# SIF data - Specialized Investment Funds, India

Machine-readable twin of https://sif.tigzig.com. Everything on the page, written
for an agent that will not render it.

Generated 2026-09-05T06:16:24+00:00. Source: AMFI. Status: observational, see the bottom.

## What this is

Every NAV AMFI has published for Indian Specialized Investment Funds since the
category began, plus the scheme master and quarterly average AUM.

SEBI created SIFs as a category between mutual funds and portfolio management
services. AMFI's main mutual fund feed carries ZERO SIF schemes, so a SIF
identifier returns nothing from the usual mutual fund sources. That is why this
exists separately.

## Coverage

| | |
|---|---|
| Funds | 33 |
| Scheme variants | 121 |
| Fund houses | 17 |
| NAV observations | 12552 |
| Trading dates | 229 |
| Covering | 2025-10-08 to 2026-09-04 |
| Industry AAUM (Apr-Jun 2026) | Rs 13583.92 crore |

FUNDS AND SCHEME VARIANTS ARE DIFFERENT COUNTS. One fund offering Direct and
Regular plans across Growth and IDCW options is one fund and four scheme
variants. AMFI assigns a separate code and publishes a separate NAV per variant,
so the data holds one row per variant. Quote the fund count for "how many SIFs
exist". Reconciled against AMFI's own SIF Monthly Report: for June 2026 they
report 27 schemes and the funds live that month come to 27, matching in all five
investment strategies.

## Files

| URL | What |
|---|---|
| https://sif.tigzig.com/sif_nav.csv | every NAV, every scheme, every date |
| https://sif.tigzig.com/sif_meta.csv | scheme master, one row per variant |
| https://sif.tigzig.com/sif_aaum.csv | quarterly average AUM per scheme |
| https://sif.tigzig.com/sif_nav.parquet | the same NAV file, typed and columnar |
| https://sif.tigzig.com/sif_meta.parquet | the same scheme master |
| https://sif.tigzig.com/sif_aaum.parquet | the same AUM file |
| https://sif.tigzig.com/manifest.json | row counts, bytes, SHA-256, generated_at |

Every data file is published as both CSV and Parquet, from the same query, so the
two always agree on content. Parquet is around 95% smaller on the NAV file and
carries real types, so a date arrives as a date and a scheme code cannot be read
as a number by accident. One deliberate difference: a missing value is a typed
NULL in Parquet and an empty string in CSV, because CSV cannot express the
distinction. Read either with

    SELECT * FROM 'https://sif.tigzig.com/sif_nav.parquet'
| https://sif.tigzig.com/summary.json | precomputed counts and series |
| https://sif.tigzig.com/README.txt | every column explained in full |

Read `generated_at` in manifest.json for freshness. Each file carries its SHA-256
there.

## Pipeline update status

Each row is measured from its own table, so a dataset that has not been refreshed
does not borrow freshness from one that has. Timestamps are UTC, ISO 8601.

| Pipeline | Last updated | Cadence | Rows | Detail |
|---|---|---|---|---|
| NAV history | 2026-09-05T06:16:23.764166 | twice daily | 12,552 rows | 121 scheme variants, 2025-10-08 to 2026-09-04 |
| Scheme master | 2026-09-05T06:16:24.490826 | twice daily | 121 schemes | 33 funds across 17 fund houses |
| Quarterly AUM | 2026-08-31T07:41:39.702245 | weekly | 173 rows | 3 quarters, AMFI publishes this once a quarter |

## Refresh

Refreshed **twice every day**, at about 05:05 and 11:45 IST (23:35 and 06:15 UTC).
The quarterly AUM file is rebuilt weekly, since AMFI publishes it once a quarter.

AMCs typically file their NAVs with AMFI through the evening and into the following
morning, and sometimes later than that. We refresh more than once a day so these
files pick up the most recent filing rather than a single snapshot. The practical
consequence for an agent: a date that is only a day or two old can still gain rows,
so re-fetching a recent window is worth doing. `generated_at` in manifest.json is
the exact age of the copy you are holding.

## Column names match the mutual fund dataset

These files deliberately use the same column names, in the same order, as the
TIGZIG mutual fund scheme master at
https://api.tigzig.com/mf/v1/download?format=latest. Code that parses that file
works on these unchanged. SIF-only columns are appended after the shared ones.

`scheme_code` is the numeric part of the SIF code: SIF-120 is 120. Every mutual
fund scheme code is 100,027 or higher, so the two cannot collide if you stack
them. `source` says which dataset a row came from.

COLUMNS PREFIXED `temp_` WILL NOT SURVIVE the merge into the main scheme master.
Do not build on them.

## Fields worth knowing before you use them

- `isin` and `isin2` are two identifiers for ONE scheme, not two schemes. isin2 is
  the IDCW reinvestment ISIN.
- `temp_plan_basis` says whether the plan was filed by AMFI (`stated`) or inferred
  by us from sibling schemes (`sibling_complement`). Filter on `stated` for filed
  data only.
- `is_active` is "true" when the scheme has a NAV within 45 days. Same rule and
  same lower-case spelling as the mutual fund dataset.
- `aaum_lakhs` is in Rs lakhs as AMFI publishes it. `aaum_cr` is that over 100.
- `upload_ts` is AMFI's publication timestamp, stored as filed, which means it is
  IST despite the trailing Z. Anything computing lag from it is 5h30m out unless
  it converts.
- `first_date` is the first NAV we hold, which is not necessarily inception.

## Known oddities in the source

Published as filed. Nothing cleaned, smoothed or corrected. Full list with
confirmation detail: https://sif.tigzig.com/#anomalies and
https://sif.tigzig.com/anomalies.json

Summary: four IDCW options stopped reporting for stretches; four NAVs filed as
zero; one AMC files to two decimals where others use four; one scheme has AAUM
filed but no NAV ever published; six schemes publish NAVs but are absent from the
AAUM filing; AMFI's upload timestamp is IST labelled as UTC. All confirmed
against the archived source bytes and all under review.

## Status and terms

OBSERVATIONAL. This is part of the beta testing of the data pipeline behind
https://www.tigzig.com/mfpro, run separately so issues surface before it is
integrated. No uptime commitment and the shape of the files may change. If you
build something on it that you rely on, keep your own copy.

A third-party copy of AMFI data. Not affiliated with or endorsed by AMFI. Refer to
amfiindia.com for the official data.

Contact: amar@harolikar.com
More TIGZIG tools and data: https://www.tigzig.com/llms.txt