Since then, we've had a lot of interest, with many developers and entrepreneurs from all over the world building the first DeFi-apps on Harmony.

This Medium post shares our thoughts on the different types of financial assets and instruments that exist on the blockchain today, and how to make protocols like Harmony work in the ecosystem. Therefore, this article will think from the perspective of Harmony, namely:

● Proof of pledge agreement

● Use BFT mechanism to reach consensus (such as Cosmos or EOS, unlike Eth 2.0 and Tezos under the "follow the longest chain" consensus)

first level title

Essential elements of a successful DeFi platform

First, let's define a DeFi "participant". At this early stage, we can assume that there are currently 3 groups:

● Dapp developers with certain financial background

● Market makers and arbitrageurs, indispensable participants in agreements such as Uniswap (liquidity pool) and market makers (CDP keeper).

● Encryption enthusiasts who use protocols that are more loosely defined

DeFi Dapp developers are currently focusing on:

● Stable platform. History is helpful because it is the best way to assess the stability of the platform.

● Stablecoins for non-speculative use cases

first level title

What is running today? what not?

Once the protocol is released, traction is mostly measured in the amount of money staked (eg: DefiPulse)

DefiPulse clearly shows that the lending business is the first use case. Borrowing does not require the processing speed of the chain, so the current Eth 1.0 is a good choice. Additionally, the KPIs used by DefiPulse, ranked by "lockup" value, are applicable to Uniswap where liquidity is locked, but not to more "traditional" exchanges such as 0x. In terms of transaction volume, Uniswap’s traffic is twice that of 0x.

Now, if we look at the centralized cryptocurrency exchanges, there are more than $10 billion traded per day. The reason we haven't reached such a high amount on DEX is that there is no on-chain throughput. A 15 second delay and usually about 50 cts+ to initiate or cancel a trade is too slow and expensive, especially to attract the necessary market makers to create a liquid market. This could be a use case very relevant to scalable on-chain networks like Harmony.

secondary title

There are three main types of stablecoins:

● Fiat currency collateral, they are well understood, and their main challenge at present is to audit the compliance work of the trust layer

● Encrypted collateral, currently used by Maker

● Algorithms, like the now-defunct Foundation, or Ampleforth, stability comes from parameters embedded in the protocol that arbitrageurs use to maintain a stable price. But unlike crypto ancillary agreements, it has no ancillary agreements. It’s similar to how fiat currencies work, except there’s no government or state behind it, and the rules are fixed and unchanging. This category has not seen meaningful adoption.

While all three categories are considered, Harmony and the developers who build on our infrastructure, in this article we will focus on the second category.

While Maker has been hugely successful on Ethereum, does the same apply to the Proof of Stake protocol? With the release of the highly anticipated eth2.0 release, the same question will also be involved in the future of MakerDAO.

Is it possible to generate a constructed dollar like DAI on a Proof of Stake protocol?

On a proof-of-work protocol like ETH 1.0, long-term ETH holders are incentivized to loan out their tokens to generate yield, whether it’s a custodial loan on Genesis Mining or BlockFi, or a non-custodial one like Compound or Dharma loan. With Maker CDP, ETH holders can use their ETH holdings to mint DAI tokens. This provides leverage similar to other lending solutions, although the role of the lender has been taken over by the Maker CDP smart contract.

Inflation tokens generated by ETH 1.0 (a PoW protocol) are provided to miners who do not need to hold any tokens and do not give any rewards for the tokens they hold. Miners still have a need to invest in hardware, but this is an off-chain resource.

Its economics are different from proof of stake agreements. Holding tokens is mandatory in order to "stake" part of the protocol inflation. Long-term token holders have a natural option for yield generation: staking.

Borrowing is still an option, as borrowing and staking have different benefits.

Staking is used to secure the protocol. The loan is matched with the borrower, who can use the national token, buy DAI, convert it to fiat currency, and use it in the "real world", or use the national token for long-term leverage.

But overall, long-term token holders will seek the highest risk-adjusted returns, which will be determined by market forces.

Assuming yields will be driven by market forces, it's interesting to think about how yields might evolve between betting and lending.

In a bull market, we can expect lending to compete with staking, as we recently saw with MakerDao and the interest rate (stability fee) has risen to 19.5% to keep the DAI peg. This is higher than the usual protocol generation rate today (https://stakingrewards.com/ average 13.3%).

Now, if we look at how Maker is operating in a bear market environment, the interest rate was 0.5% for most of 2018 last year. This was enough to anchor DAI at $1 as there was little selling pressure on DAI.

But what happens if ETH holders have a way to earn 5% - 10% through staking? Will the supply of DAI not keep up? Eventually it will lead to increased volatility, loss of the peg, and stablecoins in DAI Wouldn’t it be a viable option to stake?

Another way of looking at it is to assume that risk is the "risk-free rate" of return, as we would have in the traditional financial system.

It's not completely risk-free due to the possibility of being slashed, but with a good validator the risk should be limited. Also, in the event of a "system failure" at Layer 1 (eg: a bug that caused token flooding, like Bitcoin in 2010), the bug will be fixed and transactions may resume as it affects the entire protocol.

On the lending side, the risks are bugs in the code (sensible contract risk) and counterparty risk (minimized risk for over-collateralized contracts, especially for pool investments). The reason lending is riskier is that a bug in the code (perhaps in a protocol that is not formally verified) is less likely to be recovered (parity is a good example).

Assuming that the borrowing risk is higher than the pledge risk, the market will demand a higher rate of return on borrowing.

I look at the rate at which ETH holders can borrow from Maker and the rate at which ETH 2.0 owners can lend their tokens.

secondary title

The ideal scenario is to issue mainstream tokens directly on the platform. This is possible with purely digital assets. Otherwise, the token will simply be a representation of another asset that exists either on another chain or in the physical world (eg: real estate).

Attracting new tokens on a new platform is notoriously difficult. Ethereum was very successful in 2017 as a platform for de facto ICOs. Binance.org is all the rage these days, but Binance has an “unfair” advantage in that Binance.com is the largest crypto spot exchange.

So, how can the new protocol get more new assets?

Consortium chain is one way, just like wBTC, but it is permissioned and requires escrow, so it will take a while to complete.

Another more feasible approach is to execute permissionless cross-chain contracts.

For example, bringing in ATOM tokens from the Cosmos hub is a great way to start. Cosmos has thought a lot about it and has published a reference guide on how to exchange messages between the protocol and ICS.

Furthermore, this is where the details of the consensus mechanism are particularly important.

In "follow the longest chain" consensus (Eth 2.0, Tezos), cross-chain transactions are more challenging due to the probabilistic nature of this consensus mechanism and the need to wait for "enough" blocks to be considered final . This is an important attack vector if low latency is required.

secondary title

An obvious consequence is that for many projects tokens create friction, sometimes without any apparent benefit. The most typical example is probably Bancor. On the one hand, Bancor has invented a very beneficial liquidity pool that works very well for token swaps, especially on low-throughput platforms. It is also very composable, allowing atomic transactions with other smart contracts.

But Bancor integrated a token as an intermediary for each deal, which was key to its $143 million funding round. It seems clear to the community now that this token will add friction and that the same kind of protocol without tokens is probably better off.

Let's take a look at an example from Uniswap:

A program sponsored by the Ethereum Foundation. With only $100k in grants, or 3 orders of magnitude (!) less than what Bancor raised, the project is now twice as liquid as before.

What lessons can we learn from this?

We should not overlook the value that Bancor creates. But it is clear that forks without tokens are the trend when tokens are not needed.

But who would make such an agreement in the first place if there were no economic incentives?

In fact, there are financial incentives in it. Uniswap benefits Ethereum and therefore ETH holders. So it is clear to us that such activities will be funded by the Foundation orCommunity-Driven DAOsLead, not by a team privately funded through a token sale.

secondary title

The atomicity of transactions is challenged in sharding protocols and more research is needed. Atomicity is not actually an option for sharding protocols due to the asynchronous nature of transactions across shards, the so-called "training and hoteling problem".

However, how are the initial steps to getting DeFi on sharding protocols different?

A token standard

The first step is a new standard for cross-shard tokens, "shard-erc20". The way tokens are transferred from one shard to another is by using a burn function on one shard (source) and a mint function on another shard (destination), which will accept a "proof of burn" as input. The first question that arises from this is how to account for the token supply? One way is that the first shard on which the token is initially deployed is the "master shard", while the other shards have a lightweight "slave shard". "chip" implementation, which only has the function of burning/minting tokens, so that the supply does not increase.

Even with a token standard, we can expect that, just like ERC20, tokens have their subtleties in their implementation. Is a set of signatures sufficient for acceptance of proof-of-burn and newly minted tokens? Could "slave" tokens also have an additional function such as tokenization/locking?

Tokens seeking cross-shard functionality will need to deploy a smart contract on each shard.

Centralization of liquidity?

For exchanges, liquidity brings liquidity. Therefore, pooling assets for trading-related activities may be an incentive. We think it makes perfect sense for shards to lead in certain combinations. Shard #1 is ONE-marked quote tokens (eg: wBTC/1, ETH/1, ...), shard #2 is DAI-marked tokens, etc.

Sharding utility can be maximized by offloading transactions to different shards. This might make sense for an orderbook model, but what about a long-term product market maker model like Uniswap?

If you think that a key property of Uniswap is composability. In the world of asynchronous cross-shard transactions, composability can only be done within a single shard. Therefore, the relevant model of Uniswap will be deployed on each shard and incentivize market makers to achieve balance on each shard. This will be very useful for price-insensitive use cases. Investors who wish to hold large sums of tokens will not use this type of transaction. But if our goal is to get a very small amount of protocol-related tokens and the fee needs to be paid in this specific token (such as MKR), then this is a good solution.

There are also some benefits to intelligently routing order processing. This is similar to how Kyber is built today, but instead of looking at the price of certain assets in different contracts, our goal is to look at the price of assets in each shard.

By incorporating a Uniswap on each shard, coupled with a kyber-like cross-shard liquidity solution, it means that if you want a lower price, you will have to do many cross-shard transactions, This will take some time. If you want your transaction to be executed immediately, you will lose the liquidity of your current shard, so it is difficult to trade at the best price.

secondary title

Besides the inherent security benefits of on-chain transactions being a viable solution, this could also solve many oracle problems. The idea was presented by Delfi during a hackathon project at ETHDenver, where it was awarded the "Most Innovative" award.

The idea is this: When there is no price available on-chain, or the price is too easily manipulated, as more transaction volume is generated on-chain, we can use the on-chain oracle as opposed to the off-chain oracle to Handle multiple DeFi smart contracts.

The advantage is that no trust is required for the on-chain oracle. The disadvantage is that the current traffic is very low, so the price given by the oracle on the chain can be easily manipulated, and as the traffic increases, the cost will become higher and higher. Lack of scalability is a key bottleneck that prevents more on-chain transactions from being possible right now.

In addition, most order systems do not specifically put prices on the chain because of the high cost of storage (on the contrary, 0x uses events to reduce costs), but another advantage of the sharding protocol is that the gas fee is less than that of Eth 1.0, so It is a little more feasible for such an agreement (such as 0x), the last price on the chain can be saved, and these data can be used with other agreements on the chain.

As we all know, DeFi is the most talked about use case in blockchain. Especially if we assume that a tool used to speculate on the future value of an asset (eg: an exchange) represents a financial product, where decentralization will be the trend!

This post was sponsored by Harmony collaboratorsNicolas BurteyA summary of the research work carried out.

The data presented is for educational demonstration purposes only and does not represent a recommendation as to the success or failure of digital assets.

Let us know your thoughts on #DEFI in our official community, or start a discussion with our team directly.