Introducing Aloe Blend

Context

On July 13th, we launched an Aloe Active Vault for the USDC-ETH Uniswap pair. This was the first test of our approach to decentralized market making on UniV3, and we had high hopes for it. But looking at the data for Aloe Active and its competitors, it’s clear that no active LP manager has been very successful so far. At the beginning of August, all major services were underperforming standard V2 LPs. These systems, which used a variety of mechanisms to manage user liquidity, were not as resilient to market conditions as an IL-minimizing, full-range position.

For a direct comparison of the active liquidity management products from Aloe, Charm, Visor, and Popsicle, see this Dune Analytics dashboard. Many thanks to community member Vividot for working on some of the first LP manager analytics, which helped us put this together.

Comparison of LP Manager performance over the time Aloe Active for USDC-WETH was live. No LP Manager outperformed UniV2, even when not accounting for V2 fees. (link)

We can’t speak for other liquidity managers, but for us the biggest problem here was scale. Without sufficient participation in the Aloe Prediction Markets, the Active Vault was unable to get out in front of price movements and avoid IL. This prompted us to look for ways of using UniV3 without the need for high-stakes crowdsourcing mechanics.

Through our research, we found a few ways to take advantage of UniV3’s finite-range liquidity while maintaining the same IL characteristics as UniV2. Here are a couple options:

• [Proposal #1] Deposit liquidity with the same density as UniV2, but only over a finite range. As long as you rebalance/re-center before the price leaves this range, behavior is identical to V2 as far as Uniswap is concerned. The leftover funds can be converted to interest bearing assets.

• [Proposal #2] Do all of Proposal #1, but deposit liquidity with some higher density. Constantly exchange earned fees for the token you have less of — thus maintaining a 50/50 inventory ratio.* The key is that only earned fees are traded; the principal is left alone. This ensures you’re not borrowing from your future self and making guesses about performance.

*This method also allows you to maintain other inventory ratios, like 70/30. We will explore this in a future post.

Aloe Blend is an implementation of Proposal #1. We chose to implement this idea first for the following reasons:
- It’s a subset of Proposal #2
- Hayden Adams expressed support for the concept

We launched the first Blend Vault (USDC-ETH) on August 4th and we will launch the second vault (ETH-OHM) on August 30th.

How exactly do these Blend Vaults work?

It was actually a lot of fun to put Blend together. From the start, we knew we had to use the same liquidity density as UniV2. The obvious question: What exactly is that density?

X axis is price and Y axis is % of funds earmarked for Uniswap. The solid blue line is the current price, and dashed blue lines are Uniswap position boundaries. An interactive version is available on Desmos.

To keep things simple, let’s look at the no-fee scenario for a UniV2 pool containing assets X and Y. At price pₗ=yₗ / xₗ we have the constant product formula xₗyₗ=k. Imagine someone trades Δy units of Y for Δx units of X. We now have xᵤyᵤ=k and a new price pᵤ=yᵤ / xᵤ, where xᵤ=xₗ−Δx and yᵤ=yₗ+Δy. Isolating Δx and Δy, we get the following:

Key Insight: In this context, the liquidity between pₗ and pᵤ is either Δx or Δy, depending on which price is considered “current.” If p≥pᵤ, then liquidity is Δy units of Y. If p≤pₗ, then liquidity is Δx units of X.

These results can be massaged to match equations 6.29 and 6.30 in the Uniswap V3 Whitepaper with the slight modification that ΔL=√k. That’s all it takes to mimic v2 on v3! This is all the background we need for implementing Aloe Blend.

Each Blend Vault has some inventory of assets X and Y, rₓ and rᵧ, but only part of that inventory is placed in UniV3 — Δx and Δy. Since the vault behaves like UniV2, we set rₓrᵧ=k and p=rᵧ / rₓ=1.0001ᵃ. The choice of pₗ and pᵤ is arbitrary as long as they encompass p, but to make the math work out nicely we use {pₗ, pᵤ}=pow(1.0001, a ± b). Combining these expressions with previously-derived equations yields the following:

Equations for computing the portion of Blend Vault inventory that should be deposited to UniV3. The symmetry and 1.0001 exponentials are quite convenient here, as they allow us to reduce on-chain computation and use existing TickMath library functions.

This means that once Blend selects the half-width b of its position, it can compute the amounts that need to be deposited to Uniswap. Once that’s done, leftover funds are converted to interest bearing assets.

How has the USDC-ETH Blend Vault performed?

For the first vault, we chose a blue chip pool (USDC-ETH, 0.05% fee) and conservative yield (Compound cUSDC and cETH). Over the past 3 weeks, this Blend Vault has consistently outperformed UniV2 — as expected. This is true even after accounting for its maintenance budget. That said, this is just the start. Once we incorporate more diverse yield protocols, Blend will be able to pull further ahead of UniV2 — the upcoming ETH-OHM vault is especially promising.

For more info, check out our new Aloe Blend Dune Analytics dashboard. But do note that this data only gets updated after a rebalance, so there aren’t many data points yet.

In conclusion…

We at Aloe see the need for efficient, passive liquidity provision that avoids the pitfalls of making impermanent loss permanent. Right now, UniV2 LP tokens are used all across DeFi: as protocol-controlled value, by VCs helping protocols bootstrap, and by everyday investors searching for maximum yield. To fully transition to UniV3 and maximize ecosystem efficiency, its crucial that we create LP tokens with similar risk profiles and higher yield — that’s Blend.

Thanks for reading. We’ll be sharing more on the Aloe roadmap, future vault launches, and feature updates soon. Stay tuned!

— the Aloe Team