ETF Mechanics: Creation, Redemption, and What Prices Track

How creation and redemption keep ETF prices near NAV, why the gap widens under stress, tracking difference versus tracking error, and how SurgeFlow uses ETFs as benchmark and credit-market proxies.

Category: Portfolio

The exchange-traded wrapper, in plain language

An exchange-traded fund (ETF) is a pooled portfolio — often hundreds of stocks or bonds — wrapped so that the whole bundle trades on an exchange like a single share. That wrapper gives every ETF two prices at once. The first is the market price: whatever buyers and sellers agree on, tick by tick. The second is the net asset value (NAV): the per-share worth of the securities the fund actually holds, struck once a day from closing marks.

A traditional mutual fund only ever transacts at NAV. An ETF trades all day at the market price, and nothing forces the two numbers to be equal. What keeps them close is not a rule but a mechanism — an arbitrage loop built into the fund's plumbing. Most of the time the loop works so well the gap is invisible. Understanding when it works, and when it stretches, is the difference between reading an ETF quote correctly and over-reading it.

The mechanics: creation, redemption, and the arbitrage loop

Start with two accounting identities. NAV per share is NAV = (assets − liabilities) / shares_outstanding, where assets are the marked values of the fund's holdings. The premium or discount is d = (P − NAV) / NAV, where P is the exchange price; d > 0 is a premium, d < 0 a discount.

The key institutional detail is that shares_outstanding is not fixed. Large dealers called authorized participants (APs) can create new ETF shares by delivering a specified basket of the underlying securities to the issuer, or redeem ETF shares by handing them back and receiving the basket. This is the arbitrage loop. If the ETF trades rich (P > NAV), an AP buys the underlying basket, exchanges it for newly created ETF shares, and sells those shares at the premium — the new supply pushes P back toward NAV. If the ETF trades cheap, the AP runs it in reverse: buy ETF shares at the discount, redeem them for the basket, and sell the securities at their marks.

The loop does not force P = NAV; it enforces a band. Roughly, |P − NAV| ≤ c, where c is the all-in per-share cost of a creation or redemption round trip: trading the basket, fund fees, and the risk carried while hedging. In a liquid large-cap equity ETF, c is a handful of basis points, so gaps are tiny and short-lived. But c is a market condition, not a constant. When liquidity in the underlying market evaporates, the cost of trading the basket rises sharply, the band widens, and visible discounts can persist — not because arbitrage is broken, but because arbitrage has become expensive.

Two further terms are often conflated. Tracking difference is TD = R_etf − R_index over a stated horizon: the average return gap, sign included, driven by fees, sampling, withholding taxes, and securities-lending income. Tracking error is TE = stdev(R_etf − R_index): the volatility of that gap, not its size. A fund can lag its index by a steady, predictable amount (large TD, small TE) or hug it on average while wobbling daily (small TD, large TE). They answer different questions: what does holding this cost on average, versus how faithfully does it move day to day.

How SurgeFlow applies it

SurgeFlow is a research data site, and ETFs appear in it as instruments to read, not products to rate. They play two supporting roles. First, benchmarking: the portfolio views at /portfolio compare names against a per-market benchmark in each of the four covered markets, and continuously priced, exchange-traded index proxies are the observable series that make those relative comparisons possible. Second, credit proxies: the Bond board on the macro page (/macro?tab=credit) reads liquid US credit ETFs — investment-grade and high-yield series in the LQD and HYG mold — as real-time stand-ins for bond markets whose individual issues trade sparsely over the counter.

The honest boundary: SurgeFlow does not compute NAV premiums or discounts itself. It does not ingest NAV or iNAV feeds, and it does not model creation and redemption flows. It treats ETF exchange prices as observable proxies and takes them at face value. That means everything built on them inherits the premium/discount noise described above — and inherits it most exactly when markets are stressed, which is when a credit board is most interesting. As everywhere on the site: this is research infrastructure, not investment advice.

What SurgeFlow data this uses

The benchmark and proxy series behind these boards are daily price and turnover histories across the four covered markets — US, China, Japan and Hong Kong — with provenance that varies by market. One caveat no provenance can fix: an ETF price is a trade price, not a NAV. The gap between the two is part of the noise floor of any ETF-based reading, and SurgeFlow's boards carry that gap rather than correct it. - US: prices and profiles from FMP; the US feed also carries the US-listed credit ETF proxies. - China and Hong Kong: Tushare. - Japan: J-Quants for market data, EDINET for filings.

Deeper: when the ETF price knows more than the NAV

For an equity ETF on liquid large caps, the underlying quotes are fresh, so NAV is meaningful in near-real time and any price–NAV gap really is mispricing or noise. Bond ETFs invert this logic. Corporate bonds trade over the counter and infrequently; many issues do not trade at all on a given day, so NAV is struck partly from stale prices or matrix-based estimates. Meanwhile the ETF itself keeps trading every second the exchange is open.

Under stress this produces a striking picture: large, liquid bond ETFs swinging to visible discounts against their own NAVs, as in the widely documented episode of March 2020. One reading is that the ETF malfunctioned. The other — supported by subsequent research — is that the NAV was the stale number: the ETF price was performing price discovery, impounding information that the underlying bond marks had not yet caught up with. In that view the discount measures the staleness of the NAV as much as any defect in the fund.

Honesty requires saying that the two explanations — a stale NAV versus a genuine liquidity discount from strained arbitrage — are hard to separate in real time, and in stressed markets both usually contribute. For SurgeFlow's credit board the practical implication cuts both ways: ETF-based credit readings tend to move earlier and further than slow bond marks, which is a feature for timeliness and a caveat for level accuracy.

Explore it yourself

You can rebuild any of this from the raw series. A free (beta) API key is available at /account with an allowance of 2,000 requests per day, and the full endpoint surface is described by the OpenAPI specification at /api/public/openapi.json. Natural first experiments: pull a benchmark's daily price history and an ETF proxy for the same market and compute a rolling tracking difference; or line up credit-ETF drawdowns against equity benchmark drawdowns across the four markets. Treat everything as research data — inputs for your own analysis, not signals to act on.

Sources and further reading

Primary, academic, and SurgeFlow data references for the claims above: - Engle, R. and Sarkar, D. (2006), "Premiums-Discounts and Exchange Traded Funds", Journal of Derivatives. - Madhavan, A. and Sobczyk, A. (2016), "Price Dynamics and Liquidity of Exchange-Traded Funds", Journal of Investment Management. - Petajisto, A. (2017), "Inefficiencies in the Pricing of Exchange-Traded Funds", Financial Analysts Journal. - Ben-David, I., Franzoni, F. and Moussawi, R. (2018), "Do ETFs Increase Volatility?", Journal of Finance. - Laipply, S. and Madhavan, A. (2020), "Pricing and Liquidity of Fixed Income ETFs in the COVID-19 Virus Crisis of 2020", Journal of Index Investing. - Haddad, V., Moreira, A. and Muir, T. (2021), "When Selling Becomes Viral: Disruptions in Debt Markets in the COVID-19 Crisis and the Fed's Response", Review of Financial Studies. - SurgeFlow routes: /portfolio (market benchmarks), /macro?tab=credit (credit ETF proxies), /research/foundations-benchmarks-excess-return.