[GIP-22] Gearbox V2 Liquidity Mining Programs

UPDATE: changes are here:

UPDATE: all the details and explanations + how to claim, are here:


https://snapshot.org/#/gearbox.eth/proposal/0x9a6d7ae302241e0132dfd0380a4e20a32eb16fe2c187e54962732eb202a73508

The final numbers have been adjusted to reflect the discussions.

Summary

The purpose of this proposal is to put in motion conversation regarding:

  • An incentive-aligned LM (liquidity mining program) that enables CA (Credit Accounts) to function
  • And a retrospective LM drop for the LPs & CAs that have been present till date

Before we begin, let’s answer the question on why LM?

Because V2 is about to give off fireworks! If you missed some snippets of the upcoming products, check out some of the latest medium Alpha series. To recap, the liquidity in the passive pools enables the CA (Credit Accounts, aka leverage takers) to function. With the upcoming Leverage Ninja mode, the demand side of liquidity has over 100 high collateral size (100K$ or $150K min borrow, to be voted on) users that’ll require liquidity for leverage. That leads to a sudden requirement in pool values which will be multiples of what we currently have. That is a great “problem” to have though, and such utilization without major incentives for the CA side will prove strong PMF. But to enable it, we need to bootstrap the other side.


1. Incentive aligned LM for passive LPs + ninjas CAs

The incentive aligned Liquidity Mining program (we can come up with a fancy name later) for passive LPs is designed to incentivize depositing funds in passive Earn pools regardless of the organic APY on the said pool. Given the upcoming Curve/Convex/Lido integrations, due to the bad market, farming yields have been compressed to as low as 3-4% for stablecoins. As such, the utilization curve is planned to be adjusted to have rates fluctuate on the Aave-like formula but in the range of 1-2%. In a bull market or as DeFi grows slowly, these rates can be adjusted to a higher benchmark as yields will also go up, say 3-5%. Compressing the curve will help ensure that the borrow-APYs don’t render the strategies added to V2 unusable. The LM will be passive in nature with no liquidation and thus will be a low risk play for the LPs.

NEW: prospective Leverage Ninjas have asked for small LM for CA side too, not as a financial incentive but just as an alignment for users to be able to get a piece of the protocol governance. That didn’t seem too unfair, and the spending required for it is extrmeely small too.

WEN & how? Logistics & claims.

LPs start on Oct 24.

LP LM will begin before the launch of Leverage Ninja mode to ensure that ninjas have enough liquidity when the protocol does open up. That is, with votes to be taking place by October 23, the LM for LPs will start on Monday October 24th at a block to be announced earlier on that day. But if you don’t want to wait, you can already LP at any time, it’s just that the clock will start ticking on Monday.

Similar to most LM programs, the DAO will vote on a certain number of tokens per block which will be divided between the LPers based on the share of value they have in the pool. The same can be looked at as a variable APY. As the overall pool value grows, the APY will drop which could lead to some LPers dropping and GEAR APY going back up again. So while the incentive is pushed by the DAO, the APY on the GEAR LM is decided organically by the participants. We didn’t find a reason yet to make any different complex program before GEAR token economics are revamped.

CAs start on Nov 4.

LM for CAs will start after multisig executes all transactions and re-connects the V2 protocol in mainnet. That is expected to start on approximately October 27. Exact block TBA.

As for “how”, the new interface will reflect the APY LM numbers based on the [your share of liquidity in a given pool / total liquidity in that pool * tokens per block allocated]. The usual simple LM really. Every 3-4 weeks, DAO devs (or any external dev) can deploy a new merkle contract claim and allow for $GEAR claims. Meanwhile, participants shall be able to check their accrued rewards & APYs in the interface, to keep track of it 24/7 even if the claims are not 24/7. As GEAR (when & if) becomes transferable, this operational overhead can be replaced with a regular farming distributor.

Numbers: 10% GEAR APY extra for LPs; 1% for CAs * IF optimal level

We have estimated an LP value requirement for Lev Ninja mode and have curated the LM structure in order to yield 10% APY at what we expect our ideal value to be: at approx $55M TVL.

Below is the table that outlines the exact details for the LM program.

You can find the link here, and just copy the sheet for yourself to play with the numbers.

Please gib feedback or your inputs after going through it.

The logic for such allocations is as follows:

  • ETH strategies are higher in APY and are more organic in external protocols, hence larger % total. It’s easier to source Ethereum yields these days rather than stablecoins.
  • WBTC not having that much demand or external farming APY (although can be used for shorting/hedging then). All can be adjusted in the next stages anyway.

Since the LM happens after significant developments post the last round, the FDV is taken at a smol increase to the 200M$ FDV for strategic rounds. The spends are calculated to produce a 10% APY on what we have determined to be an expected TVL of the pools (55m$).

The calculations at a monthly level translate to 0.24% GEAR emissions / Month.

Total Spend? Duration?

The total spend depends on the duration for which the DAO decides to run the program. Do we do it for 2 months till V2 is launched and then focus on balancing APYs organically? In that case we spend 0.48% of the supply. Or does the DAO run it for a full year? In that case we spend 2.91% of the supply. This is for the DAO to decide, it’s an open-ended program.


2. Small retroactive LM for early LPs

The early LPs are the addresses that provided liquidity to the pools between the DAO launch in December 2021 to the 13th September pre-merge period. These LPs have helped us bootstrap initial liquidity, test out the security of the pools and helped the CAs become functional.

In return, the APYs on the pools have constantly been less than 1% while the fee on withdrawal was 1% and thus made exiting a hassle. That is about to go away in a few weeks, the roadmap for that will be posted in about a day or two. Stay tuned to confirm this.

The retroactive LM rewards the people for the staying and helping the protocol with the above with a small 5% APY in GEAR pro rated to the time they were in the pool. You can also see a smaller part being dedicated to Credit Accounts users, on a liquidity-borrowed-over-time basis. So there was no way to sybil as it all depended on the liquidity or your borrowed amounts.

You can find the link here, and just copy the sheet for yourself to play with the numbers.

What are the expenses?

Well, none! The entire supply here comes from the savings from the bot and sybil detection that took place so nothing new or additional to come from the DAO wallet. So it’s coming from saved GEARs and repurposed for this cause! See more in the token part of the docs: GEAR Overview - Gearbox Protocol


Conclusion

The LM will cost 0.24% monthly to ensure CAs get the funds required to function while the retrospective LM will have a one time 0.5% spend for a course of the last 9 months - that come from early tester tokens, and are not (!) a new DAO expenditure. As CA demand either goes up or down, this can be adjusted by the subsequent DAO votes. The tokens for future LM come from the DAO reserves. In order to proceed, the DAO will discuss:

  • Whether or not to do LM
  • Both LMs to be done or not
  • The parameters for the LM
  • The duration for the LM

Let’s go

4 Likes

Hi,

Maybe I have my numbers wrong, but assuming the following APR’s (at time of writing) of Convex related pools, wouldn’t stable-asset farmers just provide liquidity if they can get 10% without risk of liquidation?

Net APR = Farming APR * xLeverage -(xLeverage - 1) * Lending APR

Lending APR 1.5% APR (assumed)

Convex 3pool
0.2% APR @ 5.5x Lev = Negative 5.65%

Convex Curve LUSD+3CRV
2.31% APR @ 5.5x Lev = 5.95%

Convex Curve sUSD+USDC+DAI+USDT
3.66% APR @ 5.5x Lev = 13.38%

Convex Curve gUSD+3CRV
2.96% @ 5.5x Lev = 9.53%

Convex Curve FRAX+3CRV
2.93% @ 5.5x Lev = 9.365%

One consideration to the above;
If Gearbox DAO borrow rate equals that of sUSD on Aave (3.48%) then the above 13.38% APR is obliterated down to 4.47%

If the Leverage rate is increased to 7x you get;

Convex Curve LUSD+3CRV
2.31% APR @ 7x Lev = 7.17%

Convex Curve sUSD+USDC+DAI+USDT
3.66% APR @ 7x Lev = 4.74%

Convex Curve gUSD+3CRV
2.96% @ 7x Lev = 11.72%

Convex Curve FRAX+3CRV
2.93% @ 7x Lev = 11.5%

Obviously, the returns are based on the value of the reward tokens (CRV and CVX) and prevailing market conditions will dictate this, but I don’t think we should be planning to incentivise lenders more than the borrowers can achieve using leverage. Otherwise, prospective borrowers will just plop their $$ into the lending side and sit tight to collect the ~10%APR in $GEAR.

Do we need to consider incentivising borrowers as well if ROI of stablecoin farms @ a specific leverage is < 10% APR?

This can be achieved through reallocating existing Lender directed $GEAR to borrowers as I think 10% APR $GEAR LM-ing is overly generous so maybe a reduction to ~7.5% APR in $GEAR?

ETH farming looks good though as is;

Convex Curve ETH + stETH
7.3% APR @ 5.5x Lev = 33.4% or 42.1% @ 7x Lev.

Just my two cents but thought I’d throw it out there.

2 Likes

GM! Amazing proposal and also the comment by koala.

  1. Retroactive: the amounts and logic make sense. It’s not too large but it’s honouring early users regardless. Especially because the system ended up being low-risk after all due to tiny limits. Since the program was clearly communicated from early days, there was no sybilling possible also due to the fact that liquidity was the factor for GEAR distribution, and not a sheer number of accounts. Disclaimer: I had some LP to support the protocol.
  2. Future LM for V2: makes sense. I think it’s quite subjective whether it should be 7.5% or 10%. I would be on the side of 10% though, because it’s better to have over-supply problem than to have less supply. This wasn’t the case for DeFi protocols which optimized for TVL previously, but for Gearbox it is. And here is where this point 2 is also a reply to koala’s worry above:

If there is 50M or 200M in the passive pools, ALL will be utilized. The reason for that is that Credit Accounts offer true organic PMF [assumption though]. That means, if you can borrow at 2% rate or so (with the suppressed utilization curve optimal level) - it is a no-brainer carry trade almost. Like, you can just be an arbitrage bot even. That can service more users, more protocols, etc.

But why allow for someone to make money, if the protocol obviously makes less? The usual answer honestly grow liquidity, use that to go to larger players and institutions, get protocol integrations done quicker. No one wants to build together with a $5M TVL protocol after all.

So if we have this “problem” of over-filled TVL pools, that’s totally fine.

But it makes sense to LM just a tiny bit for Credit Accounts side, let’s say in the 1% range. Not as an interest rate game at all, but to align users and have users have ownership of the protocol. It doesn’t add much to spending, but it aligns both sides then.

1 Like

The market will set the rate. Ie, we can’t just “reduce to 7.5%” APR. We offer up an amount of GEAR, and we think we will get 70m for that, but we could get less or more. We don’t really control it. So if CA strategies are meh, then more ppl will deposit to the LPing side, thus making CA strats better. The only way this doesn’t happen is if there’s just not enough capital at any level. In that scenario (limited total capital available/interested in gearbox as a whole), you could end up with a majority of ppl depositing to the LP side and no one using CA’s. For that reason I (the writer of this proposal according to Ivan - mugglsect, that’s me :wink: ) agree with Ivan that we should use 10% of the funds to LM on the credit accounts side of things.

Yup yup, I understand we can’t ‘reduce to 7.5%’ APR as it’s all about actual deposits and protocol usage by participants. I probably should have phrased it differently, i.e.;

“Propose we reallocate 25% of LM incentives initially earmarked for Lenders to Borrowers (CA users).”

This would, assuming assumptions in Mugglesect’s calc’s held true in practice/deployment, reduce APY from 10% to 7.5% for Lenders in $GEAR.

Anyway; we’re all thinking along the same lines and seem to be in agreement, which is good.

I am supportive of a 10% reallocation of incentives from Lenders to Borrowers as proposed by @ivangbi (assuming you meant reallocation? Or did you mean an additional $GEAR allocation?) to align users with long term goals of protocol.

Either or I’m easy. Just think CA farmers need something considering the depressed returns in this bera market. I wouldn’t want to see everyone sitting on the Lender side and see the $GEAR go to waste from under utilisation of credit as the protocol wouldn’t be earning any revenue then.

I was suggesting a 1-2% on CA side extra. Doesn’t move the needle for the spending that much, but makes everyone aligned. Move the LP APY to 9% then maybe? Idk, it’s fairly same same. We are really just spitting in the wind and seeing how it goes. There is no data? Or maybe I am just not looking

1 Like

  1. I would adjust market expectations and plan TVL for $50M. The markets are so rough, that expecting more is a bit hard now. If it happens, the DAO can always adjust. But it would not make sense to print 20% APY if we target a 75M TVL and only get 40M TVL. 10% on $50M sounds good.

  2. Please help: if there is 1-1.5% of LM on the CA side, how does it work? I assume same as for retroactive rewards for CA suggested? Basically, depends on your time-horizon of borrowed capital (excluding your collateral). Targeting the 100 CA * $100K collateral * x5 leverage - 100 CA * $100K.

Shouldn’t do CA side still imho, I like the idea of trying to find PMF on that side tbh. If we don’t because conditions aren’t great, we can incentivise anytime in order to get it done.

Also, I think we are under calling leverage a bit, we are running a 3.5X on 4X max. at ~10X max, we should be higher esp with major borrowing being stables

1 Like

I also don’t see the logic for CA side LM buuuuuuut!

It’s the CA ninjas who said they would love to have just a tiny bit of GEARs as they are users who want to have exposure. It kinda makes sense after all idk. Like 1% literally, not as incentive, but as alignment.

Ok, anyway, seems like there is no objection for 2 weeks… what’s the way to move forward?

  1. I would suggest snapshot to have 7.5% and 10% respectively as choices for passive LM side.
  2. I would suggest saying “1% APY for CA side on borrowed capital”

So it can have transparent votes as:

  • Passive 7.5% and CA 1%
  • Passive 10% and CA 1%
  • NO, am against LM

And in terms of “how can people claim” is another question. I can’t code or do dev stuff, and devs are busy with the protocol. Maybe the easiest way is to just say “The DAO will make new merkles every 1 week”? And then people can come and claim, similar to how Discord process was done?

1% is very doable, but should be on collateral no?

On borrowing we effectively incentivise higher leverage and risk. On collateral used we incentivise bigger position sizes, can make it 2-3% on Collateral tbh

1 Like

On another note, I think we could also discuss the following:

  1. Reduce the target TVL especially for stables: currently, 5% of USDC marketcap has bene lost since Tornado incident, people are scared. Capital that is willing to jump around has majorly reduced. So I would say target 10M on each stable instead of 17M as it currently is.

Overall, required TVL to be at optimized level I would bring down from $70M to about $55M TVL. I would even say go slash down to $50M due to market conditions. Otherwise we might be 3x overpaying for the liquidity simple because demand is smaller in the current market, which is not necessary to do.

1% is very doable, but should be on collateral no?
On borrowing we effectively incentivise higher leverage and risk. On collateral used we incentivise bigger position sizes, can make it 2-3% on Collateral tbh.

Collateral doesn’t make money for the protocol on CA side, so you can literally just ape 3M without any risk for yourself at 1.01x leverage and make APY? No, not cool. There is no point incentivizing that.

We are not trying to run incentives on CA side - as per my message above, it’s just that Ninjas asked for small skin in the game too. SO it’s really not incentives but just alignment. Keep it small as to not ruin the PMF-testing model. Maybe just saying “1% APR on the amount you borrow” is best?

Update: would be just this much extra. For the sake of aligning with CA users, not large spending at all.

This is very doable honestly.

Think we should turn this into an incentive proposal

  1. 10% for LPs
  2. 1% for CAs
  3. 2 month boost to 300k$(2X) for white hats on critical captures
1 Like

fixed calculation bug on gSheet - seems correct numbers should be:

2 Likes

Played with numbers a bit more, to reduce the desired TVL (pool side only, not total) to $55M. I think this is good and achievable, yet not too retarded of an expectation. What do others think?

Are you comparing to any other specific protocols here, or is the $55m more of just a gut feel for what you think we can attract and how much is in defi rn?

https://snapshot.org/#/gearbox.eth/proposal/0x9a6d7ae302241e0132dfd0380a4e20a32eb16fe2c187e54962732eb202a73508

Changes: [GIP-30] LM Adjustment №1