The algo stablecoin landscape

Published in

Malt Protocol

12 min readMay 12, 2021

The beautiful landscape of algorithmic stablecoins

Having a stable asset for investors to move into and account against is of critical importance to any mature finance ecosystem. This is just as true in crypto and defi than anywhere else. It is likely that at some point investors and users of crypto will want to get out of a potentially volatile asset e.g. ETH to realise profits, balance portfolio risk or simply as a medium of exchange with a known value over time. However, it is also likely that those people will also want to be able to stay inside the crypto ecosystem and not have to rely on third parties.

This naturally leads to stablecoins popping up inside crypto to serve this purpose. However, up to this point, stablecoins have largely relied on the backing of a centralized custodian with a form of mint/redeem mechanism to maintain the peg of the coin. The use of such a centralized custodian is entirely antithetical to the ethos of defi, and of crypto more broadly — we are trying to decentralize the financial system, so reliance on a centralized custodian is contradictory to our vision.

The emergence of “crypto native” stablecoins like DAI improved the situation but ran into issues of capital inefficiency due to the requirement for over-collateralizing the positions. It comes as no surprise that the next logical progression would be attempts at creating capital efficient stablecoins.

These attempts at capital efficient stablecoins are known as algorithmic stablecoins due to the fact that instead of relying on the value of the capital collateralizing them to maintain their peg, they rely on a protocol of user incentives that is baked into code — hence, “algorithmic”.

If you have 100% collateral backing a stablecoin, maintaining peg is simple. You can just allow users to mint new coins by depositing collateral and redeem the collateral by giving back the coins. This mint/redeem mechanism creates an arbitrage any time the floating price of the coin deviates from the peg. Aggressive competition for profits will always seize that arbitrage opportunity quickly and close the price gap.

On the quest for capital efficiency you lose the possibility of having a mint/redeem scheme as there are insufficient funds for redemption. This makes things more complex and the protocol must align incentives to encourage users to take actions that bring the price back to peg when it deviates away.

In the most general sense any algorithmic stablecoin has to consistently achieve the following to actually maintain stability (let’s call them the critical criteria):

Create selling when the coin is above peg to push the price down.
Create buying when the coin is below peg to bring the price up.

Of course there are many other things they can do to help promote stability, to smooth things out, create protocol longevity or more long term value. However, maintaining stability is impossible without meeting these two critical criteria.

Here are some attempts at algorithmic stablecoins and the mechanisms they used to achieve the critical criteria.

Ampleforth

Above peg supply is expanded and distributed to every user’s wallet proportional to their ownership of the total supply. This addition of extra coins incentivizes short term traders to sell the coins creating sell pressure.
Below peg the supply is contracted and removed from every user’s wallet proportionally to their ownership of the total supply. The reduction of supply should encourage opportunistic buying to bring price back up to peg.

Basis/Basis Cash

Above peg BAC (the stablecoin) is minted and distributed to users who have bonded their Basis Shares (BAS). These holders will then likely sell the additional BAC creating selling pressure to return to peg.
Below peg any user can burn some of their BAC in return for Basis Bonds (BAB). The exact amount of BAB returned is algorithmically determined. When price returns to peg BAB can be exchanged for BAC at a 1:1 ratio. The hope is that users will buy BAC to burn it for BAB — the buying pressure from which would drive price back to peg.

ESD/DSD

Above peg the total supply is increased and distributed to bonded liquidity providers and those who bonded their ESD/DSD into the DAO. After some lockup period these users could withdraw the rewards and sell them — thus creating sell pressure.
Below peg users could burn some of their tokens in return for coupons. The number of coupons received included some premium so you always receive more than 1 coupon for each ESD/DSD burned. When back at peg the coupons could be redeemed 1:1 for ESD/DSD. The hope here is that users will purchase ESD/DSD to burn for coupons, creating the required buy pressure to return to peg.

OlympusDAO

Above fair value the protocol mints new OHM and sells into the market.
Below fair value it uses its 1:1 DAI reserves to purchase and burn OHM.

FEI

Above peg users can purchase FEI via the bonding curve (with a max price of $1.01) creating an arbitrage opportunity for anyone to buy from the bonding curve and sell on the open market.
Below peg interactions with Uniswap are incentivized. Users buying FEI from Uniswap receive additional FEI minted by the protocol and users selling have some of their tokens burned. This is meant to discourage selling below peg and increase buying, thus tipping the net flow towards buyers to bring price back to peg.
FEI also controls the majority of the liquidity in the dex pool which allows it to reweight itself and bring the price back in line as a final resort. This mechanism is a very powerful innovation that FEI are bringing to the table.

Many of these individual mechanisms succeed at their specific role; and some of them have not. Generally, the main challenge for algorithmic stablecoins is the under peg mechanism. Encouraging users to sell tokens when above peg to realise profits turns out to be relatively easy. In contrast, convincing them to take the risk of buying under peg has been less successful overall.

Let’s take a closer look at what each coin does achieve and where it falls short.

Ampleforth

Ampleforth has generally done a pretty good job at maintaining peg, albeit with a lot of short term volatility around it. At the time of writing the price of AMPL is $1.0233281.

The positive rebasing (proportionally increasing every wallet’s balance) is a good way to encourage selling pressure as anyone who sells the additional tokens can extract profit.

However, under peg the rebase is negative (proportionally decreasing everyone’s wallet value) which fails to encourage buying as effectively as the above peg mechanism encourages selling. The only buying under peg is the result of swing trades expecting price to return to peg.

The rebasing mechanism has an amplifying effect — when price goes up it gives everyone more tokens (amplifying the increase in the wallet balance), and when price goes down they remove coins from everyone’s wallet (amplifying the decrease in the wallet value).

The main issues we see with AMPL are:

Speculators are the heart of the stability mechanism but they only make profit when the coin is volatile around peg.
The user experience of a constantly fluctuating wallet balance is poor. Especially when the changes to wallet value are amplified in both directions. Users may wish to sell tokens, only to find they have “lost” coins due to the price dipping under peg.

Basis/Basis Cash

The above peg mechanism works very well by distributing rewards with the hope the recipient users will sell and bring the price back down. The issues arise when price falls below peg.

Users can purchase the Basis Bonds below peg by burning Basis Cash (which reduces supply, thus improving the economics), however, this does not actively incentivise buying Basis Cash. The initial assumption is that users will buy Basis Cash to burn it for the Bonds. In reality, existing holders can burn their Basis Cash instead. This does nothing to directly improve the price, it simply reduces the supply of Basis Cash.

This results in the amount of Bonds being issued to rise with no impact on the underlying price of Basis Cash. This accrues debt that needs to be paid off above peg. The more debt accrues the harder it will be to pay it back using above peg expansion of supply. Eventually the debt becomes so large that it fundamentally destabilizes the coin.

The core shortfall is the lack of direct buy pressure when the price is under peg, which unduly prolongs the cycle time. The longer the cycle continues the more unstable it will be.

Another drawback of the system is there are three coins: BAC, BAS and BAB, which adds complexity for the end user to understand. Seamless user experience is critically important for the adoption of these protocols so adding any user-facing complexity should be considered carefully.

ESD/DSD

The ESD/DSD model is a good progression from the Basis Cash model. Instead of three different coins to consider there is only a core stable token + a debt coupon that acts much like the Basis Bonds.

Instead of supply expansion going to Basis Share bonders, ESD/DSD distributes these rewards to bonded LPs and those bonded to the DAO.

This achieves the same mechanism with one less token. Unfortunately, the distributed rewards are only claimable after a lockup period. This bottlenecks the sell pressure and allows the price to balloon considerably above peg before this accumulated selling hits the market.

ESD/DSD still do not address the core failings of the Basis Bonds mechanism under peg. Users can burn ESD/DSD to receive coupons with some premium that can then be exchanged 1:1 for ESD/DSD during expansion. Much like with Basis Cash there is no direct incentive to buy the token under peg, letting debt mount to the point of destabilizing the price. Furthermore, coupon premium is essentially a promise to mint even more ESD in the future (to cover that premium), which further undermines the actual contraction of supply. This forces every expansion to be larger and larger, which destabilizes the entire system.

To the detriment of human users, only a limited number of coupons can be redeemed each epoch — rewarding the fastest users. Bots consistently win this race, punishing regular users. DSD tried to offset the bot issue by taking a “tax” on coupon redemption based on how soon after buying the coupons they are being redeemed. Longer holders got taxed less and that encouraged a form of a first in first out system that also allowed anyone to skip the queue by paying the penalty.

While this did address the bot issue it still did not address the lack of buying pressure under peg.

Another fatal flaw is that coupons can expire. This threatens those who purchase coupons with the possibility of 100% capital loss. It will only take one batch of coupons to expire to see the community lose faith in the mechanism. This was observed in both ESD and DSD.

OlympusDAO

The underlying mechanism of OlympusDAO is very strong. Every OHM is backed by at least 1 DAI, meaning the protocol can always stabilize by either minting and selling, or by buying back and burning.

The project is still young and the team are not directly targeting immediate stability. They are first prioritizing growing the protocol and community. Hence, it is too early to fairly comment on the success of the project just yet.

Theoretically however it is a very strong mechanism. The trade-off they made was capital efficiency — the coin is backed 1:1 so it is more capital efficient than DAI but significantly less efficient than a completely algorithmic stablecoin. In return, they have a mechanism which is many times stronger and more robust than all existing 100% algorithmic coins out there.

We have very little direct criticism of OlympusDAO and think it is a strong project to watch.

FEI

FEI is the new coin on the block that is trying to maintain stability. Above peg FEI has a bonding curve that allows people to mint new FEI in a similar way to the mint/redeem schemes of centralized custodian stablecoins. This provides an arbitrage opportunity for users that will create the required selling pressure to bring price back to peg.

Conceptually, this mechanism holds strong keeping the price down when it creeps above peg. At the time of writing however, this mechanism has yet to be used at scale.

Below peg the incentive to buy is that you will receive additional minted FEI, essentially giving the buyer a discounted price to the market price. The amount of discount increases as the price falls below peg. This in effect creates an implicit dutch auction that lets the market price the premium they want in return for providing the buy pressure the protocol desires.

This is a great incentive to get people to buy more but would not work on its own. In isolation it would mean the supply is constantly growing in times of lower demand (under peg). This is not sound economics.

To fix this issue the FEI team introduced a negative incentive (a tax) on people selling under peg. The amount of tax on sellers and the amount of mint for buyers is balanced out to provide net supply contraction under peg.

There is a big problem with this mechanism. Formulating a kind of dutch auction that allows the market to set the price of the premium necessarily has to increase the tax on sellers to continue ensuring price contraction. In simple terms, it is a tax on the sellers to pay the buyers. This is a punishment both to those actively using FEI as a stablecoin but now wanting to deploy capital elsewhere, and to those losing faith in the protocol.

Punishing your fans is a good way to incentivise them to look elsewhere, and punishing those who are losing faith is a good way to solidify that lack of faith.

Another serious mistake (in our humble opinion) made by FEI was having an unlimited genesis event. This meant anyone who wanted FEI could obtain it in the genesis, pre-launch. Meaning by launch time there were a whole lot of people who already had FEI who could potentially sell. That selling is highly reflexive — as the price drops more and more people want to exit despite the tax on the sale, which drives the price even lower. A tough situation to recover from when there is already a lack of buy side.

The FEI liquidity rebalancing mechanism is very powerful and is always a tool the protocol will have at its disposal. FEI highlights the power of having a lot of capital controlled by the protocol. This will give them significant amounts of time to improve the protocol. We expect FEI will iron out the kinks and become a very strong player in the stablecoin space in the coming months and years.

Malt

With everything we have discussed so far in mind, let us distill the most important points which will make the Malt mechanism successful.

All of the existing coins manage to control price above peg, so Malt doesn’t need to reinvent the wheel here. Malt will just mint new MALT and sell into the market. What the protocol does with the proceeds from the sale will be described in more detail in upcoming articles.
Interacting with the protocol must be an intuitive and straightforward user experience.
Below peg Malt cannot simply buy back and burn as much as it wants due to its aim to not be 1:1 backed. Therefore, Malt will provide a strong incentive for users to buy under peg, while ensuring humans are not pushed out by competing bots (unlike the redemption of coupons in ESD).
It needs to achieve this incentive without punishing the loyal users of Malt.
It also needs to ensure supply contraction under peg despite any incentives given to buyers.
The mechanism must be able to react rapidly and the cycle time must be short.
It also must not pose additional risk to those engaging under peg (like the risk of 100% capital loss from expiring coupons).
The distribution of tokens must try to skew towards buying pressure in the early stages of the protocol to give the protocol time to find a stable level of liquidity.

We are confident that Malt achieves many, if not all of these goals.

Check out the previous article for an overview of how the Malt mechanism works. We will share more details on the inner workings of our Malt protocol over the coming days. Stay tuned!

Looking forward to it