Translator's Note: According to the original design of the anchor coin DAI, the original author Stefan Ionescu proposed a Reflex-Bond that resembles a hedging strategy. DAI and other stable currency collateral can greatly reduce system risks, thus laying a stronger foundation for the DeFi ecosystem.
(Picture from: tucheng.com)
- Attackers make a net profit of 350,000 US dollars. Can DeFi not only immoral arbitrage, but also centralization?
- DeFI: Ethereum's currency Lego
- Ethereum's killer app could "kill it"? Researchers think DeFi will eat into PoS cybersecurity
- DeFi ushers in new milestones, finance is still Ethereum's main use case
- Lightning loan: what can an Ethereum transaction do?
- Ethereum 2.0 will wait until 2021? DeFi may help Ethereum to "ignite the second fire" after ICO
The following is the translation:
It's time for an old idea to shine.
In the past few years, the term "stable currency" means "anchor currency". In the long run, its value is closely related to a specific asset or a basket of assets. However, recent market dynamics indicate that the narrative of the anchor currency actually poses problems for the stablecoin projects with mortgages . In other words, it limits the economic incentives that the system can use to influence the market price of stablecoins. One of them is the implementation of negative interest rates on all token holders. If the system has a built-in interest rate, the stablecoin will be decoupled, thus breaking the original "anchor" statement.
A notable example is that MakerDAO ’s DAI is currently working to maintain its anchored exchange rate. During the 312 "Black Thursday" period, the market price of DAI was more than 10% higher than the predetermined anchor price of $ 1. In response, holders of governance tokens (MKR) must reduce the stabilization fee and savings rate to zero. Even with these stringent measures, the market exchange rate of DAI is still higher than the anchor exchange rate. Then, many people in the Ethereum community will ask, "Why can't DAI rates fall to a negative value so that governance can put more pressure on the market?" The simple answer is that a meme of 1 DAI = 1 USD must be Staying alive at all costs, the system is not designed to support negative interest rates.
At least, this is not the solution adopted by the current DAI system. In the original design of Maker (as described in the Purple Book), DAI does have a negative interest rate, but, more importantly, the market price of DAI does not always return to the redemption price of $ 1. On the contrary, the redemption price itself is floating. To put it bluntly, the stable currency here is defined as a low-volatility asset (compared to its own collateral), and its redemption price and interest rate are determined by the market, not by "decentralized" governance .
The initial motivation for the floating redemption price is for the "Black Thursday" scenario. If the system is designed according to the original specifications, it will respond more decisively to offset the sharp increase in market prices. Sadly, this robust mechanism design was abandoned.
Well, it has been abandoned until now.
In the next section, I will describe the overall mechanism of a new system that incorporates the original ideas of the Dai Purple Book. The assets created by the system are not anchored to anything, but its purpose is to eliminate most of the volatility present in its underlying collateral . I will use the term "reflex-bond" (reflex-bond) coined by Nikolai Mushegian, an early contributor of MakerDAO, to refer to this new asset.
The purpose of reflective bonds is to represent their collateral more stably while maintaining a high degree of distrust . If used in other protocols, reflective bonds can protect its users from major and sudden changes in the cryptocurrency market. For example, if Maker uses “Reflective Bonds” as collateral before “Black Thursday”, CDP creators will have more time to avoid being fully liquidated.
"Reflective Bond" ETH = Delicious ETH
Before we go through a few examples to understand how "reflective bonds" maintain low volatility, let us familiarize you with the terminology you will encounter in the rest of this article:
- Reflective bond : A collateralized, non-anchored asset with lower volatility than its own collateral;
- Redemption price : The system hopes to reflect the price of the bond. For example, the goal of DAI is to always remain at 1 US dollar, and the price of the reflected bond is variable;
- Market price : Market valuation of reflective bonds;
- Redemption rate : The interest rate per second (can be positive or negative) used to incentivize CDP creators to issue more bonds or repay their debts. The redemption rate gradually changes the redemption price. The concept is similar to the interest rate, but Not the same
- Borrowing capacity : how many reflective bonds can be lent by a unit of collateral. Each time the system receives a feed price update that accepts the type of collateral, the feed price data is divided by the redemption price, and then divided by the liquidation ratio to calculate the borrowing capacity. If we use ETH as an example, assuming that the price of ETH is $ 100, the liquidation ratio is 150%, and the bond redemption price is $ 1, then the borrowing capacity of ETH is about 100/1 / 1.5 = $ 66.67;
For example, here are two graphs that show a scenario similar to "Black Thursday" and how the reflective bonds managed by the proportional controller will react:
- Before the 30th day, the redemption price and the market price of the reflective bond are both $ 1. Because the deviation between the two prices is zero, the redemption rate is 0%. In addition, the average mortgage rate of the entire system is 190%. For simplicity, we assume that the mortgage price is always $ 100, and the liquidation rate is always 150%.
- Between 30 and 40 days, as people began to pay more for reflective bonds (for example, they wanted to pay off debt, but there was not enough liquidity in the market), their market price rose to $ 1.1. In order to reduce market prices, this system needs to create more bonds. To encourage bond issuance, the redemption rate is -10% per year (Figure 1A), and the market price has soared by 10% (Note: The redemption rate of -10% will result in a redemption price of 0.9 USD after one year);
- Between the 40th and 75th days, the market price remains unchanged, but the redemption price declines slowly because it is affected by the redemption rate. As the deviation between the market price and the redemption price increases, the redemption rate also continues to decline. At the same time, the borrowing capacity of all CDPs will rise (Figure 1B), because reflective bonds are being repriced (for example, the bond redemption price changes from $ 1 to $ 0.95, and the new borrowing capacity is 100 / 0.95 / 1.5≈ $ 70.17);
- As CDP holders realize that their borrowing power is growing, even if their collateral prices remain the same, they will be incentivized to generate more bonds. This is why on the 75th day, as more and more bonds enter the market, we begin to see market prices slowly decline. At the same time, as the gap between the market price and the redemption price narrowed, the redemption rate began to approach 0%;
- On the 180th day, the gap between the two prices became 0 (Figure 1A), the redemption rate is now 0%, and all CDP's borrowing capacity ceases to grow (Figure 1B);
If the market price is lower than the redemption price, a similar situation will occur (Figure 2A). The redemption rate will become positive, so the redemption price will start to rise. As the redemption price rises, the reflective bond will be repriced, so the cost of creating the reflective bond will become higher (for example, at a 15% annual redemption rate, a starting redemption price of $ 1, and a collateral value of $ 100 It is calculated that within one year, the reflective bond will be re-priced to $ 1.15, and the collateral borrowing capacity will be 100 / 1.15 / 1.5≈ $ 57.97), and then all current open CDP mortgage ratios have declined (Figure 2B). This means that CDP creators will realize that they will either need to add more collateral to their positions or repay part of their debt to avoid liquidation.
For simplicity, I deliberately omitted some details. In most cases, there will be a small deviation between the market price and the redemption price, and the controller can be set to noise classification. In addition, ideally, the PI controller should be used to modify the redemption rate. This PI controller can help minimize the market / redemption price deviation (as shown in the ideal simplified scheme above), and can even reduce the integer controller through the PID controller . integral.
The controller can be started with specific parameters and then fully autonomous. In this case, the system's dependence on governance is reduced, or its parameters can be adjusted manually over time. However, a warning to be given here is that if you choose the option of full autonomy from the beginning, it may prove to be invalid because of the lack of real-world data as a parameter basis.
Another important detail is that the redemption rate mechanism replaces the savings account (in the case of Maker, the Dai savings account), and implicitly deletes the savings interest rate (Dai savings rate) for interest earned by stablecoin holders.
Finally, even if the redemption rate is negative, the system can still charge a stable fee. When the reflective bond is repriced due to the redemption rate, the amount of stabilization fee charged (in terms of reflective bond) will not change. But this is not all: a version of the system can charge a fixed fee, but can only modify the redemption rate. The fixed fee and redemption rate estimated based on the market price / redemption spread give CDP creators greater predictability in terms of the costs and risks associated with their open positions.
You can always predict the direction of the redemption rate because it is the exact opposite of the deviation
How does reflection bond achieve trustlessness?
In order to create a reflective bond, we need to combine multiple parts together:
- A reliable oracle system, which is responsible for providing price information for bonds and collateral;
- Procedures for setting interest rates that change the redemption rate;
- A loan mechanism;
- A liquidation mechanism;
- Set the components of the loan interest rate (stable fee procedure);
As for the detailed parts of these contents, I will discuss them in detail in the next paper and expand them in future posts. Now, suppose that depending on the type of collateral selected by the bond, governance can set many parameters at once (stabilization fees, liquidation ratio, etc.), and then cancel most of the control of the system. In addition, multiple reflective bonds can also be introduced, each of which has its own collateral type and governance level in order to give the market as many options as possible.
When designing the oracle and redemption rate setting program, the difficult part came. One option for collateral feeding is to build an aggregator for different oracle networks, store the results of each network in a sorted array, and then choose the median. By using Uniswap v2 to exchange bonds with a specific fee token for each oracle, part of the stable fee can be used to pay for oracle calls. The price of bonds can be obtained from the bond / COL Uniswap v2 pool. COL is one of the types of collateral that supports bonds. It can also be provided through the governance whitelist oracle until the oracle network provides bond feeds.
As for the rate setting procedure, a simple implementation is a PI controller, where the smart contract automatically calculates the proportion, while the governance only sets the integral item. A more complex implementation method is to use a PID controller, where the integral and derivative terms are calculated by the contract using the market price deviation accumulator. With PID, governance may still need to adjust some parameters, although in general, they have minimal impact on the system.
The end result is that after it is launched, you can handle most of the components of reflective bonds without trust governance.
How does the agreement use reflective bonds?
After starting the system, anyone can deposit ETH atomically to create a reflective bond, and then deposit the reflective bond ETH in another protocol to borrow or create other crypto assets, as well as synthetic gold, synthetic oil, synthetic stocks, or even The currently popular synthetic dollar, while the reflective bond acts as a middleware between the initial crypto assets and the final agreement. The main benefit of using these bonds as collateral is that they can mitigate volatility.
To encourage the use of reflective bonds in other systems, they can be set using the default positive redemption rate. On the bright side, the redemption price of these bonds will be higher in the long run. On the other hand, the creators of CDP will gradually need to add more collateral to their positions to avoid liquidation due to bond repricing. Of course, the default redemption rate may be small enough that liquidation is not an imminent threat.
to sum up
DAI is defined as an anchored asset, which means that it must maintain its market price close to a single, illiquid level. But in any case, its original goal is different, and more similar to reflective bonds.
Reflective bonds are similar to hedging strategies, and their purpose is to be less volatile than the collateral that supports them. You can think of the system that creates the reflective bond as a washing machine, but it “cleans” the volatility of its own collateral, not the clothes. With this feature, reflective bonds can actually be used as collateral for anchor coins, which will then reduce exposure to crypto assets such as ETH.
Special thanks to Ameen Soleimani, Fernando Martinelli, Grant Hummer and Nikolai Mushegian for reviewing the draft of this article.