Abhilash Dhumatkar

AI trading research · Systems engineering · Founder, X-STOX

Abhilash Dhumatkar

I study markets like a scientist and ship trading systems like an engineer.

Every strategy starts as a hypothesis, gets written down as a rule, and has to survive backtests, walk-forward tests and live capital before anyone else runs it. Since 2016 I have built MetaTrader and Indian-market trading software this way at X-STOX — and I now research where machine learning genuinely improves systematic trading.

See the workRead the research
Portrait of Abhilash Dhumatkar, trading systems engineer and founder of X-STOX, in a grey blazer and black shirt
Founded X-STOX
2016
Launched products
6
Awards & recognitions
7
Core stack
MQL5

In short

Hypothesis. Rule. Test. Then ship.

I started trading at eighteen and later trained in software engineering and data science. X-STOX is where those halves met: strategies I was already trading, specified and tested properly so they could run without me. Ten years on, the discipline is the same — the rule goes on paper first, the code comes second, and the risk is stated before the price.

Read the longer story

Fig. 01 · A rule, written down

The system stands aside when the market stops behaving.

A trading rule written downA price line moves calmly, then spikes during a news release. A dashed threshold marks where the spread exceeds three times normal. A note reads: if spread is greater than three times normal, stand aside. The system re-enters once conditions calm.spread 3× normalif spread > 3 × normal: stand asideresume when calm for 3 barsre-entrycalmnews releasecalm again
Illustration, not market data. The point of a system is the rule you write down before the moment arrives.

Open research

Where does machine learning actually help?

  • Q1
    Regime classification

    Can a learned model tell when a strategy's conditions are present at all — better than the rule's own filter, out of sample?

  • Q2
    Execution estimates

    Can spread and slippage around news releases be forecast well enough to change sizing or stand-aside decisions?

  • Q3
    Signal filtering

    Which of a rule's signals failed under similar conditions — and does filtering them survive walk-forward testing after costs?

Read the research brief

Method

Seven steps. Every system.

  1. Written specThe rules go on paper in plain language before any code exists.
  2. BuildNative MQL4 or MQL5, with Python where the research needs it.
  3. BacktestAcross market regimes, not one good year.
  4. Forward testOn a live account at small size, long enough to see a bad week.
  5. DocumentA manual that states the mechanism, including the parts that carry risk.
  6. ShipWith installation support, presets and a person to ask.
  7. MaintainBrokers change, platforms update, markets change character.

Stack

Native where it trades. Python where it learns.

MQL5MQL4PythonMetaTrader 5MetaTrader 4BacktestingWalk-forwardRisk engines

Markets

From gold on M1 to NIFTY options.

  • Forex majors and crosses
  • Gold and silver
  • Index CFDs
  • Indian equities, futures and options
  • Crypto pairs on MetaTrader

Selected work

Trading software, designed and shipped.

All 8 projects

The company

X-STOX. Since 2016.

A Pune trading software company that sells software, not signals. Every product page states what the system does — including how it can lose — before the price.

How X-STOX worksVisit x-stox.com

Writing

Notes on systems, risk and research.

Featured in

Features and mentions.

Business TalkzMagazine feature · Mar 2025The Deccan MessengerRoundup · Mar 2023Fox InterviewerContributed article · Jan 2022

Media, recognition and press kit

FAQ

Questions people ask.

Who is Abhilash Dhumatkar?

Abhilash Dhumatkar is a trading systems engineer and AI trading researcher based in Pune, India. He founded X-STOX in 2016 and writes the strategies and MQL4/MQL5 code behind its MetaTrader and Indian-market trading software.

What is X-STOX?

X-STOX is a Pune-based trading software company founded by Abhilash Dhumatkar in June 2016. It builds MetaTrader 4 and 5 Expert Advisors, technical-analysis software for Indian and global markets, and custom trading systems, sold on x-stox.com and Amazon.in.

Does X-STOX use artificial intelligence?

The products X-STOX ships today are rule-based, and each product page describes its actual mechanism. Abhilash's current research asks where machine learning genuinely improves a systematic strategy — for example regime classification, execution estimates and signal filtering — tested out of sample and after costs.

How does Abhilash test a trading system?

Every system starts as a written specification, is built in MQL4 or MQL5 with Python for research, backtested across different market regimes, forward tested on a live account at small size, and documented — including the parts that carry risk — before release.

Which markets does he work with?

Forex majors and crosses, gold (XAUUSD) and silver, index CFDs, crypto pairs on MetaTrader, and Indian equities, futures and options.

How can I contact Abhilash Dhumatkar?

Use the contact page on this site or message him on LinkedIn. He is open to partnerships, custom trading-system projects, interviews, speaking and research collaboration.

A closing note

A trading idea is not a trading system. Everything between the two — sizing, slippage, a rejected order at 2 a.m., the week that breaks your backtest — is the work I find most interesting. If you are building in that gap, I would like to hear from you.

Start a conversation