Introductory Context
"Quantitative market-making and order book microstructure strategy on Indian derivative exchanges. Modeling Avellaneda-Stoikov inventory skew and toxic flow markout."
The Business of Liquidity Provision: Earning the Spread while Surviving Adverse Selection
Electronic market-making is the cornerstone of modern algorithmic trading on the National Stock Exchange (NSE). Market makers continuously provide liquidity by posting simultaneous two-sided quotes—bidding to buy at price b and offering to sell at price a. When retail and institutional market participants hit those resting orders, the market maker captures the quoted bid-ask spread s = a - b.
However, naive market making—posting static quotes around the midpoint—is guaranteed to fail due to Adverse Selection (Toxic Flow). When informed algorithmic traders possess directional signals or institutional execution algorithms (such as TWAP/VWAP) aggressively sweep the order book, the market maker's resting quotes are picked off just before the market price runs aggressively in the adverse direction. A robust market-making engine must dynamically skew quotes based on Inventory Risk and detect toxic flow before it creates catastrophic inventory drawdowns.
• The Core Market-Making Economics
Expected Markout per Filled Share:
E[Markout] ≈ (1 - π) × (s / 2) - π × δ
where π is the probability of interacting with informed (toxic) flow, s is the quoted spread, and δ is the adverse price move following an informed fill.
Reservation Quotes & Inventory Skew (Avellaneda–Stoikov)
Let m_t be fair value (midpoint or microprice), q be the signed accumulated inventory (positive for long, negative for short), and σ be short-horizon price volatility. Reservation bid and ask quotes are defined as:
b_t = m_t - δ_0(σ) - κ × q
When accumulated inventory is long (q > 0), the bid backs off further from the mid, reducing the likelihood of buying additional inventory.
a_t = m_t + δ_0(σ) - κ × q
When long (q > 0), the ask is tightened closer to the mid, aggressively incentivizing incoming market buyers to lift the offer and offload the inventory.
Here δ_0(σ) is the baseline half-spread (widened during elevated volatility regimes) and κ > 0 is the inventory risk aversion coefficient. When inventory reaches the absolute risk cap |q| ≥ q_max, the quote engine immediately pulls the inventory-increasing side to avoid exceeding balance-sheet and SPAN margin constraints.
NSE Colocation & Microstructure Realities in India
On Indian exchanges, quantitative market makers operate under rigorous regulatory and technological conditions:
- NSE Co-location (Colo): Tier-1 market makers host high-frequency server racks directly inside the NSE Data Center at Bandwidth Hub, Kurla (Mumbai), processing 10-Gbps Tick-by-Tick (TBT) UDP multicast data with sub-microsecond latency.
- SEBI Algorithm Registration: All algorithmic strategies operating on NSE/BSE must be approved by the exchange and assigned a unique SEBI Algo ID. Spoofing, layering, or deliberate order-book manipulation is strictly illegal under SEBI Prohibition of Fraudulent and Unfair Trade Practices (PFUTP) regulations.
- Security Transaction Tax (STT) & Exchange Fees: Quoting wide spreads in single stocks is penalized by high STT on cash deliveries. High-frequency market making is primarily concentrated in Index Options, Nifty Futures, and liquid ETF market-making schemes.
Structural Failure Mode: Toxic Inventory Accumulation
During rapid macro gap events (such as election results or unexpected central bank rate announcements), informed institutional flow sweeps the order book in one direction. A market maker that fails to widen spreads or disable quotes during high-volatility spikes will accumulate maximum adverse inventory right at the top/bottom before the market crashes, suffering severe mark-to-market losses.
Educational and Research Simulator Only
This specification is intended strictly for academic study and quantitative simulation environments. Live market making requires specialized exchange memberships, SEBI registration, dedicated infrastructure, and substantial risk capital.