Understand the evolution of the four major financial primitives of DeFi: liquidity, leverage, risk and arbitrage

The development of DeFi is advancing by leaps and bounds, and various innovations based on composability are emerging one after another, but there are still traces and reasons behind it.

The venture partner of investment agency Playground pointed out in an article on the Medium blog that liquidity, leverage, risk and arbitrage are the main factors driving the development of the DeFi market, and elaborated on how the DeFi protocol evolves from their perspective. Chain Catcher translated this article and made edits that did not affect the original intention.

Although DeFi is still in the experimental stage, it has been very mature since I first wrote about it in June 2019. With the successive launch of V2 and V3 of the DeFi protocol, I re-examined the industry and how it is developing. After working in finance for many years, I realized that most of the factors affecting the DeFi market can be attributed to four basic financial elements: liquidity, leverage, risk and arbitrage. This article outlines the evolution of the logic of the DeFi market through these few lenses.

This article outlines financial primitives, the core building blocks of financial markets, which are different from software or encryption primitives. This article assumes that you have a certain understanding of finance, encryption and DeFi, so the introduction to the protocol is not comprehensive.

1. Liquidity

Almost all factors boil down to liquidity, but we have always underestimated the importance of liquidity. Higher liquidity leads to smaller slippage and higher market efficiency. Lower liquidity exaggerates market movements and expands selling. It produces a flywheel effect during its ascent, and a cliff phenomenon during its descent.

The V1 DeFi agreement relies on the liquidity vacuum of restricted capital. The term restricted capital refers to capital that is underutilized and inefficiently allocated in the agreement.

This leads to opportunity costs, because the capital could have earned higher returns inside or outside the agreement. The V1 DeFi agreement relies on exclusive capital: MakerDAO requires a minimum mortgage rate of 150%, the loan agreement has not yet accepted ultra-liquid collateral (V2 innovation led by Compound), and Uniswap’s liquidity is very inefficient in all price curves.

The opportunity cost of restricted capital has expanded along with DeFi, and now V2 and V3 are striving to achieve higher capital efficiency.

As liquidity collateral

Stable coins play a huge role in DeFi. The stable currency backed by fiat currency has many problems, such as centralization, supervision, and potential competition from CBDC. Crypto-collateralized stable currency is the most viable option. However, V1 DeFi relies on over-collateralization to maintain its pegged USD exchange rate, so it cannot expand its scale.

Many of the latest stablecoin iterations (FEI, OHM, FLOAT, FRAX) utilize Protocol Control Value (PCV). This is a concept in which the collateral supporting stablecoins cannot be redeemed by users, but belongs to the agreement (the agreement determines whether/how to invest, it can be used to restore anchoring, etc.). This is similar to a treasury or insurance fund, but the difference is that PCV can be converted to liquidity immediately (AMM pool).

Regardless of development issues, FEI’s PCV makes it the largest liquidity provider (LP) on Uniswap. AAVE V2 similarly blurs the line between collateral and liquidity, allowing borrowers to use existing collateral to repay debt. Proof of liquidity creates a similar motivation for pledged derivatives, which will be described in more detail below.

DeFi V1 solves how to use liquidity as collateral (LP tokens). The V2 and V3 agreements are studying how to convert collateral into liquidity.

Unfortunately, liquidity alone cannot bring stability. The real reason for the collapse of the stable currency system is the crisis of confidence under the anchor exchange rate system. Solving the problem of restricted capital is not enough. Over time, a sound economic mechanism is needed to keep stablecoins near an anchored level, and mechanism design is difficult. Pure algorithmic stability mechanisms (Basis Cash, Empty Set Dollar) have been struggling, and these mechanisms require users to maintain confidence in future anchoring mechanisms.

The so-called direct incentive is to punish transactions that deviate from the pegged exchange rate, and reward transactions that are pegged to exchange rates, and ultimately remove liquidity from the system when it is most needed. Severe selling penalties imitate restricted liquidity and control more capital. This is the first problem we try to solve.

Other models try to solve the confidence problem first and improve capital efficiency over time, starting with fully collateralized tokens and allowing the market to dynamically adjust the collateralization ratio. As confidence increases, capital efficiency will also increase.

FRAX uses this model, but is currently supported by a basket of stablecoins backed by fiat currencies. Although this mechanism seems to be working, if it is fully secured by unregulated assets, it is unclear how it will be sustained.

Although the latest generation of stablecoins focuses on reducing mortgage requirements, we need to resolve the issue of confidence in the anchoring mechanism before improving capital efficiency.

Liquidity as a liability

When it comes to liquidity, we cannot but talk about the dominant Uniswap. Uniswap played an important role in the rise of liquid mining, which in turn played an important role in the development trajectory of Uniswap. After we realized in a difficult way that liquidity is not a moat, the focus shifted from acquiring liquidity to retaining liquidity.

Competitors are beginning to move forward, adding higher moat/margin services, such as Kashi Lending, borrowing from the traditional financial technology approach: acquiring users in a cheap way and upselling credit products.

Instead, Uniswap fundamentally reconsiders the liquidity mechanism of asset management, resulting in v3 that improves capital efficiency by up to 4000 times under special circumstances.

Although AMM V1 is an innovation from 0 to 1, they are also very inefficient because they need to provide liquidity for prices that may never be reached. For example, if ETH is in the range of US$200 and there are US$10 million in the ETH/DAI pool, then the liquidity pool for buying ETH under US$10 or more than US$5000 may be as high as 25%.

In this example, there is little chance that liquidity will be required at these levels. Maintaining this constant price curve will result in a lower volume (~20%), because a locked capital of US$5 billion is equivalent to a volume of only US$1 billion.

The even distribution of liquidity across the entire range also means that very little liquidity is concentrated in the places with the most transactions. Curve realized this early and created AMM specifically for stablecoins, which have a narrow trading range.

Uniswap v3 solves these problems and is closer to restricted orders in the process, because LP can now specify the price range for providing liquidity (that is, ETH/USDC from 1800 US dollars to 2200 US dollars). This change should cause almost all transactions to occur in the interval near the mid-price, thereby improving the places where liquidity is most needed.

More concentrated liquidity will also reduce asset risk. For example, if an LP bids $500 for an ETH/DAI pair, but is convinced that the price of ETH will rise, then they assume the risk exposure (ETH opportunity cost) of the asset (DAI) they do not want to hold. The concentrated liquidity of Uniswap v3 enables LPs to significantly increase their exposure to priority assets, thereby reducing this risk.

Balancer V2 also attempts to reduce inventory risk by introducing an asset manager, which allows LPs to lend out assets on one side of the trading pair without being used for trading liquidity. Yearn’s Stablecredit uses a similar function.

The next-generation asset management system requires less capital, but it will bring more liquidity.

Liquidity transaction

Hyper-liquid collateral is a V1 concept, which refers to the ability to tokenize locked assets (collateral or liquidity) to obtain liquidity or leverage for that asset. Pledged derivatives extend this concept to pledged assets (assets that ensure the proof-of-stake network) through pledged tokens (stETH, rtokens), essentially allowing users to deploy pledged assets to other places more effectively.

Proponents of pledged derivatives believe that without these derivatives, the liquidity of pledged tokens will be affected. There are also concerns that if the verifier can obtain higher returns from the DeFi protocol, the verifier will not be incentivized to pledge tokens.

In theory, staking derivatives can increase the staking ratio of ETH from 15-30% to 80-100%, because it eliminates the additional cost of staking compared to non-staking.

Pledged derivatives also allow the creation of new financial instruments. For example, rewarding “guaranteed” cash flow with proof of equity can make the product look similar to interest-bearing bonds (Terra’s bAssets, Blockswap). These tools can be used to generate sustainable, stable, and relatively high yields (such as Anchor Protocol), which may allow more mainstream users to support DeFi.

Compared with collateral, pledged assets are unique in that they are not just a payment promise, but a security mechanism. Some designs, especially those that enable pledged derivatives to move across chains, allow for risk conversion (from endogenous to exogenous), which may affect the basic game theory that helps protect public networks. In contrast, proof of liquidity converts pledged tokens into the liquidity of the underlying network tokens in a way that balances capital efficiency and network security, which should be the main priority.

2. Leverage

Leverage magnifies gains (which is the ultimate result of capital efficiency), but it also magnifies losses. It is easy to build leverage, but difficult to control. We like leverage until we hate it.

Due to excessive or implicit leverage, traditional financial markets have collapsed countless times. In the past six months alone, we have seen the market impact of Archegos’ high leverage and Gametop’s short selling leverage (140% of Gametop’s liquidity is short-sold).

In the cryptocurrency market, we have seen $10 billion in liquidation in the last 24 hours, partly because of the serial liquidation of leveraged bulls. This is a stress test for the crypto market, and some DeFi protocols are under considerable pressure.

Create leverage

In the DeFi summer of 2020, crazy activities are largely driven by the active leverage strategy of liquidity mining. Although this activity has subsided since then, we have begun to see the emergence of new leverage mechanisms. Element’s revenue token is an example.

When a user deposits collateral through Element, two tokens will be generated: the main token and the income token. Assuming that a user deposits a principal of 10 ETH at an annual interest rate of 20%, the user can sell his main token at a discount, and at a fixed interest rate of 10%, the user will receive 9 ETH, which can be maintained by the income token. Exposure to all 10 ETH interest over time.

After that, the remaining 9 ETHs establish a new position and repeat the operation to achieve a leverage ratio of up to 6.5 times. Compared with earlier loan agreements, the ability to obtain full principal and interest and the ability to obtain the net present value of the loan are unique.

Although there are some agreements aimed at realizing under-collateralized loans in DeFi, they are mostly conceptual. Although CREAM V2 strives to achieve unsecured loans from agreement to agreement through Iron Bank, it is only applicable to whitelisted partners. The parameters are directly determined by the CREAM team, highlighting the current limitations.

On the contrary, Alchemix approached this issue from a completely different perspective, allowing borrowers to benefit from overcollateralization. For example, users who deposit 1000 DAI can access 500 alUSD. Put 1,000 DAI in the Year’s vault to earn income, which can be used to repay the loan over time.

Another option is to buy $500 first, and then invest the unspent $500, which is obviously lower than the return of $1000 (a return of $280, assuming a two-year annual interest rate of 25%, then $280 less). Similarly, the borrower can still obtain the discounted portion of the principal while earning the full principal interest.

The new lending agreement takes advantage of the time value of money and the separation of principal and income, allowing users to benefit from (over)collateralization.

The complexity of cross-collateral

Although the early DeFi ecosystem mainly relied on a cycle denominated in one asset (ETH), V2 expanded the complexity of lending protocols by allowing the use of a multi-collateral system.

In this system, n assets can be borrowed with m types of collateral, single collateral DAI evolves into multi-collateral DAI, Compound supports the cross-collateral currency market, and Aave and CREAM expand by supporting more and more assets . Yearn’s StableCredit protocol allows users to create synthetic debt positions, which are essentially swap collateral (a feature supported by AAVE V2 through lightning loans).

Some agreements go one step further by centralizing the risk exposure of all assets and spreading counterparty risk to users. On Synthetix, when the value of any synthetic asset minted in the agreement increases, it will increase the value of the total debt in the system, while the user’s ownership of the total debt pool remains unchanged. This may cause the user’s debt balance to increase due to the increase in the price of assets that the user does not have direct exposure to.

This complexity of cross-collateral and cross-asset exposure improves functionality, but it also increases the possibility of market transmission, that is, the selling of one asset may cause the selling of other assets.

Control lever

Composability enables rapid innovation, but it also means that DeFi Lego bricks can quickly become dominoes. Although the chain is transparent, it is difficult to create a unified view of cross-protocol comprehensive leverage, which means that there is currently no simple way to understand the ratio of commodity credit to circulating credit, which has an impact on the solvency of the system.

Solvency is particularly important in systems where there is no lender of last resort. In addition, although centralized venues clear underwater mortgages on their own to avoid counterparty risks, decentralized agreements rely on third-party liquidators to eliminate underwater debt from the balance sheet. These liquidators can choose to buy underwater collateral from the agreement at a discounted price, but they can also choose not to buy it due to volatility, network congestion or other market factors.

Liquidity attempts to solve this problem by creating a pool of funds that can be used for liquidation. Under this model, the LP agrees in advance that its liquidity will be used to purchase collateral at a discount during the liquidation period. Although this allows the agreement to reduce the loan mortgage rate to 110% and provide a fixed interest rate of 0%, due to the price drop, LPs may eventually purchase collateral, thereby paying higher costs.

3. Risk

In the financial sector, there is an inevitable ratio: risk/reward. It is simple to understand: higher returns correspond to greater risks, and it is very difficult to break this ratio.

When the market is a new market, risk is usually expressed in binary form: risk taking or hedging. With the development of the market, people have a better understanding of the composition of risk, and the risk transfer mechanism has also developed. At the same time, a sliding risk that allows market participants to express their personal risk tolerance has been formed.

Dual risk

Volatility products make market participants have a dual view of market risk and are an important part of market infrastructure. The volatility index represents the market’s estimate of future volatility and is the cornerstone of the traditional financial market. The CeFi market has already provided some VOL products (FTX MOVE), but there are almost no similar products in the DeFi market. Protocols like Volmex are working hard to create a volatility index, while the stablecoin of the Benchmark protocol uses VIX as an input to its stability mechanism. Index Coop seems to be a natural candidate for DeFi’s inherent volatility index. Opyn also expressed interest in creating ” DeFi VIX”.

Floating risk

The emerging DeFi protocol is developing a “risk matching engine” to match market participants with lower risk appetites with market participants with higher risk appetites. Most projects achieve this through a multi-token system, which separates the speculative and non-speculative aspects of the protocol and redistributes cash flows accordingly (Saffron, BarnBridge, Element).

For example, Element separates principal and yield, allowing users to purchase the net present value of the principal (basically a zero-coupon bond), or assume a leveraged yield of up to 10 times. BarnBridge divides cash flow into fixed rate of return and variable rate of return. Variable-yield holders earn higher than fixed interest rates, but subsidize fixed-yield token holders in the event of a gap.

Similarly, Saffron divides risk into priority and inferior, and priority bonds have lower yields. Higher returns = higher risks.

These programmable risk protocols support peer-to-peer risk transfer and allow users to bet on different parts of the capital structure. Risk-averse capital (such as the company’s financial department) may prefer to buy priority products, while gambler traders who chase yield may prefer inferior products. Enhancing risk tolerance is a positive development for the DeFi market, and tiered transactions are not inherently dangerous.

Fourth, arbitrage

Arbitrage is necessary for the normal operation of many DeFi protocols, and the design of any protocol needs to consider arbitrage. Uniswap relies heavily on arbitrageurs to maintain price discovery. Some people believe that the scale of MakerDAO is limited due to the infeasibility of arbitrage CDP. Yearn’s StableCredit relies on arbitrage rather than governance to maintain the stability of the system.

1inch, Year, ParaSwap and Rari can all be used as arbitrage tools. ThorChain and Serum will be the key to achieving cross-chain arbitrage, because the DeFi protocol will be launched across the first layer of blockchain.

As DeFi expands, reducing gas costs will provide a temporary arbitrage opportunity, which is increasingly shaping the V2/V3 design of the DeFi protocol (Aave V2 collateral transactions, dydx gas-free orders, Balancer V2 collective assets, Sushiswap BentoBox).

V1 focuses on cross-protocol arbitrage, and V2/V3 focuses on cross-chain arbitrage.

Arbitrage improves market efficiency (the extent to which market prices reflect all available relevant information), but it also leads to narrowing of interest rate differentials and normalization of returns, eliminating the temporary advantages between agreements. As market efficiency improves, so does competition. To maintain a dominant position in this environment requires true protocol innovation, not just gradualism.

The V2 and V3 DeFi agreements will require innovation in all four financial primitives, because their interconnected nature is the factor that drives the market. Without arbitrage, stablecoins cannot expand. Without liquidity, there is no arbitrage ability. Leverage allows you to take more risks, and risk transfer allows you to take more leverage.

The important thing is that DeFi supports brand new financial primitives, such as flash loans, and these financial primitives have no traditional counterparts. Agreements that address all these primitives and/or create new primitives will lead the development of the next generation of DeFi.

I can’t wait to watch V4 and V5.