Proprietary quantitative trading · Coimbatore, India

Markets,
around the clock.

MarvelQuant is a proprietary quantitative trading firm based in Coimbatore, India. We research, test and run systematic strategies with our own capital. Our day starts with the Indian open and ends with the US close. Crypto markets fill the hours in between.

See open roles Meet the team The markets we trade

16people across strategy, technology, operations and administration
4market groups: India, United States, crypto and event contracts
24 hcoverage across Indian, US and crypto sessions
Own capitalonly. We hold no client money.

Lab reel · one trading day, 00:00 to 24:00 ISTIllustration, not market data

01 / About

About MarvelQuant

We are a research-driven trading firm. We build our own trading system, test every idea against real market rules, and trade only our own capital.

Mission

Turn careful research into reliable trading

A strategy earns live capital only after it has survived fair tests, a second reader and a run without capital.

What we build

One system from research to live

Backtests, sandbox runs and live trading share one engine, one clock and one order path. The code we test is the code that trades.

How we work

Evidence settles arguments

Every claim about a strategy or a system comes with the test that proves it. When the test fails, the claim goes.

Where we trade

India first, then the world

Indian derivatives during the day, US markets in the evening and crypto through the night, all from one team in India.

02 / Markets

The markets ledger

Every market we trade, with its regular session in the venue’s local time and in Indian Standard Time. All are listed derivatives and exchange-traded contracts.

Regular sessions · IST shown for US winter time
Region	Venue	Instruments	Local time	IST
India	NSE, BSE	Index and stock derivatives	09:15–15:30IST, Mon–Fri	09:15–15:30
India	MCX	Commodity derivatives	09:00–23:55IST, Mon–Fri; 23:30 close in US summer	09:00–23:55
United States	CME	Futures	17:00–16:00CT, Sun–Fri; one-hour daily break	04:30–03:30next day; Mon 04:30 to Sat 03:30
United States	S&P 500 options	Index options	09:30–16:15ET, Mon–Fri	20:00–02:45next day
Crypto	Deribit	Options and futures	24/7UTC; expiries at 08:00	24/7expiries at 13:30
Event markets	Kalshi and other venues	Exchange-traded event contracts	Per contractset by each venue	Per contract

IST times are shown for US winter time, in force from 1 November 2026. During US daylight saving, US sessions open and close one hour earlier in IST, and the MCX evening session closes at 23:30.

Regular sessions only. Each exchange publishes its own holiday calendar and special sessions. Exchanges are named for information only. MarvelQuant is not affiliated with or endorsed by any of them. No exchange or venue logos are used.

03 / Research

Machine learning and deep learning

We use statistics, machine learning and deep learning when they beat simpler methods in testing. Every model goes through the same lab steps as any other idea, and every model is judged on data it has never seen.

Models are hypotheses.
Markets grade them.

Deep learningNeural networks trained on GPUs over long histories of order-book depth up to 200 levels, trades, option chains and the Greeks, across India, the US, crypto and event markets. They are compared with simple rules on the same data and face the same out-of-sample tests.
Volatility and optionsModels of implied volatility and the Greeks across strikes and expiries, and of how option positions change as markets move.
Market microstructureFeatures built from order-book depth up to 200 levels, trades and open interest, recorded with the time each update arrived.
Execution qualityMeasuring what each order cost against the market at the moment it was sent, and learning where large orders should be split.
Research at scaleThousands of parameter sets tested in parallel on years of versioned data, then judged on periods the model has never seen.

Complete data across every market we trade

We hold complete data across every platform and market we trade. It covers order-book depth up to 200 levels, trades, full option chains with every strike and expiry, option behaviour and the Greeks. Deep learning models train on GPUs on this data. Research goes deep, and every model is reviewed before anything trades.

Order-book depth up to 200 levels
Trades
Full option chains
Option behaviour
Greeks

India

NSE, BSE, MCX

Index, stock and commodity derivatives.

Session times United States

CME futures, S&P 500 index options

Futures and index options, every strike and expiry.

Session times Crypto

Deribit

Options and futures, around the clock.

Session times Event markets

Kalshi and other venues

Exchange-traded event contracts.

Session times

A looping 12-second illustration on a dark canvas with five panels, built from synthetic data. Order book shows bid and ask depth over time with a mid-price line and trade prints. IV surface shows an option chain lattice rising into a swaying implied-volatility surface across strikes and expiries. Greeks shows four heatmaps for delta, gamma, vega and theta. Feature streams from these three panels feed a neural network labelled Model. Training shows training and validation loss curves with an early-stop marker at the best validation epoch. — **Illustration. Not market data.** Synthetic order book, option chain, Greeks and a neural network in training.The data and trading systems are built in Rust.

Walk-forward testing

A model trains on one window of history, is checked on the next, and is scored on a period after that which it never saw while training. Then the windows move forward and it starts again. Training stops when the validation score stops improving, even if the training score is still falling. The test window is scored once, after training ends.

Lab illustration · how a model is testedNot market data. Not results.

We do not publish models, signals or results. A model whose failures we cannot explain does not trade.

We hire researchers and engineers who test their own ideas hard. Send your CV to admin@marvelquant.com.

See open roles Machine learning role

04 / Method

The strategy lab

Ideas move through the same seven steps, in the same order, every time. An idea can stop at any step. Most do.

01
Write it down
Each idea starts as a short written hypothesis about how a market behaves, with the evidence that would prove it wrong.
02
Build the data
We keep our own versioned record of prices and contract terms for every venue we trade, expired contracts included. Lot sizes, fees and exchange limits are part of the record.
03
Test it fairly
Simulations charge real costs and respect exchange rules. Parameters are chosen on one period and judged on another the idea has never seen.
04
Review it
A second person reviews the code and the results before anything moves on. The reviewer did not write the idea.
05
Run it without capital
The strategy trades live market data in a sandbox first. Its orders and fills are checked against the simulation.
06
Run it with limits
Position, order-size and loss limits are set before launch and enforced in software. Large orders are split to stay inside each exchange’s quantity limits.
07
Watch it, then retire it
Every live strategy is monitored through its sessions. When live behaviour drifts from what we tested, we stop it and go back to step one.

Written before tested.

The hypothesis and its failure test exist before the first simulation runs.

Reviewed before run.

No strategy reaches a market on one person’s judgement.

Limited before live.

Every limit is in place before the first order is sent.

What we have built

One system. One event-driven trading system for research, simulation and live trading. The code that is tested is the code that trades.
One model for every contract. Contracts from every venue map to one instrument model. A strategy reads an Indian index option and a US future in the same way.
Data as it was known. Market data is stored with the time it arrived. A test sees only what was known at each moment.
Monitoring. Alerts follow the sessions, from the Indian open to the US close and through the crypto night.

05 / Quality

Testing and quality

Trading software fails in the moments that matter most: an opening auction, a lost connection, an exchange rejection. We test for those moments before a strategy ever meets them.

Unit and integration

Tests that fail first

A fix starts with a test that shows the fault. The fix is accepted only when that same test passes and nothing else breaks.

Exchange connections

Every connection, every case

Each broker and exchange connection runs a fixed set of data and order cases: subscriptions, fills, cancels, rejections, reconnects and restarts.

Rejections on purpose

We make exchanges say no

Before live trading, we send orders the exchange must refuse, outside the price band or the trading hours, and record its answer.

Review

One reviewer, one separate judge

Every change is read by a reviewer who did not write it, then judged on the test evidence before it reaches the trading system.

Replay

Sessions we can play back

Every state-changing event is logged in order. A trading session can be replayed to explain what happened and why.

Restart and recovery

Restarts that know their positions

After a restart, the system reconciles orders and positions with the broker before it trades again.

Every order takes the same path

Check 1Strategy limitsSize, position and loss limits for that strategy.
Check 2Freeze quantityOrders above the exchange freeze limit are split into child orders.
Check 3Price bandPrices outside the allowed band are refused before they leave.
Check 4MarginThe account must hold the margin the order needs.
SendExchangeOnly then does the order reach the exchange.

06 / Team

The team

Sixteen people work across strategy, technology, operations and administration, with a mentor in Mumbai. They come from engineering, finance, teaching and broking.

Founder

Marutha Kumar

Founder and Chief Technology Officer

Marutha Kumar founded MarvelQuant and leads its technology. He has spent 19 years in quantitative analysis. He builds trading systems and takes them into live markets.

He started with AFL, the formula language of the AmiBroker platform, moved to Python, and now builds the firm’s trading system in Rust. He is also an expert in Pine Script, the strategy language of TradingView. At MarvelQuant he designs the system that research, simulation and live trading share, and works with the strategy, technology and operations teams that run on it.

AFLWhere he started
PythonNext
RustToday

B.Tech, Computer Science
MBA, Finance
Pine Script
19 years in quantitative analysis

Leadership

Senthil Kumar
Chief Administrative Officer
Senthil runs administration at MarvelQuant. He has 15 years of experience in administration and client handling.
MBA, Marketing
Joshua Samuel
Head of Quantitative Strategy and Systems
Joshua leads strategy development and supervises the strategy, technology and operations teams. His role grew from algorithmic engineering at the firm. Before that he built indicators, execution engines and risk systems in Pine Script and MQL5 as a freelancer.
B.E. · MBA · PGDIB · NISM Series VIII

Capital markets

Broking, advisory and branch management

Nataraj
Senior Manager, Capital Markets
Nataraj has 28 years across engineering production, banking, insurance and capital markets. He was a relationship manager at IIFL, a regional development manager at three life insurers and a branch manager at Zebu.
DME · BBA
Karuppasamy
Senior Manager, Capital Markets
Karuppasamy has 19 years in capital markets. He worked at IIFL, ITI Financial Services and Way2Wealth, where he was a branch manager, then on the PCG desk at Kotak Securities.
BBM · MBA

Mentor

Ajit Rambhia
Mentor · Mumbai
Ajit mentors MarvelQuant. He has 37 years in trading, including platform trading and consulting on trading matters, and has handled a wide range of business situations.
37 years in trading

Strategy and research

Design, backtest and track the strategies

Arun Prasad Madan
Derivatives Market Analyst
Arun has 28 years in Indian equity markets, 12 of them in derivatives. He studies open interest: build-ups, unwinding, rollovers and how price, volume and open interest move together.
MCA · NISM Series I, VIII, XVI
Nagapoojith U S
Strategy Developer
Nagapoojith designs and backtests strategies for Nifty, Bank Nifty and Sensex. He has built straight and reverse Renko models, indicator-based systems and trend-reversal strategies.
B.E., Civil Engineering · NISM Series VIII
Shrutheena S
Strategy Developer
Shrutheena develops, backtests and refines options and automated strategies. From 2012 to 2016 she taught risk, financial and tax management to MBA students.
MBA, Finance (silver medal) · NISM Series VIII · SEBI Resource Person for Financial Education (2016)
Pratheena S
Backtesting and Performance Reporting
Pratheena backtests the strategies the firm runs and compares their live results with the tests. She writes the daily, weekly and monthly performance reports.
D.Ed.

Operations and execution

Run the trading day

Varun S M
Operations and Execution
Varun handles daily trade operations and order execution. Through the session he watches positions and margins, then reconciles trades with brokers.
M.Com (Computer Applications) · NISM Series VIII
Manojkannan S
Operations
Manojkannan manages operating workflows, reporting and the team’s software applications. He joined in 2025 after five years at Computer Warehouses.
B.E., Mechanical Engineering
Parivallal S
Operations
Parivallal works in operations. He came from relationship management at Ventura Securities and business development at Maitra Commodities.
B.E. · NISM Series VIII

Technology

Data, tools and the systems around the trading engine

Ajith
Python Developer and Data Analyst
Ajith writes Python tools and analyses market data. His work covers data processing, MySQL databases and web applications.
B.Sc.
Ishwarya
Python Developer
Ishwarya builds Python applications and data workflows, from processing raw data to database operations. She started as a Java developer intern.
B.Sc., Computer Science
Sathish
Data Analyst
Sathish analyses financial market data in Python and works on web applications. He joined after a Python internship.
BCA · NISM Series VIII

Administration

Senthil Kumar’s team

Sukumar
Administration
Sukumar works in administration in Senthil Kumar’s team. He has 25 years of experience in administration.
25 years in administration

One group

Strategy, technology and operations work as one group. Everyone can see the limits that apply to their work.

Short distance from idea to trade

The person who proposes a strategy works directly with the engineer who builds it and the trader who watches it live.

We are hiring across strategy, technology and operations.

See open roles

07 / Careers

Join the lab

We hire researchers, engineers and operations staff, both graduates and experienced people. Experience in finance helps. It is not required.

Research

Quantitative researcher

Form hypotheses about market behaviour, test them on clean data and take the ones that hold up to live trading.

We look for: Statistics and probability, careful code, and the habit of trying to break your own result.

Engineering

Software engineer, trading systems

Build and run the systems that record market data, simulate strategies and trade them across Indian, US and crypto sessions.

We look for: Strong systems programming, thorough testing, and care for software that runs through the night.

Operations

Operations and execution associate

Supervise live strategies through their sessions, manage risk limits and bring what the market shows back to research.

We look for: A working knowledge of derivatives, discipline inside limits, and clear written notes.

Any discipline

Open application

No role above fits? Tell us what you would work on.

Write to: admin@marvelquant.com

Send your CV and one piece of work you are proud of. Tell us which role you want and why.

We never charge candidates a fee at any stage.

admin@marvelquant.com

All roles at careers.marvelquant.com

08 / Contact

Contact

We do not take client enquiries. Write to us about careers or anything else related to the firm.