An arrow pointing to the top left



Index Tokens - Past, Present, and Future

Co-founder & CTO

Index Tokens - Past, Present, and Future

Special thanks to Dan Boneh, Tim Beiko, Max Bronstein, Riyaz Faizullabhoy, and Sam Ragsdale for feedback.

We’re very excited to announce AMKT, a benchmark of the entire crypto market and the first low-fee index token that tracks that benchmark. We want to not only enable people to get easy, diversified exposure to crypto but to get that exposure through an ERC-20 token so they can enjoy all the benefits that come from DeFi.

We think the future of finance will be built on crypto because of its ability to minimize requirements of trust. In the spirit of maximal transparency, this post outlines the trade-offs we faced when confronted with the task of making a useful index token and how we can eventually move to architectures with fewer trust assumptions.

What makes a good index token?

Before comparing architectures, we need to share what our design criteria were. You may not require all of these properties of an index token but we think a token without all of these properties could not succeed in bringing sane diversification to a significant portion of the crypto market.

🏃 Broad Asset Support – The index token must safely give the holder exposure to as many of the top crypto assets as possible. Otherwise, prospective holders would face the same issue they face today: needing to pick ecosystems or actively rebalance positions across many chains.

🔬 Low Net Asset Value (NAV) <> Market price spreads – This refers to the price the token trades at when compared to the price of all of its constituents. Ideally, they should be equal. While no index-tracking token can be perfect in this regard, some mechanisms have persistent problems keeping this spread small.

⏫ Scalable – The supply of the index token must be uncapped. If there are soft or hard upper bounds on the supply, many potential holders will not have access to the token. It also makes the previous property more difficult to achieve because large, persistent price premiums can emerge if the supply is capped.

🌲 Low & fixed fees – Indexes are, first and foremost, buy-and-forget assets. Holders shouldn’t have to worry about variable fees to determine if holding the token is still of interest to them.

tl;dr before we get in the weeds

Given the tech and resources available to us today, the criteria above can only be satisfied by using a custodian to hold the large variety of underlying assets and creating associated tokens that track those assets.

Tokens are created and destroyed by addresses with a special role called “merchants”. They provide the underlying constituents of the index to a custodian, Coinbase Prime, in return for index tokens. They can also destroy index tokens to receive the corresponding quantities of the index’s constituents. In this manner, every index token in circulation is always backed 1:1 and in-kind.

An overview of the Alongside system.

This structure is reinforced by the use of Coinbase Prime, which holds customers assets on a 1:1 basis and, by electing Article 8 of the UCC, is required to maintain the custodied assets in a manner that satisfies customers’ claims to those assets. In other words, they have very strong disincentives against undirected use of assets.

For increased transparency, we’ve published a dashboard where you can monitor the state of the assets backing the index. In the coming weeks, there will also be a Chainlink oracle attesting to the same so you don't need to trust us.

The remaining sections are for the technical reader who wants to know exactly what we thought about different mechanisms and what our current view is on the road to more trustless index-tokens.

Options Considered

Collateralized Debt Positions

Collateralized Debt Positions (CDPs) are positions that track an oracle by allowing mints and redemptions of a token against some overcollateralizing set of underlying assets. MakerDAO is the best-known CDP protocol, over-collateralizing more than $5B of Dai against various assets.

The original architecture for AMKT

We spent months designing and attempting to improve on existing CDP designs by taking advantage of the fact that some of the value of the token can be backed 1-to-1 because the collateral is natively available on Ethereum. Only the value of the assets that are not available on Ethereum need be overcollateralized.

Ultimately, we discovered the same thing Hasu did in his amazing post about DAI’s scalability limitations. Namely, capitally-inefficient and risky arbitrage would cause the token to have a persistent price premium. There would also be a supply soft cap as demand for minting the token (effectively taking a short position) saturates.

Additionally, we feared how the system would behave under periods of extreme volatility. To deal with this, MakerDAO has used governance token auctions to raise collateral when the ecosystem of liquidators was not prepared for the ETH price crash in March 2020. We’re not as established as Maker, however, so any governance token we issued at this time would be very speculative and would likely fail to hold its value exactly when we needed it to. Even if it didn’t, we’re in this for the long-term and don’t want to create tokens until they have clear value propositions.

You can view the full paper here. Note that the spec likely has conceptual flaws and is not ready for implementation. However, we’d love to see alternatives to our product out in the wild so if anyone wants help running with it, feel free to DM me! If you want to buy a similar index token with all the benefits/drawbacks of this mechanism, check out TCAP.


Perpetual futures are an oracle-tracking derivative constructed via periodic funding rates between long and short traders.

We discarded perpetuals because these funding rates vary and can be large at times. This violates our requirement for an index token with low, fixed fees.

They are much better suited for shorting and actively managing leveraged positions. That said, we plan to have perpetuals that track the AMKT benchmark in the future for exactly the cases where they make sense!

Bridged Constituents

An interesting alternative is to bridge all the index’s constituent assets over to Ethereum and then use something like Set protocol to bundle them together into an index token. The biggest issue with this approach is the massive amount of both engineering effort and smart contract risk. The number of bridges we’d have to develop and maintain given our goal to support the top 25 crypto assets on an ongoing basis was far too high. As we’ve seen in recent years, keeping even one bridge safe is hard enough.

1:1 Backed with Custodian

Holding all the assets in a custodian and issuing redeemable tokens satisfy all our criteria for a good index token.

  1. Custodians have support for the vast majority of the assets we need by market cap.
  2. There are no limits to how many of each asset can be custodied, which means the supply on-chain representing these assets is also uncapped.
  3. Fees can be implemented as low, steady, inflation of the index token.
  4. The arbitrage loop is low risk, which results in short-lasting price premia/discounts relative to NAV.

The drawbacks are the trust assumptions. There are two groups that need to be trusted.

  1. The custodian – They need to be trusted to be willing and able to process any withdrawals from the account (some delay to account for their withdrawal SLAs is acceptable). If they are rehypothecating assets they could be unable to process withdrawals due to liquidity constraints. If they want to abscond with the private keys, they can.
  2. Those with the right to withdraw assets from the custodian – They need to be trusted to not withdraw assets from the account for any reason other than to process redemptions or perform OTC trades for the purposes of rebalancing/reconstituting the index.

To fill these roles we have

  1. Coinbase Prime. This structure is reinforced by the use of Coinbase Prime, which holds customers assets on a 1:1 basis and, by electing Article 8 of the UCC, is required to maintain the custodied assets in a manner that satisfies customers’ claims to those assets. In other words, they have very strong disincentives against undirected use of assets.
  2. A group of approvers composed of some members of Battery Labs. We’re keeping their identities private but we can share that 3 people are required to move funds (with larger amounts requiring video confirmation), and the approvers are located across multiple continents. They convene to approve transactions approximately daily.

For increased transparency, we’ve published a dashboard where you can monitor the state of the assets backing the index. In the coming weeks, there will also be a Chainlink oracle attesting to the same so you don't need to trust us.

Given the above, we strongly believe the custodian-backed architecture to be the best to allow for widespread, passive exposure to crypto today. Next, I’ll go over our view of what it might take to build a protocol that is as trustless as possible.

Future Directions

The spectrum of problems we see and the relative difficulties in decreasing their trust assumptions.

Smart Contract Non-upgradeability

While smart contracts can be deployed without upgradeability, in practice many projects, ourselves included, want to retain the ability for upgrades to provide users with improvements. This requires trusting those with the ability to upgrade the contracts. This risk is often taken for granted but we think it should be explicitly acknowledged.

To fill the admin role we have the AMKT Governance Stewards – a group from different organizations chosen for their integrity and their understanding of the promise of crypto/DeFi. The multisig is 4/8 at launch with 4 people external to Alongside.

Empowering the Governance Stewards with the ability to upgrade the smart contracts is necessary to support trust-minimizing iterations on the protocol.

Remove Custodian Trust Assumption

Right now index token holders need to trust both the users of the custodian account and the custodian themselves. By migrating asset custody from a system that holds private keys directly to one that simply participates in secure multi-party computation signing, we can remove the custodian’s ability to unilaterally direct the index’s underlying assets.

The Chasm

The Chasm is the set of problems that we have to deal with because broad-market exposure necessitates supporting assets on many different chains. This could go one of two ways:

  1. All economic activity is rooted in a single base layer for security (multi-chain)
  2. Economic activity is fragmented between different base layers (cross-chain)

We will need to be able to bridge assets in either case but we hope to see the first situation come to pass because of the additional problems with cross-chain bridges when compared to multi-chain bridges.

The technologies I mention here point to how these problems could be accomplished cryptographically. However, where there are difficult and robust cryptographic solutions, there are usually easier yet less robust cryptoeconomic solutions that leverage incentive engineering. The cryptoeconomic solutions have more degrees of freedom, however, so this is one of those cases where predicting a long-term end-point is actually easier than predicting the medium-term. Whether or not future versions of the protocol will leverage cryptoeconomic solutions will depend on the robustness of those solutions and the availability of practical cryptographic approaches.

🧊 Trustless Minting – This could be the first thing to fall into place thanks to advances in practical zk-based cross-chain messaging (like zkBridge and Succinct) and increasing representation of light-client friendly protocols. Anyone should be able to deposit constituent assets on the required chains and provide the proof of deposit to the smart contract that mints the token.

🔥 Trustless Burning – Burning the index token to receive its constituents back, however, is more difficult. An example cause of this is Bitcoin’s limited scripting functionality. There’s no way yet to verify a proof that you burned an index token directly on Bitcoin to receive the underlying bitcoin back in a manner that’s as seamless as in the minting case.

There could be some mechanisms that use atomic swaps but even so we haven’t found a way around the significant friction it would add to transferring the index token. One option could be to enable people to opt-in to burning based on a multi-sig, in case they are unwilling to accept the friction incurred by trustless burning of Bitcoin.

⚖️️ Trustless Rebalancing – This requires secure message passing like the above two and additionally, thoughtful design of what needs to be proven on the source chain to allow a cross-chain trade to occur.

To remove privilege from a small group of traders, it could be possible to prove that certain triggers (like deviation between target weights and ideal weights in the index) are present and add incentives to create a keeper network of rebalancers. One way to think about this is like a cross-chain, non-fixed-curve, Balancer pool.


These three are all difficult but theoretically solvable, even if the applied cryptography hasn’t caught up to the theory yet. We’ll work with the cryptography community to pull this future forward. Once the applied cryptography is ready, we’ll also need to give the message-passing infrastructure time to get very lindy since the smart contract risk at the outset will be substantial.

The end result will be a trustless way to bundle/unbundle assets across the most popular chains.

Asset Inclusion Governance

We think index construction – choosing themes and interpreting assets for fit against those themes – will continue to involve value-aligned individuals participating in governance for a long time. With thoughtful governance design, this can be decentralized but probably not made completely trustless, which is why it’s at the end of the spectrum of difficulty.

We took inspiration from the Optimism Collective and chose an iterative approach with a working constitution and a high-integrity set of initial Stewards.


In 1960, Edward F. Renshaw and Paul J. Feldstein put forward the idea of an “Unmanaged Investment Company”. Since the genesis of the idea, index products have reached their massive level of adoption after a 60+ year slog. What we are embarking on, an attempt to create a borderless, interoperable, trust-minimized successor to what they described, will be difficult and take time. But what seems like ages in crypto years will be a short chapter in the long arc of passive investing.

Buy the entire crypto market in a single token.