[PRE-GIP 1] START Gearbox Liquidity Pools

Summary

Proposal for initial pools deployed on the launch of Gearbox Protocol

Abstract

Details of the launch were described in Gearbox Launch Article (The path to launch and first proposals | by Gearbox Protocol | Gearbox Protocol | Dec, 2021 | Medium). In this proposal, I want to raise the question of pools to be deployed and all base parameters of the protocol.

List of pools to add at the initial stage

I suggest creating 4 pools:

  • WETH
  • WBTC
  • USDC
  • DAI

These are fairly standard pools that have the most liquidity and are most commonly used in DeFi protocols.

ERC20 token addresses for pools

Interest rate model for pools

Borrow rate for pools is calculated by the formula described in docs. It is a standard model many lending protocols use. As a reference I consider Aave parameters (https://docs.aave.com/risk/liquidity-risk/borrow-interest-rate#variable-interest-rate-model-parameters). In Gearbox, the pool size is much smaller, so you need to set the utilization ratio a little lower in order to reduce the risks of a liquidity crisis.

Optimal Utilization ratio value Minimal borrow rate R_0 Borrow rate at optimal utilization ratio R_1 Maximal borrow rate R_2
WETH 70 0 4 60
USDC 70 0 5 100
DAI 70 0 5 100
WBTC 70 0 4 60

Pool Caps

Pool max Min collateral size Max personal borrow
WETH 600 WETH 0.3 WETH 4 WETH
USDC 3M USDC 1000 USDC 20000 USDC
DAI 3M DAI 1000 DAI 20000 DAI
WBTC 50 0.02 WBTC 0.4 WBTC

At the start, it is worth limiting the amount of capital to $2-3M. The problem that arises here is the risks of liquidity crisis if one whale takes out a large loan to his Credit Account. To avoid it I suggest limiting the max borrow amount per Credit Account to approximately $20k. This means that at least 125-150 credit accounts will be created per pool. Seems that this is enough diversification so that the system is not so turbulent at the start.

Other pool and credit manager parameters

  • Max leverage for all pools I suggest as 4. More aggressive leverage can increase the risks of the protocol at the start, while 4 looks like sufficient leverage for financial transactions.
  • Withdrawal fee 1%. This withdrawal fee can be considered as a backstop against system turbulence in the first weeks post-launch
  • APY spread is 10%.
  • Fee liquidation is 2%.
  • Liquidation discount 5%.
  • Default Swap protocol: Uni V3 as a leader in DEX trading volumes
  • Chi Threshold and Fast Check – need to be clarified by the team

Conclusion

I suppose the protocol launch won’t happen without active community participation, so let’s discuss all the necessary stuff including the list of pools (maybe some more pools should be added or maybe some of them need to be excluded for the initial stage) and their initial parameters.

Nick S from Mellow Protocol

17 Likes

Well done! Just a few comments and questions:

  1. Liquidation fee of 2% seems quite low. Abracadabra and Aave are both at 7.5% for ETH pools for example. I would propose bumping this up to 7% (2% to protocol, 5% to liquidator like here: Protocol Fees - Gearbox Protocol)
  2. Liquidation discount is the same as liquidation premium , correct?
  3. I’d like to see a larger launch. A smaller pool makes sense if we add an entirely new asset type, but all the ones you listed were part of the testnets / audits. I’d propose 25m per pool (proportional for all pools), for a total TVL of 100m at launch. Also, max borrow needs to be higher, or else you’re limiting people to depositing 1 ETH maximum - not exceptionally “worth it” with current gas fees. Increase that proportionally to the pool max increase, so 160k max borrow, etc. Otherwise you’re severely limiting the # of people who can even attempt to use gearbox in a capital-efficient manner. If we aren’t ready for people to use the protocol as intended, then we can delay launch and give more time to audits.
  4. What’s the reasoning for a minimum collateral size?
6 Likes

Noteworthy. I endorsed the strong comments with fair enough logic and reasoning behind it

Yes, eth fees must be taken into account.
Many people will not ape, unless they can reasonably expect to earn back at least tx fees.

  1. Current rates will give 7% in total – according to the WP, liquidation fees and liquidation discounts will summarize.

  2. Yes, in my understanding it’s the same.

  3. I think it’s always good to have some kind of guarded launch despite the audits for the new protocols as it anyway helps to improve the protocol security as the pools size limits increase gradually.

  4. Regarding the minimal collateral size – small positions are usually the most harmful because the same amount of gas is needed to close them as to close the bigger ones. This leads to the small positions becoming the bad debt. Thus in the case of, let’s say, $1000 with medium leverage of 2.5x will lead to a 5% liquidation discount equal to $125. It’s not so much, especially at the market surge with high gas price, but still enough to liquidate the account with the medium gas price.

2 Likes

re Pool Size - agree more with @0xn1ck - it’s better to start with small pool limit and then increase it every week in order to reach something like 50-100M in 1 month depending on demand of liquidity from borrowers side.
Maybe would be great to propose some formulas for pool caps like:
1st week: 3M
2nd week: if utilisation ratio > 70%, increase it to 10M
3rd week: if UR > 70%, increase it to 20M
4rd week: if UR > 70%, increase it to 40M

idk, @mikael wdyt on pool caps?

3 Likes

I’m in favor of this compromise. Lets us start small but also keeps us nimble enough to keep up with demand. If UR<70% then just check again the following week.

1 Like

Looks good, run it and let’s make history and become king of the gas wars

“Champion gear that I rock,
you get your boots knocked
Then attack you like a Lobster
And then lock sh*t down–doooown.
As I come and freaks the sound
Hardcore, but giving you more and more like ding!”
–Raekwon

I think because the proposed max borrow amount is $20K a more reasonable leverage to start with is 3, base on a traditional Maintenance Margin Requirement which is approximately 2x, or 50%.

With the current high gas fees you might want to consider increasing this to 3-4%, depending on an analysis based on what liquidations will cost during moderate, to high, to severally high (gwei) gas prices.

1 Like

5% goes to the liquidator, who pays the gas fees. 2% goes to the DAO.

3 Likes

I think we need to pay more attention to stablecoins LP

Definitely agree with Ilgiz, we should start with lowest possible limits, cause the first weeks liquidators will test their bots to liquidate accounts on time.

Thinking about caps, we should find balance between gas which makes liquidations profitable:
5% of total value > gas for liquidatin (400-800K per tx)

On the other side, big accounts could unbalanced system, so, lets move step by step

We use borrow rate model like in Aave, so my suggestion is to be close to them. Aave team is brilliant one and invested a lot of resources to figure out them. So, my advice is to start with numbers close they have in their models, which is proved in production

Unfortunately, Uniswap V3 is not an option. defaultProtocol should implement UniV2 interface. So, please choose between Sushi & UniswapV2 for ETH deployment. This design is done for make possible to deploy Gearbox on other L2 networks without UniV3 requirement.

Gearbox uses gas saving technic which is called “Fast check”. After each operation, it checks that credit account has enough value to pay borrowed amount back. In math sense, it means that hf >1.

This check requires a lot of gas. so we implemented a tecnic which is called FastCheck. (more info there: Fast check and healthFactor protection - Gearbox Dev)

This parameters are interconnected with Liquidation fee with following rule:

1 - chi^n < Liquidation fee (check this page: "Risk free long" attack - Gearbox Dev)

So, for 2% liquidation fee, which we checked on Kovan network, I can suggest following:

uint256 constant CHI_THRESHOLD = 9950;
uint256 constant HF_CHECK_INTERVAL_DEFAULT = 4;

However, all things should be setup by dao. If you found that its better to increase chi & hf_check_inferval to make less slippage alloweed without full check but increase period between checks - feel free to mention it

4 Likes

Wdyt? Looking to post GIP-1

1 Like

imo starting caps at $3M and increasing weekly as per ilgiz makes sense, as well as the $40k borrow limit - I think that’s a reasonable number to maintain stability at launch while also giving traders/farmers room to make it worth it for them

What about Min collateral size? 0xn1ck proposed ~$1000 Min collateral, Mikael suggests “we should find balance between gas which makes liquidations profitable: 5% of total value > gas for liquidatin (400-800K per tx)” – if I did calculations correctly 800K gas at 100 gwei is ~$320.

If I’m understanding correctly, that would require Min collateral size to be ~$6400? And if anything we’d like it higher than that since gas can ofc be >100 gwei? Thoughts?

On withdrawal fee (proposed temporary 1%), I worry that potential LPs will be turned off by having to pay nontrivial withdrawal fee after already taking a risk on a new protocol. It could ofc still work, I’m just trying to get in head of potential LP - please let me know of any comparable examples where non-degen farms included significant withdrawal fee, I could definitely be missing something.

Other possible options:
Withdrawal “cooldown” similar to Tokemak where withdrawals are free but take eg 24 hours to unlock.
Rewards claimable eg weekly to gently incentivize longer LP durations.
Some combination of fee + above.

Not sure how difficult it is to implement cooldowns technically, I assume a simple fee would be simpler. How confident are we that enough liquidity will be provided with eg 1% fee? Thoughts?

Moved to: