Independent algo engineering · Bengaluru
We build, test and run trading systems. We never promise returns.
From an idea on a napkin to a monitored strategy running in the cloud — development, walk-forward validation, execution tuning and 24×5 operations. Your strategy stays yours. Our job is to make it correct, testable and survivable.
What we do
Every engagement is a fixed scope with a written deliverable. No revenue share, no signals, no managed accounts.
Strategy development
Your rules, turned into deterministic, testable, production-grade code.
Backtesting and validation
An independent test harness that tells you whether the edge is real — or that it is not.
Strategy improvement
Take a working-but-fragile system and make it steadier, cheaper and easier to trust.
Execution engineering
Order routing, slippage control and the failure handling that decides your worst day.
Cloud deployment and hosting
Move off the laptop. Containerised, monitored, restartable, on your own cloud account.
Monitoring and market-hours support
Someone technical watching the system while you watch the market.
Validation, not decoration
A backtest is a hypothesis test, not a brochure
Most strategies that arrive here look profitable. Most of that edge disappears once we add real costs, honest fills, and a data split the strategy has never seen. We rebuild your test harness so the number you look at is the number you can defend.
- Point-in-time data, corporate actions and expiry handling checked first
- Brokerage, STT, exchange charges, GST, stamp duty and impact cost modelled per leg
- Walk-forward folds, parameter surfaces, regime and stress windows
- Written verdict — including "this does not work", when that is the answer
How an engagement runs
Five stages, each with an exit gate you approve before we spend your budget on the next one.
Discovery call
30 minutes, free. You describe the idea, the market, the capital and the constraints. We tell you plainly whether it is buildable, and where the hard parts are.
Written spec
Entry, exit, sizing, risk caps, data sources, edge cases, failure behaviour. Nothing is coded until the spec is signed off, because ambiguity is where money is lost.
Build and unit-test
Clean, readable code in your repo. Deterministic replay, logged decisions, tests around expiry, holidays, partial fills and reconnects.
Validation
Costs, slippage, walk-forward, stress windows. You get the report and the raw trade log, not a screenshot.
Deploy and monitor
Containerised on your cloud account, your broker keys, your control. Heartbeats, alerts, kill switch, daily reconciliation against the broker statement.
Engineering, not scripting
The parts nobody demos
A strategy is 20% of a live system. The rest is the boring machinery that decides whether a bad afternoon costs you a rupee or a fortune: reconnect logic, idempotent orders, position reconciliation, margin checks, a kill switch that actually kills.
| Failure | What we build for it |
|---|---|
| Feed goes stale or latches | Staleness watchdog, quote age gate, halt-and-alert |
| Duplicate or lost order | Client order IDs, idempotent retries, broker reconciliation |
| Process dies mid-position | State persistence, restart recovery, position resync |
| Strategy misbehaves | Loss cap, order-rate cap, manual kill switch, flat-by-time rule |
Code shop
Ready-made connectors, harnesses and reference strategies. Source included, licensed per developer, delivered as a download.
Broker API connector (Python)
Production-grade REST + WebSocket client with re-auth, reconnect and idempotent orders.
Event-driven backtest harness
One strategy class, two modes: replay history or trade live, with the full Indian cost stack built in.
Reference strategy: opening range breakout
A complete, readable intraday reference implementation — structure to learn from, not a money machine.
What we will not do
We do not give buy/sell calls, run your account, share profits, or quote expected returns. We are not a SEBI-registered Investment Adviser or Research Analyst. If a vendor shows you a "guaranteed 8% monthly" algo, walk away — and read our regulatory disclosure for why the 2025 SEBI algo framework exists.
Research notes
What we are testing, breaking and fixing.
The SEBI algo framework, in plain terms
What the February 2025 circular actually asks of retail algo traders, what your broker has to do, and which parts affec…
Nine ways a backtest lies to you
A field guide to the failures we find most often when auditing strategies that looked excellent right up until they wen…
Slippage is a strategy parameter, not an accident
Once you measure fills properly, entry timing stops being a detail and becomes one of the cheapest improvements availab…
Bring us the strategy that almost works
Send the rules, the code or the messy spreadsheet. We will tell you what is wrong with it before you pay us anything.