Author: 798.eth

The Bound exchange, which has been generating a lot of buzz in the Bitcoin ecosystem lately, was formerly known as RadFi. It's a platform I started using at the end of 2025 due to its NodeStrategy monkey strategy tokens. Now upgraded and renamed Bound Exchange, it features a unique asset: the monkey strategy itself—a flywheel of node monkeys backed by underlying assets. The key is whether these underlying assets can sustain the exchange.

Today I'll mainly talk about its asset deposit, withdrawal, and trading logic. But before delving into the mechanisms, I'll first explain the differences between the upgraded Bound and the original Radfi.

I. Past and Present: From Radfi to Bound

Bound didn't just appear out of thin air; it's a complete rewrite of radFi. The radFi website now displays the statement, "radFi has evolved into bound.exchange." It's the same team, but the technology stack was completely rewritten, changing from Create-React-App to Next.js.

Bound's real additions are a built-in multi-chain wallet for the passkey, term loans using Bitcoin as collateral to borrow bUSD, and SODAX cross-chain functionality. The trading engine remains the same concentrated liquidity AMM from radFi, and the Runes swap remains unchanged.

This product line has two 2-of-2 schemes with timelock. The account-level scheme protects you, allowing you to circumvent the bound after three months; the lending scheme protects the lender, allowing them to take your collateral directly after the grace period. (See the diagram for a detailed comparison of radFi → Bound).

II. Core Mechanism: Asset Deposit and Withdrawal and 2-of-2 Multi-Signature Custody

Having understood its positioning, let's look at its asset management logic. There are two main points: first, 2 of 2 multi-signature custody; second, timelock countdown protection in the event of platform collapse or downtime.

Let's set up the scenario first. You have BTC and want to trade it on a platform. The traditional approach is to deposit your coins into the platform's wallet, and the platform holds them for you. The problem is that the coins are no longer in your hands. If the platform disappears or is hacked, the coins are gone too. That's how FTX and Celsius went. Bound aims to solve this contradiction: it wants the trading speed of a platform-level exchange (CEX) but doesn't want to relinquish control of the coins.

The first step is to understand what a deposit is.

After you register, Bound gives you a Bitcoin address, starting with "bc1p". To make a transaction, you first transfer your BTC from your own wallet to this address. A typical Bitcoin transfer takes about half an hour with three confirmations. Up to this point, it's no different from using a regular wallet. The real design is hidden in the next question.

The second step is to find out who has the private key for this address.

This is the core of the whole system. There are only two cumbersome methods: either the platform keeps the private key (which is escrow, and goes back to FTX's old problem); or you keep the private key, in which case the platform can't help you complete transactions quickly, and you have to manually sign each transaction yourself, which is slow.

The solution to the bound is to assign two keys to the address, not just one. To access the funds within, both keys must be used for signature. This is 2-of-2 multisigning.

The third step is to find out whose two keys they are.

One key is yours, stored on your own device, accessed via your passkey; the private key never leaves the device. The other key resides in the Bound backend. During daily transactions, you sign with your passkey, and the Bound backend automatically generates the other key. Once both keys are combined, the transaction is completed in seconds.

Take a moment to reflect on this step. Bound only has one party, so it can't unilaterally access your coins. Therefore, it's not escrow and can't use your coins for misappropriation or lending. However, when you two combine your funds, the transaction can be completed instantly, restoring speed. This is the desired effect of 2-of-2: achieving both security and speed.

The fourth step is to determine which hole the timelock is used to fill.

The 2-of-2 approach introduces a new problem. Since two keys are required, what if Bound runs away, crashes, or is shut down? The backend key will be permanently lost, leaving you with only one key. Does that mean your coins are locked in the address?

Timelock is designed to close this loophole. The rules are simple: the bound token has a 3-month validity period. Within 3 months, you need to withdraw the tokens twice. After 3 months, you can withdraw all the tokens in a single transaction without needing a bound token anymore.

The key point is that this isn't just a verbal promise from Bound; it's written into the Bitcoin script and enforced by the Bitcoin network. Bound's agreement is irrelevant; once the time is up, you can sign off and leave. This is the source of its confidence in advocating self-custodial.

Step 5, withdrawing now goes smoothly.

For normal withdrawals, you need to use your passkey and Bound backend to sign together; both are required for instant withdrawal. For emergency withdrawals, if Bound is unavailable, wait until the 3-month timelock period is reached, then use your own single signature to scan the coins to any address.

If you look at these five steps together, the first four words form a chain: you deposit coins into a 2-of-2 address, this is deposit; you spend them together with Bound in daily transactions, this is the fast path of withdraw; the Bound transaction expires in 3 months, this is timelock; after it expires, you can leave in one transaction, this is the escape route of withdraw.

III. Underlying Logic: Boundex Transaction Ranking

Now that we understand how to store assets, let's talk about Boundex's transaction sorting.

There are no smart contracts on Bitcoin. So where exactly does the AMM (Agent Matcher) on a native Bitcoin DEX match your transactions, and who decides your order's order number? Essentially, it's a centralized exchange's (CEX) matching backend, connected to the Bitcoin blockchain as the settlement layer. Let's explain step by step:

First, the AMM curve is not on-chain . BTC L1 has no contract, so there's nowhere to store states like reserves, ticks, and liquidity intervals; they can only be stored in the Bound backend database. The on-chain UTXOs are simply the result of managing and settling LP funds.
Second, on-chain transactions are merely settlement receipts . I pulled the transaction records from that pool, and every inflow and outflow is a fixed, hard-coded amount, containing no curve calculations. The multiplication along the curve was completed in the background before being recorded on the chain. A block contains over twenty transactions, each using its own UTXO, completely independent of each other. This is because the price was fixed off-chain; recording it on-chain simply involves recording a set of calculated numbers.
Third, how is the price determined for a single transaction? When you want to buy or sell on the front end, the front end requests a price quote from the back end . The back end uses the concentrated liquidity curve it maintains to calculate the price. Your transaction consumes liquidity along the tick, deducts fees, and gives you a number. You sign the PSBT (Purchase and Transaction Token) and specify this number. The back end verifies the information and broadcasts it, simultaneously pushing its own curve up one notch. The curve only exists in the back end.
Fourth, the ordering is a mathematical necessity imposed by this curve, and there's no circumventing it. The curve is stateful; at any given moment, it only has one current position. Each new transaction consumes a segment of liquidity, pushing the curve to a new position, and the next transaction can only be executed at that new position. This means that all transactions cannot be settled simultaneously; they must be arranged in a strict sequence. The first transaction executed at its original position pushes the curve one space, the second transaction at the new position pushes it another space, and so on.
Fifth, however, who determines the order and by what criteria is another matter. Here, we need to distinguish between two orders. One is the actual time you confirm, and the other is the processing order used by the backend to push the curve. Neither the blockchain nor the protocol guarantees consistency between these two. Your order enters the Bound backend queue; whether it declares it, whether it processes orders according to arrival time, and whether it pushes orders forward or backward are all internal matters, which you cannot see.
Sixth, this is the entry point for the operator's MEV, even more crucial than ordinary on-chain MEVs. Generally, the order of transactions on a DEX is at least recorded in the block; whoever precedes the other is caught in the middle, and the order can be reviewed afterward. Bound, however, occurs off-chain, leaving no trace of the order after the block is written. Over twenty transactions are unconnected, and the order is impossible to discern on the chain. But at the moment you confirm, the price has already been calculated based on a point on the curve. You just don't know the exact position of the transaction within that block.

Therefore, Bound isn't without order; the order is mathematically mandated and inherently present. It simply hides the power to define this order entirely in a black box. The principal layer is indeed self-custodied, using a 2-of-2 structure with a time lock, meaning the platform can't touch your coins. However, the price and order of your transaction are recorded off-chain by the platform alone. Ultimately, it's essentially a centralized exchange's matching backend, connected to the world's largest blockchain for settlement, with a block issued approximately every ten minutes, and confirmation is irreversible.

IV. Practical Tips and Feedback

Finally, here are two practical details to avoid pitfalls. This 2-of-2 address only accepts BTC and Runes; sending BRC-20, Alkanes, or other chain coins will result in permanent loss. Alternatively, you can bypass Bound and connect directly to your own Unisat or Xverse. In that case, there's no 2-of-2 or timelock, and the coins remain in your single-signature wallet. The trade-off is that you have to manually sign every transaction.

Currently, there are still some minor issues with the system. I tested it, and although the previous transaction has been confirmed two blocks ago, the Bound Exchange frontend still shows "pending," preventing me from placing the next order. I hope the official Bound Exchange team can look into this issue.

Coinbase will launch CHECK and SIGN spot trading pairs.

Popular Articles