Four provisions do most of the work, and reading them takes about twenty minutes per index.
Which shares count, and who may enter?
Free float and the investible weight factor decide which shares count. Promoter holdings, strategic corporate stakes, government holdings and locked-in shares are excluded, and the surviving fraction sets the constituent's weight.
What matters to a tracker is not the definition but the revision schedule. A float revision is a forced trade with no price move behind it, so a provider that revises quarterly imposes four such events a year and one that revises annually imposes one.
Eligibility screens decide who may enter: size, liquidity, listing history, domicile and sometimes profitability. The liquidity screens are the ones a tracker feels, because a screen that admits thinly traded names admits expensive ones. An impact-cost threshold stated for a specific order size, which is the form the Indian screens take, is a direct statement about what a large fund will pay to hold the index.
Advertisement
How do buffer zones and review dates work?
Buffer zones break the churn a pure rank rule produces every time two names near the boundary swap places. A non-constituent must rank meaningfully higher than the cutoff to enter, a constituent must fall meaningfully below it to leave, and between the two thresholds nothing happens. That costs index purity and saves turnover. Worked example 1 computes both sides of the exchange.
Review dates and notice periods are friendly to a large tracker when the calendar is published and the notice is real, because the fund can work its orders over days rather than minutes. The same calendar is an open invitation to anyone who wants to trade ahead of forced flow. Providers know it, which is why notice periods are short, why some indices reconstitute as needed rather than on a calendar, and why several large funds now trade part of their reconstitution away from the closing auction.
Read all four before you compare two trackers on the same nominal index. Two funds tracking the same benchmark can have materially different costs purely because one of them is large enough to trade the reconstitution over a week and the other has to take the close.