Liquidity Mining 2.0 (LM2)

The super hot DeFi summer 2020 was fired up with the launch of Compound’s governance token and was totally driven by the concept of liquidity mining (aka yield farming). Without any doubt, liquidity mining propelled DeFi into the spotlight so more people can see the power of DeFi over CeFi and TradFi. On the other hand, the abusive usage of liquidity mining and its damage to many unsophisticated token buyers absolutely hurt the reputation of DeFi. The net benefit of liquidity mining for DeFi as a whole is still up in the air.

Copy trading bots from the top traders. Try it for FREE.

There were a few novel usages of liquidity mining such as YAM’s pool1 design to bootstrap a crypto community, Sushiswap’s vampire attack to bootstrap an AMM’s liquidity. But overall, most projects were simple copycats or straight forks. Sadly, there are also many scams riding the market narrative.

Clearly the current design of liquidity mining is not optimal and is one of the main root causes that lead to the unsustainability of almost all projects’ liquidity mining programs. To make things even worse, during the last 2 years, there was no effort whatsoever to even try to fix the flaws in the liquidity mining design.

Liquidity mining, in the narrow context of liquidity to support trading on Automated Market Makers (AMMs), is a token incentive program that is designed to attract Liquidity Providers (LPs) to provide liquidity for specific trading pairs/pools on AMMs.

Synthetix pioneered the on-chain implementation of reward token distribution to LPs for its sETH Uniswap pool. At a high level, to earn rewards, LPs need to first provide liquidity for sETH pool on Uniswap and then stake their Uniswap liquidity tokens into the staking reward contract, which was created in 2019. (This smart contract might be the most widely used and forked contract ever due to the DeFi and yield farming mania.) Reward tokens are distributed fairly to LPs based on their % of staked liquidity tokens against all tokens staked by all LPs.

From tokenomics incentive design perspective, the liquidity mining approach pioneered by Synthetix distributes reward tokens based on the size of liquidity positions and let’s define this approach as Liquidity Mining 1.0 (LM1). Based on the results, such an incentive program worked for Synthetix to achieve its goal to attract more users to mint more sETH.

LM1 becomes the de facto liquidity mining design and implementation. It enables many projects to solve the liquidity challenge to a certain degree, at least in the beginning. However, LM1 has many issues that contribute to its unsustainability.

First, reward tokens are being distributed to LPs even though there might be no trade or few trades, meaning the liquidity is not really used. From tokenomics point of view, using project tokens to incentivize liquidity is expensive for most projects using it, since the incentive won’t contribute much to the growth of the protocol economy. And when the liquidity is not used, it makes the incentive program worse.

Second, in many cases, multiple pools need to be incentivized. The existing approach is to allocate a certain amount of reward tokens to each pool without considering the contribution of each pool, e.g. how many trades and how much trading volume are executed in each pool. The reward allocation decisions are either made by governance vote like in Curve, Balancer or by the team like in Sushiswap, which are sometimes political or arbitrary.

LM1 can be improved and a much better liquidity mining incentive design is to distribute reward tokens based on the AMM trading fees earned by liquidity positions. This design is fundamentally different from distributing tokens based on the size of liquidity positions, and let’s define this approach as Liquidity Mining 2.0 (LM2). Clearly, LM2 resolves the two big flaws in LM1 raised in the above section.

First, during the fixed token distribution intervals, if there is no trade, then LPs will not earn AMM trading fees. No fee, no distribution of reward tokens. In addition, it also discourages LPs to supply more liquidity than a project needs. Using LM2, projects won’t waste their precious tokens for non used liquidity and hence reduce the token inflation and downward token pricing pressure due to liquidity mining.

Second, there is no need to allocate reward tokens to multiple pools manually either through governance token votes or team decisions. These manual approaches create wrong incentives to LPs and treat liquidity in different pools unfairly. Using LM2, if a LP position in any pool earns more AMM trading fees, then more reward tokens will be distributed to that LP position, as simple and fair as that!

Typically, projects issue an ERC20 token (mostly used as a governance token) with a token amount cap and allocate a portion of these ERC20 tokens for liquidity mining programs. During liquidity mining programs, a fixed amount of tokens are distributed during a fixed time interval, for example per block.

In LM1 implementations, the fixed amount of tokens per time interval are distributed evenly over the total amount of all LP tokens staked for liquidity mining. Each staking LP will earn the amount of reward tokens based on the amount of LP tokens it staked. Anytime there is a change in the amount of LP tokens staked during the time interval, ratios are updated accordingly and rewards are updated accordingly as well. This implementation ensures a fair distribution of reward tokens among all LPs that participate in liquidity mining programs.

Unfortunately, distributing the fixed amount of tokens per time interval based on trading fees collected by LP positions is actually very hard. The trading fees are driven by two dynamic unpredictable parameters during the fixed time interval: 1) when trading fees are generated and earned by LP positions is dynamic and unpredictable because nobody can predict when traders will trade; 2) how much trading fees are generated and earned by LP positions is also dynamic and unpredictable since the size of trades are dynamic and unpredictable too. Hence distributing a fixed amount of tokens based on two dynamic and unpredictable parameters will create an unfair distribution of reward tokens among all LPs that participate in liquidity mining programs.

One possible solution is to adapt the relevant data distribution models for the two dynamic parameters — the time of trades and the size of trades. And then develop an on-chain implementation that updates the model dynamically based on each new trade and distribute reward tokens accordingly. The solution will be much closer to a fair distribution of reward tokens among all LPs that participate in liquidity mining programs. It is not rocket science but may require Ph.D. level big brains.

There are definitely other approaches to implement LM2. One better approach is to adopt a new token model that not only has better tokenomics but also makes it easy to implement LM2. Innovative solutions are coming and please stay tuned! Follow Double on Twitter.

Liquidity mining propelled DeFi into the spotlight. Current liquidity mining design and implementation has a few flaws and contributes to the unsustainability of the liquidity mining programs. Improvements can be done and reward tokens should be distributed based on the trading fees earned by LP positions instead of basing on the size of LP positions. It is a hard problem to distribute reward tokens based on the trading fees due to the current token model and reward distribution schedule. Innovative solutions are coming. Please stay tuned.

Join Coinmonks Telegram Channel and Youtube Channel learn about crypto trading and investing