The data behind every page
Every chart and table on tipsheet is drawn from a file you can download: 3,055 datasets, with market data to 20 Nov 2026. They are plain JSON, free to fetch from any program or browser, and refreshed every trading evening.
Where to start
/data/v1/manifest.json lists every dataset by name, with the address of its current file. Names are stable (activity/exchanges/monthly); file addresses carry a hash of their contents (activity/exchanges/monthly.35025b80b188.json) and change when the data does, so a file never changes once published and can be cached forever. Read the manifest, then fetch the files it names.
/data/readings.json lists every reading with today's value, its date, what it means in a sentence and where its history lives.
/data/v1/status.json says when each source was last updated and whether the latest refresh succeeded. The machine-readable description of all of this is /openapi.json (OpenAPI 3.1).
Both files are fresh for a minute; dataset files are cached for a year. Every file allows cross-origin requests, so a page on another site can read them directly.
Three shapes
Series (example): dates is a list of ISO dates, and series maps each column name to a list of values of the same length, with null where there is no value. resolution is the frequency (daily, weekly, monthly, quarterly), and meta carries the source, the method and notes on units.
Statistic (example): one reading against its own history. latest and as_of are today's value and date, value_ago the value a week, a month and a year earlier, and history its start, lowest, highest, median and percentiles.
Table: columns names the fields and rows holds one list per row, in the same order.
A few datasets (portfolio results, trend studies) hold several of these together under named keys; their meta says what each part is.
Every reading as CSV
Each of the 65 readings has a permanent page at /i/<slug>/ and its full history as CSV at /i/<slug>/data.csv, for example /i/nifty-50-pe/data.csv. Every indicator lists them.
For language models and agents
Every page has a Markdown twin with its text, tables, sources and a ready citation: add index.md to the page's address, as in /valuation/index.md. /llms.txt maps the site and /llms-full.txt holds every reading in full.
What is published
| Family | Datasets |
|---|---|
etf/ | 426 |
indices/ | 371 |
lab/ | 315 |
trend/ | 276 |
macro/ | 257 |
models/ | 223 |
funds/ | 179 |
economy/ | 149 |
credit/ | 126 |
companies/ | 85 |
pulse/ | 71 |
directories/ | 61 |
valuation/ | 58 |
leverage/ | 53 |
activity/ | 46 |
markets/ | 46 |
bonds/ | 45 |
depositories/ | 43 |
rbi/ | 29 |
barometers/ | 27 |
flows/ | 24 |
ipo/ | 22 |
ownership/ | 17 |
business/ | 16 |
fpi/ | 16 |
capital/ | 14 |
world/ | 14 |
derivatives/ | 12 |
evidence/ | 9 |
factors/ | 7 |
fo/ | 7 |
dossiers/ | 6 |
currencies/ | 3 |
commodities/ | 1 |
events/ | 1 |
Terms
The files are free to fetch. Please cite tipsheet and the original source named in each file's meta: tipsheet, "<dataset or page title>", data to <date>, <URL>. Values are derived statistics; raw index levels and licensed feeds are not published. Upstream sources keep their own terms. This is historical research, not investment advice.