Dragonfly Partners: Can ordinary liquidity providers really make money in Uniswap V3?

Author: Haseeb Qureshi, Translation: Lu Jiangfei.

Uniswap V3 has just unveiled its mysteries, featuring many insightful concepts. I remember that at the beginning of 2020, Uniswap founder Hayden discussed related ideas, and now he has finally turned those ideas into reality! Here, I hope to elaborate on my understanding of Uniswap V3 and the potential impacts it may bring in the future. The thoughts are somewhat rough, so let's get started.

In the current Uniswap V2, all liquidity is provided according to the constant product curve from 0 to infinity, which leads to very low capital efficiency in Uniswap V2. We have already observed this issue in practice—Uniswap now has a locked amount of 5 billion dollars, but the daily trading volume is only 1 billion dollars (only 20%), which is quite low.

This also means that very little liquidity is centralized, but one trading pair may handle the vast majority of trades. A lot of capital is reserved to ensure whether assets can achieve 2x, 5x, or even 100x growth, and according to the constant product growth curve trend, there will still be some liquidity left for the market.

Uniswap V3 optimizes this situation, allowing liquidity providers (LPs) to set a price range for providing liquidity (for example, providing liquidity when the ETH/USDC trading pair price is between 1600 and 1800 dollars). When I heard this idea, a question arose: How will Uniswap complete all these operations on-chain? Because you need a better tree structure to track all liquidity providers' positions.

Uniswap V3 provides the answer: quantization!

They did not perform inefficient, fancy mathematical calculations but instead broke down the overall price curve into a bunch of "tiny buckets," referred to as "ticks" (though I prefer to call them "buckets"). Each "tiny bucket" acts like a small automated market maker (AMM), with a very small price range, such as the price range of the USDC/ETH trading pair being 1705-1710 dollars.

You can imagine that each "tiny bucket" has its own independent liquidity provider. If a trade exhausts the liquidity in one "tiny bucket," the remaining trades will enter the next "tiny bucket." Low-slippage trades will not be allowed to cross multiple "tiny buckets" (each "tick" has a small capacity, such as only occupying 0.1% of the liquidity pool).

This means that as a liquidity provider, if you provide liquidity in the price range of 1600-1800 dollars for the USDC/ETH trading pair, this liquidity will actually be deposited into 100 "tiny buckets" (depending on the calculation results). Many people have noticed: this model looks a lot like an order book, and yes, it is somewhat similar!

In a sense, Uniswap V3 is tailor-made for Layer 2. Once Uniswap V3 goes live, there will be less capital on Layer 2, but active trading and position management, as well as the costs of order book-style trading, will be lower. Uniswap V3 can "elegantly" enhance concentrated liquidity and capital efficiency.

However, for trades that cross multiple "tiny buckets," the costs may become higher (a token swap between 3 "tiny buckets" is like swapping tokens on 3 Uniswap platforms). Additionally, the transaction costs for withdrawing and depositing liquidity between multiple "tiny buckets" will also be more expensive (though it doesn't seem to be a big issue on Layer 2). Trail of Bits provided some insights in their audit report, as shown in the image below:

Assuming there are many liquidity providers in one "tiny bucket," will the accounting costs become higher? For this question, audits may pay more attention, but I don't fully follow the logic here because I don't see any cycles with liquidity providers (is this issue only applicable to very small "tiny buckets"?)

Uniswap claims that V3 has higher gas efficiency than V2, but this statement may only be correct under certain conditions (for example, when the number of "tiny buckets" crossed in a trade is not many). Providing liquidity will become more expensive, but fortunately—it's meaningful for users to optimize their gas costs and liquidity provision costs (see the audit opinions below):

Okay, now providing liquidity is more "intelligent"—you need to set a price range—but if this model can really generate more liquidity, wouldn't that be a good thing for everyone?

Traders can make money! Market makers can make money! However, ordinary liquidity providers may not necessarily make money!

Some people have already noticed this issue.

If you are a relatively lazy liquidity provider, you might set a wide price range, such as 1000-3000 dollars, and then deposit funds within this range. Next, your liquidity will be evenly distributed across each "tiny bucket," but the amount of funds allocated to each "tiny bucket" is actually very low.

The problem is that very few trades can be completed in one "tiny bucket," and the vast majority of trades will cross several "tiny buckets" (the pricing of these "tiny buckets" is basically set around the mid-market price), which means these "tiny buckets" will become very crowded, as almost all fees will be generated in these few "tiny buckets," so that may be where all the professionals keep their liquidity.

For market makers, miners, and other DeFi professionals, they will set their liquidity provision positions near the mid-market price and move in and out of some high trading volume "tiny buckets." Meanwhile, ordinary retail trades will be covered by a large number of "tiny buckets." This means they will ultimately receive a relatively low share of the fees!

Notice anything? In the old version of Uniswap, the liquidity provision mechanism that required no thought and allowed for "buying on the go" made it very easy for anyone to provide liquidity, as there was no competitive mechanism for providing liquidity in the old version of Uniswap. In this "foolproof" strategy, everyone shared the liquidity pool's profits proportionally.

But in the new version of Uniswap, every liquidity provider needs to compete, and if your strategy is not good, you will receive very low returns. Under this model, it is estimated that almost all fee income in Uniswap v3 will be taken by organized professionals. In a sense, Uniswap V3 has become quite strange, somewhat like Kyber—many market makers have to frequently update on-chain prices.

Uniswap explained this in a blog post, stating that they expect roles like "Uniswap strategy integrators" to emerge in the future, similar to Yearn, where these "Uniswap strategy integrators" will aggregate retail deposits and then distribute the profits to them, thus achieving cost rebalancing. However, this strategy is really inconvenient to implement!

Well, this makes me think of NFTs. Now all NFT positions are like snowflakes; no two are the same, and each NFT has different price limits, so they are no longer fungible tokens. Looking back at Uniswap V3, doesn't it feel like each Uniswap V3 position is like an NFT?

My first thought is: oh, a bunch of Uniswap derivatives are coming. In the future, you might not be able to easily borrow Uniswap tokens on MakerDAO or Aave, because each Uniswap position is different. Of course, if someone marks a generic Uniswap LP position (perhaps through one of the aggregators?), it might still be somewhat useful, but doing so would disperse liquidity.

My second thought is: wait, won't this disrupt liquidity mining programs? Because now Uniswap NFTs do not prove that you are actually providing any liquidity. For example, I could create a very outrageous limit order that never falls within the range, which should mean I would never receive liquidity mining rewards.

In fact, the reason Uniswap has achieved its current success is largely due to liquidity mining. Fortunately, there is a way to address the issues mentioned above: you cannot allow users to subscribe through Uniswap and only reward those who hold LP NFTs; this model is incorrect. Instead, they must deposit into Uniswap through a shared contract, which will specify a reasonable price range (e.g., 80-120 dollars) and ensure that the value remains effective at all times. If the mid-market price gradually declines, it may be necessary to occasionally reset the new price range.

This mechanism can also make market-making supported by the protocol easier, as Fei has already successfully done! (I would love to see @MakerDAO try something like this, such as using a strict price limit of around 1 dollar on DAI to deposit Uniswap LP positions.) In fact, many great ideas can be extended from here.

Each Uniswap liquidity pool will feature adjustable fees (Uniswap will eat Curve) and "cute" anti-fork permissions.

Finally, I want to say that Uniswap V3 indeed leaves a deep impression, and my understanding is not absolutely correct.

Thanks to Uniswap co-founder Hayden Adams and Paradigm research partner Dan Robinson.