All Projects
Case Study

QuantOI

Systematic options-trading intelligence for Indian index derivatives

Type
In-house R&D Product
Markets
NSE / BSE index derivatives
Status
Paper-trading validation
Interface
Telegram bot

Our own R&D product: a system that watches live Nifty and Sensex options chains through the full trading day, separates genuine institutional conviction from noisy open-interest spikes, and turns confirmed signals into defined-risk spread trades — every one simulated through a realistic paper broker with real exchange cost modeling. It runs unattended and reports to a Telegram bot. Currently in disciplined paper-trading validation before any capital goes live.

The Problem

Raw open-interest and volume spikes — the signals most retail tools alert on — are frequently false moves ("stop-hunts"). Acting on them directly is how retail traders bleed money.

Naked option buying carries undefined risk, and most retail tooling looks at the options chain in isolation: no institutional positioning, no macro context, no cross-session persistence.

Testing a strategy honestly is itself an engineering problem: naive simulators ignore fill quality, order-size limits, and transaction costs — and quietly overstate results.

What We Built

We built the whole pipeline: real-time market data in, scored and filtered signals, risk-defined trade construction, realistic simulation, and a Telegram bot as the operations console — with a hard safety flag ensuring no path to live orders during validation.

QuantOI performance report with equity curve — illustrative demo data
The weekly report format: equity curve, win rate, R-multiples, cost tracking. Numbers shown are illustrative demo data — not actual trading results.

Observation-window signal scoring

A candidate signal isn't trusted on arrival — it's scored over an observation window against OI persistence, volume confirmation, VWAP alignment, IV percentile, futures positioning, and higher-timeframe trend.

Defined-risk spread systems

Two strategies, both risk-capped by construction: directional debit spreads on confirmed breakouts, and credit spreads defending major open-interest walls. No naked exposure.

Realistic paper broker

Simulated fills respect order-book depth, exchange freeze-quantity order splitting, and the real F&O transaction cost structure — so results aren't quietly flattered.

Institutional context engine

SEBI participant-flow data, max pain, gamma walls, and cross-session OI persistence give every signal the context most retail tools ignore.

LLM-generated macro briefing

A daily pre-market briefing synthesized by an LLM from global markets and news, feeding sideways-day detection that suppresses trade suggestions in choppy conditions.

Telegram bot operations

Trade alerts, active position tracking, portfolio Greeks, institutional summaries, and system health — pushed automatically or on demand, from anywhere.

Self-learning calibration

Every signal outcome is logged into a monthly calibration pipeline that proposes — never silently applies — threshold adjustments as evidence accumulates.

Stock swing-trade screener

An independent daily module screens ~210 F&O stocks through Minervini's trend template, volatility-contraction detection, and a strict reward-risk gate — with near-misses shadow-tracked to keep thresholds honest.

QuantOI system architecture from market data to Telegram alerts
The pipeline: live market data → scored signals → risk-defined trades → realistic simulation → Telegram.

The Outcome

  • Runs unattended through every market session, logging roughly a thousand market observations a day into structured datasets.
  • Every trade idea is costed and simulated realistically — order-book-aware fills and real exchange charges, not optimistic assumptions.
  • Every outcome feeds the calibration loop, so the system's thresholds improve from evidence rather than intuition.
  • A hard safety flag guarantees no execution path to real money until validation is complete — discipline first, capital later.
QuantOI Telegram bot showing a confirmed signal alert
The operations console is a Telegram bot — alerts, positions, Greeks, and health checks from a phone.

Interface visuals are illustrative mockups — client data is not shown.

What We Did

Real-Time Data EngineeringQuantitative System DesignAutomation & AlertingLLM Integration

Technology

PythonKite Connect APIWebSocket StreamingTelegram BotClaude (LLM)Pandas

What's Next

  • Cloud deployment so the system runs on managed infrastructure instead of a local machine.
  • A web dashboard UI over the Telegram bot — positions, equity curve, signal history, and calibration reports in the browser.
  • Live broker integration behind strict risk limits, once the paper track record has proven itself over a meaningful period.
  • Productizing the alert engine so other systematic traders can subscribe to the intelligence layer.

Self-initiated R&D product, currently in paper-trading validation. No live-capital performance is claimed, and nothing here is investment advice.

Want something like this for your business?

Catalogs, lead capture, automation, AI tools — we design and build systems that do real work. Tell us about your problem and we'll propose a solution in plain language.

Start a Conversation