Technology that fits your trading process.
Ready-to-use terminal capabilities, broker integrations, and custom algorithmic or research solutions.
Quantitative trading technology
From market data to trading decisions.
One integrated platform for NIFTY options: research ideas, engineer strategies, validate them against evidence and execute through a flexible trading terminal.
Research-led developmentReproducible validationDiscretionary + algorithmic execution
I'm a
Pick one to highlight the parts written for you. Nothing is hidden.
All displays on this page use demo data.
The platform
One connected workflow, from the first tick to the last order.
Each stage feeds the next. Use the whole platform or engage with the capabilities that fit your workflow.
01 / DATA↗
Capture & check
Tick-level capture, checked every trading day.
02 / FEATURES↗Engineer features
1-second bars and 8 proprietary indicator families, built nightly.
03 / RESEARCH↗Investigate ideas
Strategy maps and logged studies. Most ideas are rejected.
04 / VALIDATION↗Test the evidence
Reproduced exactly, then tested on unseen dates.
05 / EXECUTION↗Act with control
One terminal, five ways to trade.
06 / TOOLS↗See what happened
Visual Alpha Profiler, chart workbench and marked-chart viewer.
07 / HOW IT IS BUILT↗Built and checked
AI agents build it. Other agents try to prove it wrong.
A differentiated technology foundation.
The platform architecture, proprietary tools, engineering approach and potential areas for collaboration.
The systems behind the workflow.
How research pipelines, strategy engines, validation systems and execution interfaces fit together.
Algorithmic trading
Turn your trading idea into a tested, automated strategy.
We build AI-assisted, statistically validated Algorithmic Trading Machines (ATMs): strategy research, reproducible large-scale backtesting, signal handling, and broker order routing — all in one connected process.
Research the idea
The idea is prototyped in Pine Script or Python, then reproduced and tested in our Python backtest engine.
Generate signals
The validated strategy runs live in our Python engine and sends its entry and exit signals to the order engine.
AlphaStar Platform
Receives triggers, applies order rules and manual controls, selects the best strike at or near the money, and acts as the broker gateway.
Route to the broker
Places orders through your own broker account's API. Built today for Zerodha Kite Connect and FlatTrade PiConnect.
Validate your strategy the way we validate ours. We set up the same pipeline for you: an exact Python replica of your strategy, large-scale backtests on data you are licensed to use, and a one-time check on dates it has never seen, before any capital is used.
See how a strategy is tested ↓Strategy implementation
Each strategy we take forward has a Python replica, run on an engine that reproduces the Pine strategy engine's next-bar fills, margin and commissions. The live strategy runs in Python.
Broker connectivity
Orders go through your own broker account's API. Current integrations are Zerodha Kite Connect and FlatTrade PiConnect, subject to account access and API capabilities.
Hands-on trading
Trade five ways, from one terminal.
A terminal to trade from, and a simulator to practise on. Ready-made for day trading NIFTY options; any NSE or BSE security can be traded with customisation.
5-way input action hub
Training simulator
Practise in the same terminal you trade with.
Replay a recorded trading day from your own data, tick by tick, through the real terminal, at real time or faster. It picks a recorded day and contract from a tick store; for clients, it connects to tick data they are licensed to use.
- The day being replayed is kept out of indicator warm-up, so the trainee cannot see the future.
- Uses the same hotkeys, gamepad buttons and voice commands as live trading.
- Replays recorded ticks, not synthetic pricing.
- A session trade log and simulated execution report straight afterwards.
Data
Tick-level capture, checked every trading day.
Built to capture every tick of a live feed, check every trading day for gaps, and repair or flag what is missing. Held for our own research, never distributed.
Distributed ingestion
Every NIFTY option tick is queued in RabbitMQ and written to the database by Celery workers, so a slow write does not lose ticks. Each day is archived while capture keeps running.
Gap check
A daily health check finds missing seconds. Gaps in the feed itself are detected and recorded.
Repair or flag
Verified gaps are filled; a day that cannot be repaired is marked, so data-quality issues are visible rather than silently ignored.
Content hashing
SHA-256 fingerprints catch any change to stored data.
Read-only sealing
Finished data is sealed so nothing rewrites it. Long runs are supervised and restart from their last checkpoint.
Market rules registry
One source for the exchange's rules: holidays, lot sizes, freeze quantities and session hours. The same file generates the Pine Script code.
Research history
History kept for our own individual research. Option ticks, and longer-run index and volatility bars, kept to research and test our own individual strategies. None of it is distributed.
Features
Every night, ticks become 1-second features, with no look-ahead.
Ticks become 1-second bars and 8 proprietary indicator families, stored as 13 datasets. A strategy only ever sees what it could have known at that second.
Time-aligned feature pipeline
Features from several timeframes are aligned to one 1-second grid. A slower timeframe is lagged one bar before it meets the 1-second grid, and a test checks the stored data for it.
No look-ahead1-second gridBuilt nightly
Stored and sealed
Each contract-day is written once, fingerprinted with SHA-256 and sealed read-only, so every result can be traced to the exact data it used.
SHA-256 fingerprintsRead-only sealingTraceability
Research
Strategies start as hypotheses, then face the evidence.
We design indicator families, test ideas as logged studies, and drop most of them because they don't meet their success criteria.
ACTIS · Situation map
Where the market sits between bullish and bearish, at three speeds, every second.
BearishBullish
Two indicator families, P4 (Prem's Precision Power Parity) and MF (Minu's Favorite), each read at 1-minute, 5-minute and 15-minute views, updated every second. Marker positions are illustrative.
A slow average on a fast chart
It jumps every 15 minutes and lags.
ACTIS, ladder-free
The slow view is rebuilt from fast prices, so it moves smoothly and timeframes compare directly.
MTFS2 · Multi-timeframe stack
Several timeframes, stacked into one decision.
One decisionIllustrative layering, not a trading recommendation.
A second ACTIS line runs on 1-minute bars. Each strategy family is tested the same way: reproduced, then confirmed on dates it has never seen.
Research library
Strategy families and the research infrastructure behind them.
ACTISSituation map at three speeds, on a 1-second grid. Families P4 and MF.
ACTIS on a 1-minute gridA second ACTIS line, running on 1-minute bars.
MTFS2Several timeframes stacked into one decision.
204Pine Script filesOur TradingView charts compendium: strategies, indicators and their revisions.
53Logged studiesSince September 2026, tracked in holdout and multiplicity ledgers. Most are rejected, and that is the point.
Reinforcement-learning platformAgents trained on the same data and cost model as the backtester.
Counts as of 31 Oct 2026.
Validation
Reproduced exactly, then tested on unseen dates.
Most of our engineering time goes here. A backtest that cannot be reproduced proves nothing.
Finalists meet unseen dates once, before any money does
Many ideas go in; at most three finalists reach the vaults, and each vault is read once.
01 · DEVELOP197 datesBuild and tune freely
02 · FREEZEat most 3Commit the finalists
No more changes.
03 · VAULT A96 unseen datesRead once
04 · VAULT B96 unseen datesA second, separate read
05 · PAPER1 weekLive market, no money
06 · LIVEthenOnly after every step above
Nothing on this page claims a strategy has passed. This is the path one must take.
Reproducibility and statistical discipline
✓
Pine Script / Python parityEvery trade is compared: entry and exit time, entry and exit price, and the result, under matched test conditions. A difference fails the check.
✓
Execution-aware backtestingThe engine reproduces next-bar fills, margin and commissions.
✓
Walk-forward evaluationOptimise on one period, test on the next. Every run records the days it used.
✓
Every try is countedWith a Holm correction, so luck cannot pass as an edge.
✓
Random-entry comparisonsEvery rule is compared against random entries.
✓
Daily-level statisticsConfidence is measured on daily results, not trade by trade.
The pass bar: a mandatory deployment gate
A strategy must meet all four.
≥ 2.0Sharpe ratio
≥ 1.5Profit factor
≥ 75%Profitable trades
< ₹10,000Max drawdown on ₹1,00,000
Mandatory targets, not results. This page shows no trading performance.
Why parity matters
The Pine Script strategy and its Python replica must take the same trades under the same data, timing and execution assumptions. A mismatch is a defect to fix, not something to explain away after seeing the result.
Pine ScriptPython replicaCheck
Entry and exit timeEntry and exit timeMatch
Entry and exit priceEntry and exit priceMatch
Trade resultTrade resultMatch
Tools
See what a strategy did. Understand why.
Built for our own research, and the basis of our custom charting work. Shown on synthetic demo data. Select any screenshot to enlarge it.
Visual Alpha Profiler
Every trade or event of a study, overlaid on one chart, with standard slice filters and distributions.
Synthetic demo study (made-up 2030 data)Chart Workbench
Multi-pane charts in IST or UTC time, with high-resolution snapshot and export.
Chart explorationMarked-Chart Viewer
Redraws a strategy's indicator plots for any past day from our own data, including expired contracts. Mark up and comment on any bar.
Review & annotationThe trading terminal and its simulator are shown under Hands-on trading.
Engineering method
AI-accelerated engineering. Adversarially tested. Human-approved.
AI coding agents build against strict architectural specifications, while independent review agents and automated test suites actively try to break every change. Evidence, deterministic testing, and accountable human sign-off determine what ships.
Every significant change goes round this loop
01 · Build
An agent makes the change and writes down every claim it makes, each with the command that proves it.
02 · Independent review
A separate agent gets only those claims and the code change, not the builder's reasoning. It re-runs every command and tries to prove each claim wrong.
03 · Fix
Whatever it disproves is corrected. If the two still disagree after two rounds, a person decides.
04 · Record
The outcome goes into a findings tracker and a daily engineering journal, so lessons carry into the next change.
The manuals are checked the same way. Numbers in them are tagged, and a script re-measures each one. A number that no longer matches, or that nobody has re-measured recently, fails the check.
2Independent review agents
20Numbered engineering manuals
72Days of engineering journal
71Test files
As of 31 Oct 2026.
Ways to work with us
Technology built around your requirements.
We build custom trading terminals, broker API integrations, and quantitative validation pipelines for algorithmic and discretionary traders.
Trading terminal
A configurable terminal for discretionary and assisted trading.
- Dual CE/PE trading views
- Hotkeys, gamepad and voice inputs
- Broker API connectivity
- Simulator and replay
Algorithmic trading
Strategy implementation, testing, signal handling and execution.
- Strategy development and Python replication
- Backtesting and validation on data you are licensed to use
- Webhook and broker integration
- Automated order management, within the exchanges' algo trading rules
Research & charting
Quantitative research and visualisation shaped around a defined problem.
- Feature and indicator engineering
- Research and charting tools
- Validation pipeline design
- Custom integrations
Scope and delivery are agreed in a requirements discussion. These are service propositions, not a statement that every feature is offered as a self-service product.
Founder
Built from experience in technology and systems.
Prem Srinivasan
Principal Architect
Senior IT Professional, and Entrepreneur with 25+ years of leadership spanning Enterprise Software Delivery, BI & Data Warehousing Architecture, International Business Development, and Multidisciplinary Ventures.
LinkedIn: linkedin.com/in/prem--srinivasan ↗Get in touch
Want a terminal, a data pipeline, or a strategy engine like this?
Tell us whether you're exploring a trading terminal, broker integration, algorithmic workflow or quantitative research solution. We can start with the problem and shape the right engagement.
Email: info@alphastar.in
Website: alphastar.in
Office: AlphaStar Consulting Services
Perumal Koil St., Kotturpuram, Chennai 600085, India