearing pool account of clearing corporation. 5) Clearing corp then credits buyer's BO account. On buy: reverse — clearing corp pool → buyer's DP → buyer's BO account. All this happens on T+1 in India.

Trading API rate limiting requirements: prevent order spam, protect exchange connectivity (FIX connections are limited by NSE per broker), and ensure fair usage. Design: 1) Multi-level limiting — per API key (client), per IP, per exchange segment. 2) Algorithm: Token bucket preferred over fixed window — allows short bursts but limits sustained rate. Token bucket: each client gets N tokens per second; each order consumes 1 token; tokens refill at rate R/second. 3) Storage: Redis with INCR and EXPIRE for distributed rate limiting across gateway pods. 4) Different limits per endpoint: /orders (stricter — e.g., 10/sec per client) vs /quotes (lenient — 100/sec). 5) Soft limits with headers (X-RateLimit-Remaining) + hard limits with 429 response. 6) Algo traders on paid plans get higher limits via a flag in the token metadata.