Explainers
Bitcoin ETF arbitrage, with the numbers
The arbitrage trade that keeps ETF shares glued to bitcoin's price — worked through with a real basket, the hedge, the profit, and the conditions that make it stop working.
Ask why a Bitcoin ETF share tracks bitcoin and you will usually get told "arbitrage". True, and unhelpfully abstract. Here is the trade itself, with a real basket size and real numbers, because once you have seen the arithmetic the whole product makes more sense — including the famous case where it was structurally impossible and a fund traded 40% below what it owned.
Why an arbitrage opportunity exists at all
An ETF share has two reference points that update at different speeds. Net asset value is struck once per business day against a defined reference rate. The share price moves every second the exchange is open. And bitcoin itself trades continuously, worldwide, including when the exchange is shut.
Three clocks running at three speeds guarantees the values drift apart. Nothing in the design prevents shares trading above or below the bitcoin behind them — no rule, no obligation, no regulator watching the tick. What closes the gap is that closing it is profitable.
The trade, worked through
Use IBIT's published figures from 2 September 2026: NAV $43.7352 per share, basket 10,000 shares, basket value $1,749,211.32, backed by 22.65 bitcoins.
Scenario: shares are trading at a premium.
- Spot the gap
Shares change hands at $43.7550 while NAV-implied value is $43.7352. That is a premium of about 4.5 basis points — roughly $198 across a 10,000-share basket.
- Buy the underlying
The AP acquires 22.65 BTC in the spot market, hedging its price exposure as it goes.
- Create the basket
It delivers the bitcoin to the trust and receives 10,000 newly issued shares. Under in-kind rules this is a direct asset transfer with no cash conversion.
- Sell the shares
It sells 10,000 shares at $43.7550, receiving $437,550 against a $437,352 asset cost. The gross spread is about $198.
- Net it down
Subtract spot execution cost, exchange and clearing fees, hedging cost and the capital charge. What survives is small — and that is the equilibrium.
Discount scenario, in reverse: the AP buys cheap shares on the exchange, redeems the basket, receives 22.65 BTC worth more than it paid, and sells the bitcoin. Its buying lifts the share price back toward value.
How the AP avoids taking a directional bet
A crucial detail: the arbitrageur does not want bitcoin exposure. It wants the spread. Between buying 22.65 BTC and selling the resulting shares, bitcoin could move 2% and wipe out a four-basis-point edge many times over.
So the position is hedged, typically by shorting CME bitcoin futures or by simultaneously selling ETF shares against the spot purchase, locking the spread and leaving only execution and basis risk. This is why a functioning regulated futures market matters so much to ETF quality — it is the hedging venue that makes tight quotes economical. It is also, incidentally, part of why the SEC's generic listing standards require the underlying commodity to trade on an established futures market. See crypto ETF regulation.
What the in-kind approval changed
| Aspect | Cash-only era | In-kind permitted |
|---|---|---|
| AP delivers | Cash; the trust then buys BTC | Bitcoin directly |
| Execution risk | Sits inside the fund | Sits with the AP, which is better equipped for it |
| Steps in the loop | More — cash leg plus a fund-level trade | Fewer — one asset transfer |
| Minimum profitable gap | Wider | Narrower |
| Observed spreads | Wider | Tighter |
| Fund tax efficiency | Lower — trust realises gains on sales | Higher — transfers avoid a sale |
The SEC's July 2025 order cited enhanced tax efficiency and minimised transaction costs among the benefits of in-kind processes.
You never see this as a shareholder. You see its consequences: a slightly tighter spread when you trade and slightly closer tracking when you hold. Both are real, and both are worth more to a frequent trader than a few basis points of fee. See fees and expense ratios.
When arbitrage breaks — the GBTC case
The cleanest demonstration of what this mechanism does is a fund where it was absent. Before January 2024, the Grayscale Bitcoin Trust could issue shares but had no redemption facility.
Follow the logic. Shares trade at a discount. An arbitrageur wants to buy them cheaply and exchange them for the bitcoin behind them, which is worth more. It cannot — there is no redemption. So the discount does not close. It widened past 40%: holders owned claims on bitcoin worth substantially more than the market would pay for their shares, with no route to the difference except selling into the same depressed market.
Conversion to an ETP created a redemption facility, the trade became possible, and the discount collapsed within weeks. That was arbitrage arriving, not sentiment improving. The whole episode is covered in the GBTC profile and NAV and premiums.
Four other conditions can strain the loop without breaking it: thin spot liquidity, custodian or settlement friction, a short list of APs, and the permanent weekend gap when bitcoin trades and the fund cannot. Those are covered in creation and redemption.
Trade the asset with no basket and no cut-off
Arbitrage exists because the fund and the asset trade on different clocks. Buying bitcoin directly means one clock, running always — CEX.IO is registered with FinCEN as a money services business, licensed for money transmission across US states, and authorised by the Gibraltar Financial Services Commission as a DLT provider.
Can a retail investor arbitrage a Bitcoin ETF?
Not the creation-redemption version. Baskets are 10,000 shares, over $1.7 million for IBIT, and they require an authorised participant agreement with the sponsor. That door is closed.
What retail traders sometimes attempt instead is cross-venue arbitrage: buying ETF shares while shorting bitcoin on an exchange, or trading between two funds on the same asset. Be realistic about the arithmetic. The gaps are one to two basis points, you pay a bid-ask spread on both legs, and you carry financing and, on any short leg, borrow cost. In almost every case the round-trip costs exceed the edge before you have started. It is not that retail arbitrage is prohibited — it is that professionals have already competed the margin down below your cost base.
The basis trade next door
A related trade deserves mentioning because it drives a lot of the flow people misread as directional conviction. When CME bitcoin futures trade above spot — contango — an institution can buy spot bitcoin exposure, often through an ETF, and simultaneously sell the futures contract, locking the premium regardless of which way bitcoin moves.
That trade generates genuine ETF inflows with no bullish view attached whatsoever. It is one of the main reasons daily flow figures are a poorer sentiment gauge than headlines suggest, and it is why we treat flow data carefully in flows explained and do ETF flows move the price.
Our take from the desk
The most useful thing arbitrage gives a retail investor is a verification system nobody had to design. Sophisticated firms have money at risk in checking, every day, that the bitcoin behind each share is really there and correctly priced. No disclosure regime achieves that with the same reliability. When we get asked whether these funds can be trusted to hold what they claim, that is the answer we find most persuasive — not the marketing, the incentive.