An open quant desk for Indian markets
Indian market data is plentiful and context is scarce. A number on its own says little; this site shows each one with its full history, where today sits in that history, how it was computed, and the data behind it.
Rules we keep. Every series is built from public, primary sources (NSE, NSE Indices, AMFI, MoSPI, RBI, FBIL and the Federal Reserve among them), checked against the source's own published figures, and documented in a method note. Every backtest and model was specified in writing before it was run, and null results are published as plainly as positive ones. Universes include stocks that later delisted. Costs and Indian taxes are included where they matter.
What we don't publish. Raw index levels licensed by index providers, prices from licensed feeds, or third-party copyrighted series. Where a measure depends on them, we publish derived statistics only. An automated gate checks every data release for this.
Not advice. Everything here is historical research. Nothing is a recommendation to buy, sell or hold any security, and model states are never shown as live signals.
Data. Every chart offers its data as CSV and JSON. If you use it, credit tipsheet.markets and the original source named on the chart.
How explanations are written
Some charts carry a box headed How to read this. We write those by hand: what the chart shows, how to read it, and the one thing people get wrong. They contain no live numbers.
A few pages also carry a box headed Today, marked with a small diamond. A language model, DeepSeek, writes it each day from the figures already on the page, and the box says so, with the date of those figures. The model never computes anything. It is handed a short list of numbers, each with the exact wording to use, and asked to describe them.
A program then checks the text before it is published. Every number must match one of the figures it was given, at the precision it was given; every year must be one it was given. The text must not predict, advise or use the stock phrases of machine writing. A text that fails is thrown away and the previous day's stays, labelled with its own date, until it is too old to show. Texts are written on our server when the data refreshes, so pages stay static and no key reaches your browser.
Terms used across the site are explained in the glossary.