Mirror Protocol: A DeFi Application that Terra had gotten right?
The Terra ecosystem’s dramatic collapse were some of the most devastating news crypto investors had to withstand in 2022. Nevertheless did the Terra-native Mirror Protocol make important advances in the field of synthetic asset tokens — an innovation that could soon play an important role for investing in all kinds of financial assets.
Written by Paolo Di Stefano — August 24th 2022
Mirror Protocol, introduced in December 2020, was the most advanced DeFi-platform for synthetic asset tokens until the entire ecosystem collapsed in May 2022 due to flaws in its algorithmic stablecoin design.
Synthetic asset tokens are designed to track the prices of real-world financial assets of all kinds and were referred to as mirrored assets, in short mAssets, on Terra. The tokens’ prices are usually soft pegged to their real-world counterparts by a system of dynamic arbitrage and liquidation incentives.
Just like for most crypto tokens, synthetic asset tokens allow for cheap and fast transactions, fractional ownership at arbitrary levels, and permissionless access thanks to decentralization. Since this sounds highly promising, this article aims to shed more light on how synthetic asset tokens function based on Mirror’s mAssets, and analyze how well they managed to track the prices of their real-world counterparts. This will help to critically answer the following two questions:
What are the differences between a synthetic asset investment and a direct investment in the underlying asset?
And based on that, to what extent are synthetic asset tokens more attractive than direct investments to investors looking for equity exposure?
1. Synthetic Asset Tokens vs. Asset-Backed Tokens
To grasp the potential of synthetic asset tokens, it is important to first understand how they differ from asset-backed tokens, which are a more widely known approach of representing real-world assets on blockchains:
Asset-backed tokens (ABTs) are fully backed by the real-world assets they represent. The main idea behind ABTs is to securely store the underlying assets with a trusted custodian and then tokenize them on a specific blockchain. This facilitates the transfer of property rights over the tokenized assets by leveraging the transactional efficiency of blockchain technology.
Contrary to ABTs, synthetic asset tokens such as Mirror’s mAssets do not require a one-to-one backing by the represented assets and thus do not provide the holder with any additional ownership rights. Synthetic asset tokens solely allow for exposure to the represented assets’ price movements (Liu & Lee, 2021).
2. The mechanics behind minting mAssets
On Mirror, new mAsset tokens could only be minted after creating a collateralized debt position (CDP) with another whitelisted token as collateral (e.g. a stablecoin). The minting process is to be understood as taking out an mAsset loan directly on the blockchain that is secured by locking eligible collateral tokens. The collateral ratio, which is defined as the value of the collateral divided by the value of the loan, could not fall below a defined minimum. This ratio typically ranged from 150% to 200% and mainly depended on the volatility of the chosen collateral token (Mirror Protocol, 2022a).
If a user’s collateral ratio fell below the required minimum, the CDP became subject to liquidation through auction. This means that other users could place bids to buy the user’s previously locked collateral at a discount (Mirror Protocol, 2022a). As soon as the auction was over, the proceeds were used to buy back the owed mAssets on the open market and burn them to close the open loan position (Liu & Lee, 2021).
The prices of both the collateral as well as the loan positions were determined via an oracle feeder. Oracles are designed to provide tamper-proof information from the outside world to the blockchain in real time (Chainlink, 2021). The mirrored assets’ real-world prices thus only affected the valuation of the borrowers’ loan positions, but not the prices for which the mAssets were effectively traded. The mAssets’ market prices could therefore deviate from the real-world asset prices provided by the oracle.
3. How did mAssets retain their peg?
The mAsset’s soft peg was maintained through arbitrage and liquidation incentives combined with governance mechanisms:
Premium: If an mAsset was trading at a premium compared to the real-world asset, users were incentivized to mint the mAsset and sell it, thereby effectively entering a short position to collect the premium once the price would move back to the soft peg (Liu & Lee, 2021).
Discount: Conversely, when an mAsset was trading at a discount compared to the real-world asset, mAsset minters that had previously sold their mAssets and entered a short position, had an incentive to buy the mAssets on the market and pay back their open mAsset loans, which was equivalent to burning the mAssets (Young, 2021).
Liquidation: To recall, when a CDP fell below the minimum collateral ratio, the collateral was automatically put up for sale in an auction, using the proceeds to buy back the mAsset. Consequently, liquidations drove up the price and helped it converge towards the price of its real-world counterpart when trading at a discount (Cryptopedia, 2021).
Governance: Finally, Mirror relied on governance mechanisms to fix situations in which the market prices deviated too much from the real-world prices. Measures taken by governance could for example be the modification of collateral requirements and/or specific transaction fees to steer the attractiveness of minting certain mAssets.
4. How well did mAssets track their real-world counterparts?
To find out how reliably the described incentives and mechanisms worked in maintaining the mAssets’ soft pegs, we analyzed the historic prices of two out of the 34 available mAssets: mAAPL and mTSLA. The necessary price data was retrieved via the official Mirror API, a GraphQL-based data service, for a total timeframe ranging from 01.01.2021 to 31.03.2022. When directly comparing the mAsset prices with their real-world counterparts, it looks as follows:
To display the relative price dispersions between the mAssets’ market prices and their real-world counterparts more clearly, we also looked at the spread based on the following formula. The ensuing plot therefore displays both the mAssets’ spreads over the same timeframe.
We can see that both mAssets had mostly been trading at a small premium between 0% and 3% compared to their real-world counterparts. In Q1 of 2021, this premium had however briefly surged to over 20%, which the Mirror team mainly attributes to design imperfections with regards to liquidity provision rewards that were initially only collectible by taking on long or neutral exposure to the mAssets.
This asymmetry was later addressed by the new short minting functionality, in which the minted mAsset was immediately and automatically sold on the decentralized exchange, and an additional un-tradable token was minted and issued to the user. This so-called short LP token could be staked to receive a (dynamic) portion of the liquidity provision rewards now also while effectively shorting the mAsset (Mirror Protocol, 2022b).
Interestingly, periods in which the mAssets were trading at a premium still lasted significantly longer than periods in which they were trading at a discount, even after the introduction of the short minting mechanism. This indicates that the incentives in place to reduce the mAssets’ spreads when they were trading at a discount had remained stronger than the incentives when they were trading at a premium.
Lastly, we can also observe that the correlation between the two spreads was very high (0.92) until the end of Q3 of 2021. Then, the spreads suddenly started diverging, reducing the correlation to -0.24 for the period of 01.10.21–01.03.22. Remarkably, the volatility of the mTSLA spread has at the same time increased significantly compared to the volatility of the mAAPL spread (standard deviation of 2.4% for mTSLA vs. 1.1% for mAAPL). An explanation for this divergence could not be found. This indicates that the mAsset markets were either not sufficiently efficient yet and/or are subject to additional factors that were not taken into consideration in this short analysis.
5. Conclusion
To recall, the goal of this article was to find out what the differences between a synthetic asset investment and a direct investment in the underlying asset are, as well as to what extent synthetic asset tokens would be more attractive than direct investments to investors looking for equity exposure.
Being blockchain-based, the advantages of synthetic asset token investments certainly lie in their transactional capabilities that are not restricted by national borders, their possibility for fractional ownership at arbitrary levels, as well as their permissionless access thanks to decentralization (meaning anyone with a smartphone can use them). Furthermore, they can be used to interact with other DeFi protocols and thereby earn additional yields for example by providing liquidity to a decentralized exchange.
Synthetic asset tokens do however not convey ownership rights over the underlying assets and hence do not come with the right to a dividend or voting rights; they solely provide exposure to their real-world counterparts’ price movements.
Since synthetic asset tokens are not hard pegged to their real-world counterparts, they also do not necessarily track the prices of their underlyings very reliably as the quantitative analysis showed. This adds an additional layer of volatility to the investment, which should however decrease with more sophisticated protocol designs and higher trade volumes.
The case of Terra furthermore showed that decentralized networks still bear substantial systemic risks, as well as technical (smart contract) risks as the developers’ sloppiness even caused multiple exploits on Mirror shortly after the hasty introduction of Terra 2.0 (e.g. Wright, 2022).
In their current state, synthetic asset tokens are thus mainly interesting for unbanked users. For them, synthetic asset tokens represent the only way to get exposure to financial markets. These investors however have to take into consideration the tokens’ special risk profile arising from their unique characteristics as well as the potential price dispersions from their real-world counterparts.
There are currently multiple projects trying to take over Mirror’s transpired pole position in DeFi for synthetic asset token trading. Existing alternatives are for example Julian Hosp & U-Zyn Chua’s DeFiChain, Secret Network’s Shade Protocol, and the recently launched Youves Platform on Tezos. The highest trading volumes for synthetic asset tokens are currently achieved on the Ethereum-based Synthetix Protocol which has existed since 2018 but whose mechanics differ substantially from Mirror’s and the other three’s. Apart from that, FTX and Binance have also already issued their own (centralized) versions of synthetic asset tokens. Due to a lack of regulatory approval in many jurisdictions, Binance was however already forced to stop offering them again (Ossinger, 2021).
Read the full paper
This article is based on the academic paper “Mirror Protocol: Synthetic Stocks on the Terra Blockchain” that was co-authored by Paolo Di Stefano & Gian Pfister under the supervision of Prof. Ph.D. Andrea Barbon & Prof. Dr. Angelo Ranaldo in the MBF-HSG Research Seminar in Financial Economics (Spring ‘22). The paper was written before the UST de-peg and subsequent demise of the Terra network. While the downfall of Terra certainly impacts the paper’s relevance, it does still provide interesting insights into primitives for blockchain-based synthetic asset models. The full paper is available here: https://drive.proton.me/urls/J62H3896DM#knauOPMZ8MG0
References
Chainlink. (14th September 2021). What is a Blockchain Oracle?
Cryptopedia. (7th December 2021). Mirror Protocol (MIR): Synthetic Asset Creation Platform.
Liu S. & Lee I. (2021). Mirror Protocol v2. [Whitepaper]
Mirror Protocol. (2022a). Collateralized Debt Position.
Mirror Protocol. (2022b). What’s new for Mirror v2?
Ossinger, J. (16th July 2021). Cryptocurrency Exchange Binance to Stop Support for Stock Tokens.
