MSCI rebalancing day: what happens inside an ETF

Four times a year MSCI indexes change skin: stocks in, stocks out, weights revised. In episode two we saw who gets in and out, and through which filters; here we look at the day it all actually happens — the logistics of the operation, written in the MSCI handbooks we read in full, with the August 2026 review as the worked example, using MSCI’s own published numbers.

The short answer first: rebalancing day is not a day of chaos but the end of a scripted countdown. Changes are announced weeks ahead, weight adjustments are frozen over the last five days, everything takes effect in a single instant — the close of the month’s last business day — and for every mishap a rule already exists. ETF holders have nothing to do: the fund executes the move.

The most announced house move in finance

A rebalancing changes in a single day the basket that replicating funds — 14 in our registry for the MSCI World alone — must mirror, yet it surprises nobody: by design. MSCI’s methodology requires the outcome of every quarterly review (the “Index Review”) to be announced at least two weeks before it takes effect, and every corporate-event change to reach clients before implementation, every evening, in a dedicated file. The declared principle is to give the market “sufficient lead time to adjust to the information”: better an announced move than a surprise in live markets.

In the August 2026 review the two weeks became almost three: announcement on 12 August, effective at the close of 31 August. The only date nobody knows in advance remains the price snapshot — the “secret day” told in episode 2: for August it had been 20 July, and MSCI revealed it only with the announcement.

The countdown and the share freeze

As day X approaches, the machine reduces the moving parts. The most curious mechanism is the share freeze: in the five business days before each quarterly review’s effective date, float and share-count changes coming from placements, private placements and block sales are no longer implemented one by one — they are set aside, and all land together on review day. Mind the name: the stock keeps trading normally on the exchange — the freeze only concerns the numbers (shares outstanding and float) used to calculate the index. Why does it matter to an ETF holder? Every avoided trade is one less execution cost ending up in the tracking difference.

The reason is practical and stated: index-tracking funds are already preparing the review operation; making them adjust the same stock twice within days would be a pointless cost. “Market neutral” events — splits, stock dividends, mergers — are the exception and proceed normally even in those five days, because they require no buying or selling: numbers change, money does not.

Why everything happens at the close

Review changes take effect at the close of the last business day of February, May, August and November — not at the next day’s open. It sounds like a technicality and is not: the old index ends its day on the same closing prices from which the new index restarts. There is no moment when the index “does not exist”: the baton passes on a single price, the official end-of-day one — which, as we saw in episode 3, is the only kind of price MSCI accepts.

“Last business day” carries the precise global definition we met in episode 2: a day counts as a business day when markets weighing more than 80% of world capitalisation are open. And for markets that trade on weekends — some Middle Eastern exchanges — the rule is written down: Saturday or Sunday prices are carried to Monday.

The index never really sleeps

“Four appointments a year” does not mean the machine is off in between. Three examples from the methodology:

• size thresholds are maintained daily: MSCI recomputes the “interim” size-segment cutoffs every day, on the previous day’s data — they are used to classify on the fly the IPOs, mergers and spin-offs that arrive between reviews;
• sector changes (a company’s industry classification) not tied to events are announced twice a month and implemented together at the close of the month’s last US business day;
• new shares of companies already in the index identified by the month’s tenth business day enter on the third business day of the following month.

These are smaller calendars running underneath the main review cycle — and the reason replicating funds receive updates every evening, not four times a year.

If something goes wrong on day X

The most striking part of the handbooks is their detailed catalogue of things that can go wrong: it is written before anything does go wrong, with numeric thresholds in place of emergency judgment calls.

• A stock suspended the day before the effective date: its change slips to two full days after trading resumes; if the suspension lasts two months, the change is cancelled (for Chinese A-shares, additions are cancelled outright and reconsidered at the next review).
• An exchange closed for a national holiday on the effective date: the change slips to the next day, using the last available price.
• An unexpected full-day market closure: additions, deletions and weight changes with an impact of at least half a percentage point on the country index are postponed and implemented at the reopening.
• The extreme case: if on review day markets weighing more than 20% of world capitalisation are closed or illiquid, MSCI moves the entire review, for everyone. Below that threshold the rules above apply: only the affected market’s part slips, the rest of the world proceeds.
• Price limits (exchanges that halt stocks beyond a maximum daily move): if a stock is limit-locked at noon, additions and deletions slip by two days — but exits for acquisition, delisting or bankruptcy do not wait: the stock leaves at the price deemed most appropriate.

And for generalised market stress there is the “light rebalancing” already told in episode 2: when spreads and volatility cross written thresholds, buffers widen and only what is macroscopic moves. Introduced in March 2021, after the Covid lesson.

How much of a portfolio actually moves

MSCI measures each review with one-way turnover, and defines it in the most concrete way possible: “the proportion of assets that should be purchased in a portfolio replicating the index”. And since the fund’s assets are fixed, every euro bought corresponds to a euro sold: turnover counts the move once.

The August 2026 numbers, as declared by MSCI: 0.4% on the developed-markets universe from large caps to small, 0.7% on the whole world (large and mid), 4.7% on small caps alone — “both lower than in recent reviews”, the document notes. Translated with arithmetic: for every €10,000 replicating a developed-world index, rebalancing day moves a few tens of euros. The bulk of the portfolio stays put — and it is episode 2’s asymmetric buffer that keeps it put.

August 2026: a review seen live

Put the August 2026 review on a timeline, as MSCI documents it: prices photographed on 20 July (revealed after the fact), announcement on 12 August, effective at the close of 31 August, new index live from 1 September. Across the complete world universe 184 securities entered and 279 left; on the MSCI World, just 8 additions and 37 deletions, while the World Small Cap counted 114 and 98. The explanation is in the document itself: most moves are companies migrating between size segments or falling below the free-float requirement — not companies vanishing from the market.

And now let’s watch the machine work inside an ETF. We photographed the official basket of iShares Core MSCI World (SWDA) — Europe’s largest ETF on the index — on 20 August and on 31 August, the day the changes took effect, counting equity lines only. A comparison you will not find in any official document: the issuer publishes each day’s photograph of the basket, but not the comparison with the past — we are the ones photographing it at every reading, archiving the photographs and computing the differences. Data by iShares; readings, archive and comparison by Rebalix. The move, seen from within:

Two things stand out. First: SanDisk alone weighs almost as much as all 32 exits combined — and it is precisely the stock MSCI named in its review summary as the most notable case born of the methodology change to the Extreme Price Increase screen. Second: the exits are dust — 0.008% each on average; per €10,000 replicated, all 32 sales together are worth €26. The bulk of the portfolio stays put, as the turnover numbers above promised.

An honest note on the counts: SWDA’s 10 additions and 32 deletions do not match the index’s 8 and 37 one for one — the fund replicates by optimised sampling (it is not required to hold every single line) and the 20-31 August window also includes ordinary corporate events. Another live lesson: an ETF chases the index, it does not photocopy it line by line. The full comparison, stock by stock, sits in SWDA’s basket history, updated at every reading.

What it means if you hold an ETF

The operational answer is: nothing. An index rebalancing requires no action from ETF holders — the fund manager receives the list of changes in advance and executes the move. The ETF’s value does not change simply because one holding is replaced by another: what changes is the composition of the portfolio.

What rebalancing leaves behind, if anything, are the execution costs of the move — trading is not free. They appear on no fee line: they end up, along with everything else, in the gap between the fund’s return and the index’s — the tracking difference we measure year by year in our ETF pages. And that contained turnover is no accident: it is the effect of the buffers told in episode 2 — fewer changes in the index, less trading for the funds that track it, fewer costs ending up in the tracking difference.

In the fifth and final instalment we close the circle: why the same index has three official returns — Price, Gross and Net — and which of the three is the right yardstick for judging an ETF.