Voltz Protocol: A Guide For LPs

June 1, 2022

The definitive guide for Liquidity Providers on Voltz Protocol.

Key Principles

Voltz Protocol is a non-custodial automated market maker for interest rate swaps. The AMM is split into two core components; a vAMM that is used for price discovery only and a Margin Engine that allows traders and LPs to trade with leverage. 

In terms of LPing, the vAMM utilizes concentrated liquidity concepts from Uniswap v3. This means LPs deposit liquidity within tick ranges. However, relative to Uniswap v3, there are a few key differences:

  • Leverage: the Margin Engine allows liquidity providers to deposit margin as collateral to create notional that goes into the vAMM as liquidity. This notional trading liquidity can then be used by traders, which generates fees for LPs.
  • Margin Recycling: LPs generate the most fees where there is balanced fixed and variable trading activity. This occurs due to the “Margin Recycling'' mechanic within the Voltz Protocol AMM, which identifies VT (Variable-Taker) and FT (Fixed-Taker) positions that net off to release LP margin back into the vAMM to continue to collect fees. This is outlined below and in further detail in the litepaper.
  • Liquidation Risk: whilst the ability to take leverage contributes to capital efficiency for LPs, it also exposes LPs to liquidation risk as outlined in more detail in the risks section below.
  • Single-Asset LPing: with rate-markets the fixed and variable rates are paid in the underlying (e.g. on a cDAI pool, both fixed and variable rates are paid in DAI). This means an LP only needs to deposit one asset (e.g. DAI) to create either side of the market.
  • Funding Rate Risk: single-asset LPing means LPs aren’t exposed to impermanent loss. However, they are exposed to funding rate risk - where notional can be locked into a swap that doesn’t get netted off. In this instance if the LPs leg is out of the money, e.g. they’re locked into a swap as a FT but the variable rate they’re paying is higher than the fixed they’re receiving, they would have negative cash-flows. In this instance there is a chance the fees generated aren’t high enough to offset the negative cash-flows meaning the LP would lose money. This is explained in more detail in the risks section below.
  • Term based pools: pools have terms, meaning capital that’s locked into a position, i.e. hasn’t been netted off via the margin recycling mechanic, can be locked until the end of the term. At the end of the term LPs can settle their position. Fees can be collected at any time.

Return Potential

LPs collect fees every time a trade is made with their liquidity. This means LPs generate the most fees where there is balanced trading activity of VT and FT positions, since the positions keep netting out such that the LPs liquidity can go back into the vAMM to generate more fees. 

This means the following variables positively impact LP returns:

  • Trading Volume: fees are generated each time there’s a trade, meaning balanced trading of VT and FT generates the most fees due to margin recycling.
  • Proportion of Activity Liquidity: If liquidity is concentrated around the area of most VT and FT trading, this means more liquidity is actively generating fees. However, tight bands also expose LPs to risks outlined below.
  • Leverage: LPs generate fees from their notional, but only have to deposit margin to create the notional that goes into the vAMM. This means LPs can generate higher fees per margin deposited by taking more leverage. However, this also exposes LPs to liquidation risk as outlined below.
  • LP fee parameter: this is a parameter set by governance and defines the fee LPs receive on their notional. 
  • Protocol fee parameter: this is also a parameter set by governance and defines the proportion of LP fees that go into the protocol treasury.
  • Time to maturity: LP fees are then effectively pro-rata’d based on the time to maturity.

As explained in detail in the litepaper, this means LPs returns are driven off the following formula:

Fee Parameters

Initially, stablecoin pools will have a fee parameter of 𝞴 = 0 (i.e. none of the LP fee will go to the protocol treasury) and 𝞬 fees, the LP fee, of 0.003 (i.e. 30bps). These fees will be built out per pool in the future, with any changes communicated well in advance.

These fee parameters were designed to give LPs sufficiently high fees to incentivise providing funds, whilst reducing the risk to liquidation and funding losses, as well as effects of excessively high leverage. When considering alternative fee parameters, it was important to balance the expected LP APYs, which are positively impacted by higher fee parameters, with the negatively-impacted trader APYs for the same higher fees. Comparisons between the LP and VT APYs, assuming 25% of liquidity is traded per day (consistent with the typical trading volume of Uniswap) are summarized below, as a function of the leverage provided. To strike a balance between LP and trader APYs under a variety of leverages, 𝞬 = 0.003 (i.e. 30bps) is seen as an effective compromise, offering strong returns for both LPs and traders in the platform. 

Analysis on these parameters will be ongoing.

LP P&L Analysis

In order to analyze the returns of liquidity providers, we have run a few simple simulations. Before we proceed to the simulation results, it is important to note that we need to make a few assumptions.

We assume that the proportion of LP notional traded per day is 25%. This is based on Uniswap v3 trading data. Additionally we assume the liquidity provider has entered into a 60 day IRS pool as soon as it is initiated.


We also assume the fee factor (gamma) is 0.30% (of time-scaled notional). To be conservative, we will assume 20x leverage. Finally, we assume an ideal case scenario where all of the LP margin is recycled, meaning that all variable and fixed cashflows are continuously netted out by VTs and FTs consuming liquidity along the fixed rate range of the LP.

A critical consideration for LPs is the fixed rate range within which they are providing liquidity. An ideal case scenario is where the fixed rate range is aligned with the range that is one or two standard deviations away from the historical equilibrium APY in the underlying yield-bearing pool in order to ensure IRS trading activity happens within the LP’s fixed rate range to capture fee income.

For example, if we wanted to capture fee income from trading activity around 2020-12-25 in the plot below, we would set the range to be around 2% and 20%.

Using the assumptions above - of 20x leverage, 25% of LP notional traded per day, and 100% recycling - we can generate the following plots for LPs.

The plot below depicts the Cumulative Fee Income from a margin deposit of 100 Dai, we can see that the cumulative P&L is increasing over the entire term of the IRS pool. However, the rate of P&L increase starts to decline closer to maturity. This is driven off the fact the LPs fee is scaled based on time to maturity. In this scenario, the LP would generate 55% APY.

Below, we added a few additional plots that show the relationship between LP APY and a few variables of interest. Plots 2 and 3 show the relationship between LP APY and the proportion of notional traded and leverage respectively.

Risks

Whilst LPs have the ability to generate significant APYs through Voltz Protocol, they are also exposed to risks. Due to single-asset LPing, there is no risk of impermanent loss. However, there are the following risks:

Funding Rate Risk

When LP notional is used by a trader it is momentarily locked into a trade. For example if a FT uses the LPs margin, then the LP effectively becomes the VT on the other side of that trade. In an instance where a VT comes in and trades the same position, it nets out and the LPs margin goes back into the AMM to keep collecting fees. This is the process of Margin Recycling explained above. 

However, LPs are exposed to a risk described as “funding rate risk” where a corresponding trade never takes place to net out the position. In this instance, the LP will remain locked into a swap until a trade arrives to net it out, or until the end of the term.

If an LP is locked into a swap, the fixed rate of the trade will be known, but the variable rate will move according to the underlying variable rate of the pool (e.g. cDAI). There is a chance this variable rate moves against the LPs position, meaning the LP would be paying out more than their receiving. There are two scenarios where these negative cash flows can arise:

  1. The LP is locked in as a VT: If an FT trades with the LPs notional, but a corresponding VT trade never arrives, the LP will be acting as the VT. This means the LP is selling fixed in exchange for variable. If the variable then drops below the fixed rate, the LP is out-of-the-money. 
  2. The LP is locked in as a FT:  This is the opposite of the above, meaning the LP is selling variable in exchange for fixed. In this instance if the variable moves above the fixed, the LP is out-of-the-money.

Scenario where LP is locked into a VT Position (selling fixed, receiving variable)

In this scenario the worst case cash flows for the LP are bounded by:

  • The average of the fixed-price tick range the LPs liquidity is deposited into. This is the maximum rate the LP is “selling”.
  • The lower bound for the variable APY, which is zero. This is the minimum rate the LP is “receiving”. 

This is reflected in the following formula for a VTs cash flow (assuming a 60 day pool): 

Cash Flow = margin*leverage*(variable APY - fixed APR)*(60/365)

For the purpose of loss modeling we’ll assume the margin is 1 and leverage is 1. If the LP deposits liquidity in a low fixed rate range, where the rate paid out is 1% on average, then their loss is bounded to 1%*(60/365) = 0.16%.

These losses would change as the leverage increases and as the delta between the fixed rate paid vs. variable rate receives increases. This is shown below, where we assume a 60 day term and a margin of 1: 

Scenario where LP is locked into an FT Position (selling variable, receiving fixed)

In this scenario the worst case cash flows for the LP are bounded by:

  • The upper bound for the variable APY, which is bounded by the interest rate model on aave / compound. Theoretically this can be unbounded, but historical data here is helpful (as shown above). This is the maximum rate the LP is “selling”.
  • The average of the fixed-price tick range the LPs liquidity is deposited into. This is the maximum rate the LP is “receiving”. 

This is reflected in the following formula for a FTs cash flow (assuming a 60 day pool): 

Cash Flow = margin*leverage*(fixed APR - variable APY)*(60/365)

Potential LP P&L can be modeled by looking at the delta between the Fixed APR and Variable APY and the leverage taken. This is shown below, where we assume a 60 day term and a margin of 1: 

Liquidation Risk

As with Traders, LPs deposit margin to create notional that goes into the vAMM. This means LPs can be exposed to liquidation risk, since they’re taking leverage. Importantly the initial margin requirement is always higher than the liquidation threshold, meaning once LPs have deposited margin they can then continue calling the Voltz Protocol SDK to analyze how close their position is to the liquidation zone. There are then two scenarios where a liquidation can occur:

  1. Where the LPs trade doesn’t get netted out and is “out-the-money” for a long enough period to enter the liquidation zone. This is how the funding rate risk plays out if the rates don’t move in favor of the LP. 
  2. Where the LP has a lot of virtual liquidity in the vAMM, but not enough margin to support the trades should that liquidity be consumed. In an extreme example, if the LP had notional in the vAMM but $0 of margin then the notional clearly shouldn’t be available to trade with since it would leave the system insolvent.

Liquidations occur differently for these two scenarios.

In the first, this works the same as a VT of FT liquidation. A liquidator bot is able to liquidate the LPs position by triggering an unwind and takes a proportion of the LPs margin as a reward.

In the second, a liquidator bot can burn the notional of the LP in the vAMM and then take a proportion of the LPs margin as a reward. At this point any unnetted (i.e. un-recycled) trades would also be unwound as above. This second scenario means the LPs notional is removed from the vAMM so it can’t be traded, but the margin is left untouched except for the LPs reward that’s taken.

LPs can monitor how close their margin is to the liquidation threshold by calling the Voltz Protocol SDK and pulling the liquidation margin requirement and comparing that to their current margin.

Just like in protocols like Aave or Compound, the liquidation penalty (or the reward of the liquidators) is dependent on the underlying asset / yield-bearing pool. For the low volatility stablecoin pools, the liquidator penalty is 5%, which is in-line with the penalties in stablecoin lending pools on the money market protocols.

Extended Use Cases

Acting as an LP can also enable sophisticated traders to implement the following strategies:

  • Caps / Floors: by placing liquidity in a tick range where you hope it never gets netted off, you can effectively create a Cap or Floor. 
  • Covered Call Payout: As with Uniswap v3 you can use Voltz Protocol to replicate a covered call payout as outlined here.

We’ll provide further analysis of these strategies in due course.

Please see the user interface guide for instructions on how to provide liquidity via the app interface.