V2 Liquidity Mining Brainstorming

[mike]

V2 Liquidity Mining

Preface

This post is meant to be a kickoff point for ideas around liquidity mining in v2. All of this is still up for discussion and is purposely not polished. Even the Balancer team has not read this as our goal is to start having more discussions in public to also bring in community members in the process. I encourage the community to take an active role in helping shape the new liquidity mining program.

Liquidity Mining Purpose

Before diving into the details about a revamp of the liquidity mining program it is important to first look at how and why liquidity mining exists. Liquidity mining is first and foremost a token distribution mechanism for BAL. This has been the idea since the beginning and will continue for years to slowly distribute BAL to users of the protocol. The goal is to have the widest distribution possible across users and time in order to achieve a decentralized ownership and therefore governance of the protocol. Liquidity mining did not come as an afterthought where BAL token supply was inflated at the expense of existing holders in order to boost stats. And because this was the plan all along, Balancer is in a unique position in that it has an extremely powerful incentive war-chest that can be used to capture liquidity, users, and ultimately value to BAL holders.

Goals

Below are some general thoughts about the evolution of v1 liquidity mining, trends in the defi space, and goals we are looking to capture as part of the new program.

The original goal with v1 liquidity mining was to be as open and welcoming as possible to all types of tokens and liquidity. By having a more open set of tokens the process was obviously gameable and evolved to include different factors over time to align BAL distributions with the most useful liquidity. This included fee factor, ratio factor, wrap factor, etc. More details can be found here: Liquidity Mining - Balancer

All of those factors were necessary and after a couple months of tweaks did a pretty reasonable job of preventing gaming and encouraging healthy liquidity. Although we found that each additional factor increased user misunderstandings and complexity of the scripts. Ex: scripts can only be run once a week because aggregate data is necessary in the calculation and therefore its difficult for users to estimate or keep track of BAL earnings until they claim. Not only that, but understanding the calculations can even be confusing for those who built it and it became a difficult process to explain to new users how everything mapped from liquidity → BAL

GOAL #1: Simple. The liquidity mining process should be dead simple to understand and at any point an user should be able to see an accumulating tally of their BAL for liquidity.

Along the same lines above about the calculation scripts, this also meant everything has to be run offchain. While this was an okay process in the beginning, the idea is to move onchain so the process can be more automated and continuous instead of weekly intervals.

GOAL #2: Automated. Liquidity mining should be calculable and distributed onchain allowing users to withdraw accumulated BAL at any point in time.

Taking a step back and looking at the broader dex space, one thing that has become clear is how fast narratives and interest shifts in the Defi ecosystem. In any given week new sets of tokens make up for massive % of overall trading volume. There was the uniswap & yfi fork phase, the algo stablecoin phase, and many more.

The idea with liquidity mining is to incentive liquidity that is healthiest for the protocol. And because accrued fees is how the protocol may capture value, real trading volume is the metric to try and align BAL incentives to. By being open in V1 to all tokens, incentives get spread incredibly thin when trading volume follows a long tail distribution. This means Balancer loses out on a lot of trading volume when it does not have competitive liquidity for the hot tokens.

GOAL #3: Agile. Liquidity mining should be agile to target high volume tokens

Lastly, there have been dozens of teams building and integrating Balancer protocol in interesting and novel ways. These projects and respective tokens if any should feel welcomed and aligned with Balancer as a partner.

GOAL #4: Strategic. LBPs and projects building on top of Balancer should be more closely tied into the BAL mining process

We believe teams that choose Balancer’s LBP to launch their project tokens should have a head start in terms of competing to get BAL incentives for their pool. This should though only happen if they keep the liquidity of their LBP on Balancer after the LBP finishes the weight switching (which is the period where the project tokens are being actively sold). A good example of a project that did that is Perp.fi. They have been great partners to Balancer and generated a lot of trading volume on Balancer.

Please join #governance in Discord to share ideas on how to best accomplish these goals and help shape incentives in v2.

[Coopahtroopa]

Great thoughts here @mike - Some thoughts based on the above:

@mike

The idea with liquidity mining is to incentive liquidity that is healthiest for the protocol.

To me, the healthiest liquidity is that which is actively being used by traders and projects on a daily basis. A pool which has $100M in liquidity but is not ever touched either due to non-desirable assets or high swap fees is not nearly as valuable as $1M in liquidity being traded against heavily (like YFI in the first week of launch).

I would push to see liquidity mining rewards distributed to the liquidity generating the most protocol fees for the network.

@mike

We believe teams that choose Balancer’s LBP to launch their project tokens should have a head start in terms of competing to get BAL incentives for their pool.

Love this thesis and think there is design space here similar to what Sushiswap is doing with their Onsen & MISO program. LBP partners could either receive grants and/or benefit from an ‘LBP factor’ in which their liquidity mining rewards are boosted for a predetermined period of time.

Overall, this is a fascinating topic and one I’m very excited to iterate on.

Taking the rest of these thoughts to Discord. Excited for whats next!

[DavisRamsey]

My thoughts are no factors, incentivize 80/20, index style pools, and curve style pools. Trades within balancer will be far more efficient than trades with other AMM’s, so I think it could make sense to incentivize the same tokens in multiple pools to encourage as much internal trading as possible. For example, an 80/20 AAVE and an index pool with AAVE.

We can have certain pools given a multiplier similar to SUSHI to reward things like LBP’s.

Some kind of gatekeeping around what gets voted on to avoid a million votes. A small committee, formal process involving discussion, soft polls, etc makes sense to me.

[delitzer]

Love this re-think of the liquidity mining program.

IMO, one of the strengths of the v1 program has been the inclusiveness. This has been hindered by the lack of automation and slightly clunky process for navigating whitelist additions and updates. I would prioritize simplicity and transparency in the process of qualifying for inclusion in the v2 LM program(s). The more this can be streamlined, the more high-quality early-stage teams will be confident in their ability to get their pools included in the program, which would help push Balancer up their list of potential primary pools to use at launch.

[tongnk]

I 100% agree with the inclusiveness piece. I do think that how we’ve had the caps has been a decent way to go around it - though we can probably improve it.

I actually think that from a liquidity mining perspective it makes more sense to be trying to target the pre-mainnet early stage projects than super well established ones. There’s 2 main reasons I believe this:

1. Stickiness is a lot easier to build for users building on top of Balancer if we get them earlier. If they join early and love the product and we have a product suite for them to use throughout their growth lifecycle, it’s a no brainer for them to stick
2. Larger projects usually have established solutions already. I think it makes more sense to use targeted BD + grants + proposals here to try and attract them over instead of liquidity mining. Liquidity mining wouldn’t bring an Aave partnership - relationship building + grants + proposals do.

Obviously you can’t just give all the rewards to the early stage projects but I do think this shouldn’t be neglected

[Fernando]

I agree with both @delitzer and @tongnk. Adding a couple more thoughts:

1. The amount of volume an incentivized pool is getting should have a direct impact on how long the LM incentives will last. The more volume in USD generated the longer it stays incentivized. We could think of a gradual decay of incentives to avoid abrupt incentive changes that may have bad consequences.
2. While I agree that small and upcoming token pools are the key to our success and ethos of inclusiveness, there are some established projects (I’m thinking of AAVE’s SM smart pool but there are others) that definitely deserve high LM incentives. AAVE is awarding 550 AAVE/day to that pool, Balancer should at least get close to matching it if not going beyond it.

I’d love to see more ideas from more community members, we have to get this new LM plan drafted soon if we want to implement it by the launch of V2. Of course the transition between V1 LM to V2 LM will be smooth and gradual, but everything needs to be very well discussed and agreed upon in advance.

[Coopahtroopa]

Great thoughts above! Highlighting my favorites:

@delitzer

I would prioritize simplicity and transparency

@tongnk

target the pre-mainnet early stage projects

@Fernando

AAVE is awarding 550 AAVE/day to that pool, Balancer should at least get close to matching it if not going beyond it.

Taking all the above into account, I see room for base allocations to established pools (like aBPT for exmaple) while having a degree of flexibility around up and coming pools. There should be a core set of pools that always receive some allocation, with a floating allocation that can be nimble according to where demand is.

Excited to keep this conversation going

[rogenth]

I would also be in favor of an on-chain method to calculate rewards, not only for simplicity, for also for saving gas costs. In general, people farming with smaller liquidity in AMMs need to wait a long time for them to be profitable to claim rewards. The current method used in Balancer makes it unfortunately every week more expensive. For instance, the guy that gets 10 BAL after 1 week pays way less fees than the guy that has to wait 10 weeks to withdraw the same 10 BAL. In comparison, in other AMM, the gas cost for claiming rewards was almost 65% less for the same period of time for similar value rewards/week. Is there any possibility to keep the current simple system (not staking), but also optimizing gas costs?

[DavisRamsey]

I think a better metric might be fee earnings instead of volume, though if every pool has the same fee then volume would work. Just have to keep in mind almost any pool can generate very high volume with a very low fee.

[followthechain]

Loving the idea of a revamped liquidity mining that allows for more simplicity, predictability and automation.

Some food for thought…

We could have different tiers of pools that get each a fixed amount of BAL. According to Mike’s idea, people could claim their BAL on an ongoing basis, from a BPT staking contract.

In a 3-tier system, governance could vote (via Snapshot) on a selection of tier 1 pools, and general guidelines for what’s expected for tiers 2 and 3. Then a smaller committee (e.g. Ballers) could be granted the authority to fine-tune the pools on tiers 2 and 3 on a weekly basis.

Tier 1:

• ETH/USDC or ETH/DAI → the main ETH/USD pool
• ETH/WBTC → the main ETH/BTC pool
• DAI/USDC/USDT → the main USD stable pool
• 80/20 BAL/ETH → the main BAL pool

Tier 2: pools that are likely capable of bringing good volume to Balancer. Would be somewhat sticky, changing not very often.

Tier 3: experimentation, for instance incentivizing potentially hot LBPs and capitalizing on emerging DeFi trends. Would rotate faster.

The end state of liquidity mining on V2 (after mining has transitioned entirely from V1 to V2) could look something like this:

Reasoning for those numbers:

• deep liquidity on tier 1 pools is paramount because it unlocks a great order routing experience for a wide variety of trades within Balancer
• it’s safer to concentrate distribution on tier 1 because it would be controlled directly by Balancer Governance

Before V2 mining starts, governance would ratify (in case Ballers are the committee controlling tiers 2 and 3):

• the number of pools and amount of BAL allocated per pool at each tier
• the initial set of pools on tier 1
• the general guidelines for how Ballers should choose pools for tiers 2 and 3
• what type of values/attitudes/knowledge would make one eligible to become a Baller, and what’s expected of them
• the initial batch of Ballers

This is just a rough draft of what liquidity mining on V2 could look like. In case this looks like a good direction, let’s think through all the details and fine-tune the parameters.

[Fernando]

Thanks for that @followthechain.

I love the idea of having a committee (or whatever we call it) simplifying the somewhat burdensome process of fine tuning the tier 2 and 3 pools every week. This would be an unbearable overhead for governance if it had to vote on it weekly. This is similar to what Synthetix did recently.

It might be interesting to add some special invited members to help the Ballers decide on the tokens weekly. Even though these invited members don’t have voting power they could bring precious insights to the discussion. Two such persons I see very fit for this are our friends @delitzer and @Coopahtroopa. Would you be interested? No requirement to be participating every week, of course, all up to you.

Another addition I would make is that we should have a process to promote new Ballers and demote existing Ballers. I would suggest this to be a governance vote as it should not happen too often.

@followthechain what do you think of having a signaling method using another space on snapshot (e.g. “Balancer Liquidity Mining” where people would suggest pools for the Ballers committee to consider for liquidity mining? I would add a minimum amount of BAL that proposers need to have to be able to add a pool suggestion. Then the rest of the community could add their support to the proposal. The final decision on the pools for LM though should IMO still be at the committee’s discretion.

[markus]

@followthechain

The end state of liquidity mining on V2 (after mining has transitioned entirely from V1 to V2) could look something like this:

I like these numbers as an end state, but I think we should be more aggressive in order to bring a lot of liquidity for tier 1 pools from the start:

• As liquidity mining transitions from V1 to V2, only tier 1 pools participate (4-6 weeks?)
• Once transition is over, another few weeks (2-4?) where all 145k BAL goes to tier 1 pools
• After that, tier 1 rewards immediately drop to 100k, 25k/pool and we introduce tiers 2 and 3

As we phase out the V1 liquidity mining program, I assume most LPs will reevaluate their positions and more strongly consider moving to other AMMs. The ETH/BTC pool is currently mining ~30k/week, and I’m afraid reducing that could drive LPs away. If we keep the mining APY high (potentially higher than today’s) for a few weeks, we can be more confident that the incentives exist for them to migrate and get the ball rolling in the liquidity-drives-volume-drives-liquidity game.

In addition to incentivizing V1 liquidity to migrate to V2 instead of moving to the competition, the 2-4 weeks of high allocation for the tier 1 pools would create some FOMO among farmers, further increasing liquidity.

And if we want to make some serious noise, we could even consider increasing the 145k issuance for those 2-4 weeks!

@followthechain

Before V2 mining starts, governance would ratify (in case Ballers are the committee controlling tiers 2 and 3):

• the number of pools and amount of BAL allocated per pool at each tier
• the initial set of pools on tier 1
• the general guidelines for how Ballers should choose pools for tiers 2 and 3
• what type of values/attitudes/knowledge would make one eligible to become a Baller, and what’s expected of them
• the initial batch of Ballers

Minor plus: the strategy outlined above has the added benefit of buying us a few weeks before we need to discuss and decide all of these points.

[TelcoinD]

Hi there.

I am still relatively new to Balancer and honestly just started reading more about it the last few days. So i apologize for asking the below questions.

Based on the 3 tiered system you mentioned, as it stands right now, do we know how many pools will land on each tier? Also, not sure if i am reading it correctly but tier 1 is the only tier that will have governance votes and tier 2 and 3 will be taken care of by Ballers?

Reason I am asking is that if most of the pools (lets say 80% - just a random #) will land on tier 2 and tier 3 then the governance capability of people have been reduced to that 20% and the ballers decision is effectively affecting 80%?

Thanks.

[markus]

@TelcoinD

Based on the 3 tiered system you mentioned, as it stands right now, do we know how many pools will land on each tier?

followthechain’s proposal had 4-4-7 pools in tiers 1-2-3, but that’s just a suggestion and it’s one of the things we still have to decide.

@TelcoinD

governance capability of people have been reduced to that 20% and the ballers decision is effectively affecting 80%?

Hm, not quite. Keep in mind Ballers would still be appointed by governance. You can think of it as governors delegating a limited set of their powers to Ballers in order to reduce overhead.

[TelcoinD]

Thanks.

Any chance that some of the activities for tiers 2 and 3 be automated based on set rules? Then Ballers can then jump in when theres a discrepancy/gap that automation is unable to discern?

[markus]

@TelcoinD

Any chance that some of the activities for tiers 2 and 3 be automated based on set rules?

I think one of the main learnings of the liquidity mining program was that rules quickly become too complex for people to understand because we are too often having to tweak them to account for some newly discovered manipulation risk.

In my opinion, a set of broad guidelines should be enough to get the committee/Ballers started; continuous community feedback will allow us to constantly improve their methods and help everyone reach consensus; and if it comes down to it governors demote misaligned committee members.

[Fernando]

Love the suggestion @markus !

I think we also need a set of directives for the selection of pools that will be made by Ballers. They still have full discretion but a set of desired attributes for candidate pools would give a lot of transparency and clarity for teams that are looking to apply for being considered for LM.

Anyone volunteers?

[Shawman]

I really like this foundation to begin discussions around. I agree with the Tier 1 pairs, but think that it would be beneficial to expand the Tier 2 pairs from 4 to maybe something like 20 to include blue chip crypto assets (high market cap) that would probably migrate off the platform if we did not continue to bootstrap. Scanning digital assets over 400 million in market cap, the following assets come to mind: Chainlink (Link), Uniswap (Uni), Aave, Synthetix (SNX), The Graph (GRT), Maker (MKR), Compound (COMP), Sushi, Yearn (YFI), UMA, REN, Loopring (LRC), Basic Attention Token (BAT), Bancor (BNT), Curve (CRV), Wrapped Nexus Mutual (WNXM), 1Inch, Kyber.

Tier 3 could be targeted strategically on mid to low cap digital assets.

I also like the idea of the #bal per pool to be adjusted by a balancer liquidity mining working group similar to what Maker does for monthly rate setting.

[rabmarut]

I love the general idea here, but I’m not convinced about concentrating so heavily on just four pools. The suggestion of 25,000 BAL per tier 1 pool leaves only 31% of the distribution for all remaining pools in tiers 2 and 3. I know everyone is worried about losing deep liquidity to other protocols, but Balancer’s flexibility can speak for itself here. V2 will have low gas costs and the most generalized framework for any type of liquidity pool one can imagine. Liquidity in these four main pools benefits not only from concentrated liquidity mining but also from multihop routing through all other pools.

Perhaps @markus’s approach to starting with only tier 1 and scaling down over time - I really like this - will reveal that liquidity is not so easily lost and that there is actually room for more pools in the program. I think there are some attractive alternatives to the proposed 4x25, 4x6, 7x3 (15 pools) structure. One is 4x20, 5x7, 10x3 (19 pools), and a more aggressive one is 4x15, 7x7, 12x3 (23 pools).

Let’s take the last one as an example. In my mind, you’d have the four main pools still earning 15,000 each for over 40% of the total. Then 7 tier 2 pools could be devoted to popular “blue chips,” especially close Balancer partners, like Aave’s smart pool. And finally you still have 12 tier 3 pools to play with: these are the ones that change most frequently and target trends and/or new token launches (preferably LBPs).

[Shawman]

I tend to agree that we will have to analyze very closely the tradeoffs between heavily concentrating bal rewards on four or five pools (Tier 1) and the benefit of incentivizing a larger basket of pools with more Bal in Tiers 2 and 3. I was curious about what Uniswap’s data reveals, and I’m not seeing clear evidence on a volume to liquidity ratio basis that smaller liquidity pools contribute less value to the protocol than the higher liquidly pools. I think once the Tier 1 pools get to a certain liquidity, the law of diminishing returns kicks in. Meanwhile, blue chip crypto assets tend to need a minimum of 5 - 10 million of liquidity or the slippage is really bad. I’m liking the 23 pool idea @rabmarut !