Skip to main content
This page collects interview questions on ETF market making, arbitrage and Delta-1. The answers are concepts and frameworks. Market rules, such as Hong Kong creation and redemption or market maker obligations, are on the matching pages of this site and are not repeated here.
Status of this page. Last checked on 15 November 2026. These answers are general frameworks, not the rules of a specific market or any firm’s internal practice. For market details see ETF creation, redemption and arbitrage (Hong Kong) and ETF market maker rules (Hong Kong).

Fair value and arbitrage

The basic approach: value the basket at live constituent prices, add the cash component, divide by the number of ETF units it corresponds to, and convert to the ETF’s quote currency.The hard part is whether constituent prices are live:
  • When a constituent’s home market is closed, its price is stale. You can estimate the current fair price with related index futures, ADRs or FX.
  • For foreign-currency constituents you need FX, and FX may not be synchronized with the stock close.
  • When a constituent is halted or illiquid, the last price misleads.
So fair value is an estimate with an error band, and your quoted spread has to cover that uncertainty.
  • ETF at a premium (price above NAV): Buy the constituent basket, deliver it to the issuer for ETF units (creation), and sell the ETF in the market.
  • ETF at a discount (price below NAV): Buy the ETF, redeem it with the issuer for the constituents, and sell the constituents in the market.
The profit is the price gap minus all costs: constituent trading costs, creation and redemption fees, financing and borrow, hedging and execution risk. So the ETF price stays inside a no-arbitrage band rather than exactly at NAV.Without access to creation and redemption, you can still lock in the spread with futures or other hedges and wait for convergence. See Delta-1: hedging with listed derivatives.
Common reasons:
  • Costs and capital constraints: The gap is smaller than trading cost, funding cost and capital usage.
  • Borrow difficulty: The constituents cannot be borrowed for a short leg, or the borrow is expensive.
  • Time zones and closures: The underlying market is closed, so NAV itself is uncertain.
  • Capacity and process limits: Creation and redemption have minimum sizes, cut-off times and restrictions on who can participate.
  • Illiquid constituents: Trading the full basket has market impact.
A good interview point: separate a premium or discount that is only information (stale NAV) from a true mispricing. The first should not be treated as an opportunity.
  • Tracking difference: The cumulative gap between ETF return and index return over a period. It mainly reflects fees, dividend handling and cash drag.
  • Tracking error: The volatility of that gap, usually the annualized standard deviation of daily or weekly return differences.
The first measures how much the ETF lags on average, the second measures how unstable the gap is. Investors look at the first, risk management looks at the second.

Market making

The spread typically needs to cover:
  • Hedging cost: Trading cost and basis risk when hedging with futures or constituents.
  • Inventory risk: Price movement while holding an unhedged position.
  • Adverse selection: Counterparties may know more than you (for example a stale NAV, or futures that have already moved).
  • Fixed costs and taxes: Fees, stamp duty and similar. See Trading rules.
  • Profit target and quoting obligations: See Market maker rules.
In practice the spread adjusts to volatility, constituent liquidity and time of day. A wider spread lowers the chance of being picked off but also lowers volume.
Shift the quote midpoint against the inventory: when long, lower both bid and ask so selling is easier and buying is harder.The Avellaneda-Stoikov model gives the intuition with a reservation price r = s − q γ σ² (T − t), where q is inventory, γ is risk aversion and σ is volatility. The larger the inventory, the higher the volatility and the longer the remaining time, the larger the shift.This is a conceptual framework. In practice, if inventory can be hedged immediately with futures, its effective risk is lower than unhedged inventory, so the shift can be smaller.
It means the counterparty has an information advantage, and the price moves against you after the trade.Detection (at a concept level):
  • Analyze price movement shortly after fills (seconds to minutes), the post-trade markout. A systematic move against you indicates adverse selection.
  • Check whether fills happen after a futures or related instrument has already moved, which means your quote was stale.
  • Compare markouts by client type, time of day and order size.
Responses: widen the spread, reduce quote size, speed up quote updates, or pause quoting in specific conditions.
Number of contracts = ETF position market value times beta, divided by the contract value (futures price times contract multiplier). See Delta-1: hedging with listed derivatives.Residual risks:
  • Basis risk: The futures-spot spread moves and only converges at expiry.
  • Tracking and composition differences: The index the ETF tracks may not exactly match the futures underlying.
  • Dividends and timing: Dividends affect the futures and the ETF NAV differently.
  • Whole contracts: Contract counts are integers, so a residual remains.
  • Roll cost: Futures must be rolled before expiry.

Leveraged and inverse ETFs

Because leverage is reset daily, the long-run return is the compounded product of daily leveraged returns and is path dependent.In continuous-time approximation, the log growth rate of a daily-reset fund with leverage L is about L μ − (1/2) L² σ², while the index’s is μ − (1/2) σ². Compared with L times the index’s log return, the gap is −(1/2) L (L − 1) σ². For L = 2 the gap is about −σ². This is volatility decay.Implications: leveraged ETFs lose value in choppy, trendless, high-volatility markets, and in strongly one-directional trends they can exceed L times the index. An inverse ETF (L = −1) has drag −(1/2)(−1)(−2) σ² = −σ², so it also suffers.
To keep a constant L times exposure, the notional to trade before the close is about L (L − 1) × AUM × (index return that day).
  • Index up, L above 1: the fund must buy more of the underlying (chasing the rally).
  • Index down, L above 1: the fund must sell (selling into the decline).
  • Inverse ETF (L = −1): L (L − 1) = 2, so the flow is also in the same direction as the index. When the index rises, fund NAV falls and the fund must cover shorts (buy the underlying). When the index falls, it must add to shorts (sell the underlying).
This flow is concentrated near the close and can amplify end-of-day volatility. For a market maker it is a predictable directional demand and something to prepare hedging liquidity for. The actual size depends on that day’s AUM and on whether exposure is built with swaps or futures.

Delta-1 swaps

Basic structure: the client receives the underlying’s total return (price plus dividends) and pays a funding rate plus a spread, while the issuer holds the hedge. The price typically includes:
  • Benchmark rate and funding spread (reflecting cost of funds and capital usage)
  • Borrow cost, especially when the client is short
  • Dividend handling and tax (withholding tax and dividend tax rates differ by market and investor)
  • Hedging cost and liquidity
The specific numbers and terms depend on the counterparty and market. See OTC derivatives.