Large, mid and small cap cut-offs: methods

Code: pipeline/tipsheet/compute/size_cutoffs.py, pipeline/tipsheet/publish/size_cutoffs.py, step size_cutoffs (pipeline/tipsheet/steps/insights_size.py). Tests: pipeline/tests/test_size_cutoffs.py. Page: /funds/size/.

What we measure

How large a company has to be to count as a large, mid or small cap for Indian mutual funds, and how that has changed since the rules began.

• Cut-offs: the average market cap of the company ranked 100 (the large-cap line), 250 (the mid-cap line) and 500, at every half-yearly list, in rupees and in rupees of the latest list’s window.
• Old cut-offs today: the rank, in the latest list, of a company exactly the size of each past cut-off.
• Concentration: the share of the summed market cap of every ranked company held by the top 100, 250 and 500, and the number of companies above ₹1 lakh crore and ₹10,000 crore.
• Moves: companies in one category in a list and another in the next (small to mid, mid to large, and back), plus entrants to the top 100 and 250 including new listings.
• Survival: the share of the first list’s top 100 (and top 250) that are in the top 100 (250) of each later list, and the share still in the list at all.

The rule

SEBI’s circular of 6 October 2017 on categorising mutual fund schemes defines large caps as the 1st to 100th company by full market capitalisation, mid caps as the 101st to 250th and small caps as the 251st onward. AMFI publishes the ranked list twice a year. Each company’s full market cap is averaged over the six months to 30 June or 31 December, and across every exchange where it trades. The list appears in the first week of the following January or July; funds then have a month to realign their portfolios. Full market cap, not free float, is used.

Source

AMFI’s “Average market capitalisation of listed companies” list, collected by the Data bank as amfi_stock_categorisation. The first list covers July to December 2017; there are 18 lists to June 2026, each with 4,900 to 5,400 companies.

Calculations

• Cut-off: avg_mcap_all_exchanges_inr_cr at the printed rank. The rank is used as AMFI printed it because it is what binds funds.
• Real values: the all-India CPI (combined, linked series from macro_cpi_monthly) is averaged over each list’s own six months. A cut-off in latest-window rupees is the cut-off times the latest window’s average CPI divided by its own window’s. The step fails if any window lacks six months of CPI.
• Rank today: one plus the number of companies in the latest list whose average market cap exceeds the past cut-off.
• Matching across lists: by ISIN, then by NSE symbol. A company whose ISIN and symbol both change, for example after a merger or a scheme of arrangement, counts as unmatched rather than as a move. The Jul–Dec 2017 list ranks only 3,452 of its rows, so the first comparison (to Jan–Jun 2018) has about 1,500 unmatched companies, all below the small-cap line.

Source quirks

• Unranked rows. 1,795 rows in the Jul–Dec 2017 list have no rank (AMFI’s unranked tail). They are excluded everywhere.
• Ties in the tail. Companies with an average market cap of ₹1 crore or less can share a rank (Jan–Jun 2025 has two ties). The step requires ranks 1 to 500 to appear exactly once and allows ties below that.
• Out-of-order rows. Four rows in the Jan–Jun 2021 list sit slightly out of market-cap order (for example rank 221 is a little larger than rank 220). They are kept as printed. The step fails if more than ten rows in any list are out of order, which would mean ranks and caps had come apart.
• Checks. Every list must have exactly 100 large caps and 150 mid caps.

What this does not show

This page is descriptive. Whether a company’s price moves when it is promoted from small to mid cap, or mid to large, around AMFI’s publication date is a separate question. It needs an event study on the price panel, pre-registered before it is run (see docs/research/amfi_reclassification_spec.md once written). No such claim is made here.