Yield Source / Vault Survey - Results and Recommendations

[Alx]

Thanks everyone for all of the feedback you provided on vault selection. I wanted to share out the results so far, and give G9’s recommendations on how to best move forward.

So, here’s the feedback we received, along with a little commentary from our team.

Top chains people wanted to launch on were clear: Arbitrum and Base were tied with 35% of the vote. Following that was Mainnet with 14% and then there were a number of other L1s/L2s that were write-ins.

Unsurprisingly the top tokens people want to deposit are USDC, ETH and POOL. Liquid staking tokens like rETH and stETH were also very popular. Behind those we have some other surprising choices. A number of people (27%) said they wanted to see stablecoin LP tokens and there are also quite a few people interested in depositing OP or ARB.

Note that there are some conflicting desires here. There is a clear preference for single asset vaults, stablecoins and ETH, but there are few options for that on Base. The biggest yield source there is Aerodrome, but that is primarily LP pools. Similarly, people want to deposit OP, ARB and POOL but there are really low yields for ARB and OP, and POOL can only be farmed as an LP on Velodrome right now.

As there are more options for stablecoin/ETH vaults on ARB, we think it would be best to launch on Arbitrum first, and then shortly thereafter on Base.

In terms of security, I think there have been some enlightening comments made on this post. MY take is we should be adding some information to the vault details page of cabana (perhaps into the vault lists?) highlighting key security considerations for each vault. Here are the answers people provided on security questions:

When it came to yield sources opinions were very diverse but a few protocols stood out including Aave, Yearn and Lido.

Recommended Yield Sources & Vaults

Based on all of this information, and the realities of what is available on various chains, these are the yield sources vaults we recommend we move forward with to start:

1: Aave

(with Yield Daddy wrapper)

Chains/Assets we gain access to:

OP: USDC - 4%, USDT - 4.7%, DAI - 4.55%, SUSD - 4.3%
BASE: USDbC - 18%
ARB: USDC - 4.6%

Estimated Cost:

Wrapper: $0 * Already exists and is audited
Integration Audit: $0 * Will be re-audited in upcoming C4 audit

Total: $0

2: Yearn

(It appears this will not require a wrapper)

Chains/Assets we gain access to:

OP: DAI - 12%, USDT - 11%, WETH - 5.67%, & a handful of liquid staking LP pools

• this list assumes we can wrap v2 vaults with v3 -technical analysis of this assumption is underway.

Estimated Costs:

Wrapper: $0 * Native 4626, no wrapper needed
Integration Audit: * $0 Included with upcoming C4 audit

Total: $0

3: Beefy

Assets we gain access to:

OP: WSTETH/ETH LP at 8%, ETH/RETH - 4.8%
ARB: GDAI (GAINS NETWORK DAI) - 25%, USDC - 22%, USDT/ARBUSDC.E - 15%, along with a variety of other liquid staking LP pairs

Estimated Costs:

Wrapper: $0 * Wrapped 4626, wrapper has not been audited (please correct me if this is wrong)

Integration Audit: $0 * Included with upcoming C4 audit

Total: $0

As these 3 yield sources are all likely to be included in the upcoming audit and don’t appear to require a new wrapper, we think these are the most impactful vaults we can launch with. This would allow us to have 14+ unique vaults.

By pursuing these integrations, along with partnerships with the organizations we believe we can make the most immediate impact for the lowest cost.

Other protocol we believe we should continue to explore integrations with:

Yield sources to integrate with bridged exchange rate:

Some assets, like Lido’s stETH and RocketPool’s RETH, have been bridged to L2s but cannot be used as a yield source. That’s because the underlying ‘yield’ is accruing on L1, and so it cannot be measured on L2 unless that value is bridged or provided by an oracle. This is additional infrastructure that ideally is run by the protocols themselves.

Lido

Requires a custom wrapper and exchange rate bridge to integrate this on an L2. Also, the partner/maintainer would need to have infrastructure and push updates on each chain. However, if a wrapper is created it is a compelling choice for ETH mainnet given the sheer scale of their TVL ($30.1 billion!).

Chains/Assets we gain access to:

Mainet: stETH - 3.3%

OP: stETH - 3.3%

Estimated Cost:

Wrapper: $?
Audit: $20,000 - 50,000

Total: $XX,XXX

RocketPool

Rocketpool requires a custom wrapper and exchange rate bridge to integrate on an L2. Also, the partner/maintainer would need to have infrastructure and push updates on each chain.

The G9 team has been in discussions with RocketPool for a few months and our understanding is there is a 4626 wrapper in development but we do not know the timeline for this.

Chains/Assets we gain access to:

Mainet: rETH - 2.94%

L2?: rETH - 2.94%

Estimated Cost:

Wrapper: $?
Audit: $20,000 - 50,000

Total: $XX,XXX

Other yield sources to consider in the future:

OETH / OUSD / Origin Protocol

Origin protocol is a yield aggregator which automatically rebalances across a number of yield sources including Morpho, Compound, Convex and Aave. It is 4626 compatible but is currently mainnet only. However, Origin will be launching on OP in the near future. Depending on their technical implementation it could be something we integrate on OP.

Assets we gain access to:

Mainnet: OETH - 4.55% , OUSD - 6.55%

(In the future)
OP: OETH - 4.55% , OUSD - 6.55%

Estimated Cost:

Wrapper: $?
Audit: $20,000 - $50,000
Total: $XX,XXX

Velodrome

This will require a wrapper, but it would get us access to great yield, especially LP tokens. But we need to prove demand for those types of tokens/partnerships and can do that at low cost with Yearn and Beefy to start.

Chain/Assets we gain access to:

OP: ALETH/WETH (alchemix eth/weth) - 11.95%, USDC/ALUSD - 15%, USDC/DOLA - 35%, WSTETH/WETH - 8.6%

Estimated cost:

Wrapper $?
Audit: $20,000 - 50,000

Total: $XX,XXX

Aerodrome

This will require a wrapper but will get us access to good LP token yields on Base. But we need to prove demand for those types of tokens/partnerships and can do that at low cost with Yearn and Beefy to start.

Chain/Assets we gain access to:

BASE: DOLA/USDC - 46.8%, USDC/USDBC - 9%, WETH/WSTETH - 7.5%

Estimated cost:

Wrapper ?
Audit: $20,000 - $50,000

Total: $XX,XXX

Maker / DAI Savings

The DSR is supposedly 4626 compliant, and a long time partner of PoolTogether. They deliver a solid 5% in a large market

Chain/Assets we gain access to:

Mainnet: DAI - 5%

Estimated cost:

Wrapper ?
Audit: $20,000 - $50,000

Total: $XX,XXX

Morpho

Morpho was originally an Aave/Compound fork but they now have created Morpho Blue a permissionless lending protocol on EVM.

Chain/Assets we gain access to:

Mainnet: USDC - 21%, WETH - 8.5%

Wrapper ?
Audit: $20,000 - $50,000

Total: $XX,XXX

[underthesea]

Thanks for sharing the results and GS’s take. A couple of immediate thoughts…

Excited for Yearn and Beefy! Awesome we can get those into the current audit. If we can have these at launch that would be really great. The partnerships are crucial, we should prioritize focussing where the integrating partner is promoting us, especially if we can get our gameification layer added to their app.

Mainnet is not something we can back-burner. This was a mistake of V4 that we should not repeat with V5. We are a yield lego and yield depth is still dominated by Mainnet. A survey does not capture the sentiment of liquidity. Most of the LSTs and LRT’s work out of the box with a vault that @trmid has already setup a factory for because they have rebasing versions on Mainnet. Otherwise the exchange rate is there to be accessed directly. One partnership on Mainnet can make huge prizes. We have seen it before and it can happen again overnight.

I agree bridging the exchange rate to L2s is not practical. It requires a Mainnet transaction for every value update which also adds risk. If the protocols themselves set this up in some way we can use it for L2 value accruing yield tokens. Or if they move native minting to L2s like EtherFi has said they will do. In the meantime let’s get those yields through aggregators like Beefy/Yearn.

An important one I do not see listed here is Compound. Not on Optimism but is on Base and Arbitrum and of course Mainnet.

We should be sure all integrations can farm incentives.

I do not think it’s relevant here to decide which vaults from Yearn/Beefy we want to focus on. Those decisions have several layers and are constantly evolving.

1 - Is there a partnership motivation
2 - Does the app want to support the vault
3 - Does the community want to promote the vault
4 - Do we incentivize (PT Inc or DAO or Other)

There may be some fragmentation here, but ideally there is alignment, I just think that’s a bit down the line from deciding what integrations to focus on. To some extent we should let the permissionless nature of it all play out.

As an aside I am for LP tokens. A lot of yield lives in token pairs, look at Beefy and Yearn. Holding USDC alone is in some ways shorting crypto. Holding ETH shorting USD. A USDC/ETH pair is an inbetween strategy. Investing has risks, best we can do is make sure those risks are clear.

[trmid]

Thanks for the survey breakdown @Alx. The list of yield sources as well as the specific assets that people want to deposit are helpful in the testing phase to ensure we catch as many possible issues before the code goes to audit.

I have been testing various integrations based off these results with the new vault code and although these tests cannot guarantee that every asset with a yield source will work, they are helpful in finding glaring issues with specific integrations. So far I have tested basic functionality across some vaults from Beefy, Yearn V3, the Yield Daddy Aave Wrapper, sDai, a generic rebasing Aave wrapper, a generic artificial-yield vault (potentially useful for a POOL vault), and the Yield Daddy Lido Wrapper. Here are the current findings:

Lido

• Network: Ethereum mainnet
• Wrapper: Yield Daddy stETH Wrapper

The integration of the stETH wrapped yield vault into a prize vault appears to work well so far; however, there is an issue with stETH itself that may lead to some unsatisfactory results. It seems that stETH can frequently incur 1-2 wei rounding errors on token transfer. This means that if a depositor withdraws 1 stETH, they may actually receive 0.999999… stETH to their balance. stETH is a rebasing token, so this error may be made up the next time rewards accrue on the stETH contract, but it can still lead to unexpected results if a contract is interacting with the prize vault and expects to receive the exact stETH balance that they withdraw.

My initial recommendation for this would be that we focus our development attention on an alternate LST that does not have this issue. If we are able to obtain a wstETH yield vault on L2s, we could also explore having a wstETH vault on Ethereum as well to avoid these rounding errors.

Yearn V3

• Network: Ethereum mainnet
• Wrapper: Official yearn v3 vaults

While the yearn vault strategies are functioning under normal operating conditions, there have been no discovered issues yet. However, the Yearn V3 vaults have a unique approach to realizing losses in the vaults where withdrawals are normal until a withdraw tries to pull from a strategy that has incurred some loss. When that withdrawal is pushed through, the loss is passed on to the withdrawer, instead of socialized across all depositors. This can lead to some odd behaviors when paired with the prize vault since one lossy withdrawal could cause everyone in the prize vault to incur the loss. There is also a chance that withdrawals on the prize vault may be completely halted when an underlying loss occurs since the yearn v3 vaults appear to break the ERC4626 spec in this state. More testing needs to be done to evaluate the behavior in this state.

The Others

No other issues have been found for the remaining yield source integrations.

[Lonser]

@Alx

3: Beefy

Assets we gain access to:

OP: WSTETH/ETH LP at 8%, ETH/RETH - 4.8%
ARB: GDAI (GAINS NETWORK DAI) - 25%, USDC - 22%, USDT/ARBUSDC.E - 15%, along with a variety of other liquid staking LP pairs

Estimated Costs:

Wrapper: $0 * Wrapped 4626, wrapper has not been audited (please correct me if this is wrong)

The Beefy Wrapper is kind of audited already I think:
https://github.com/beefyfinance/beefy-audits/blob/master/2023-08-03-Beefy-Zellic-4626-Wrapper-Audit.pdf
If it’s good enough for us or we maybe wanna do a second one before rly using it is up to us though, Beefy itself haven’t rly used their ERC4626 Wrapper much itself asfaik.

Apart from that, agree with the importance of a ETH launch, lots of yield sources and we could deposit the treasury there too then.

Also adding to that and Morpho Blue as a potential mainnet yield source they are also using ERC4626 asfaik
https://docs.morpho.org/concepts/morpho-optimizers/morpho-ecosystem/user-tutorials/use-vaults-ma-mctoken

@Alx

Morpho

Morpho was originally an Aave/Compound fork but they now have created Morpho Blue a permissionless lending protocol on EVM.

Chain/Assets we gain access to:

Mainnet: USDC - 21%, WETH - 8.5%

Wrapper ?
Audit: $20,000 - $50,000

Total: $XX,XXX

Instead of building a Wrapper for Velodrome directly I think it’s easier to just use Beefy, they are already doing the hard work of swapping all the yield the LP earns and converting it into more Beefy shares, then the prize vault can just sell those Beefy shares. Long term a native Velodrome Wrapper would be nice though. One thing we should make sure of is talking to the Beefy guys and getting their help setting up a POOL/WETH LP Vault, it’s pretty easy using their general Velodrome strategy asfaik and the LP currently gets OP boosted by DeFi Collective (I am providing LP there myself, would be nice to contribute to the prize pool that way )

Also regarding the topic of a POOL Prize Vault. Thx to @trmid we got a wrapper to convert ERC20 into no yield ERC4626 Vaults. That way it is very easy to build a no yield POOL Prize Vault. I will propose (in a separate Gov Post) to incentivize that Vault with some treasury funds so we can make it eligible to win prizes! Even like $10k in WETH would be enough at POOL prices like 0.50$ for 200k POOL to earn 10% yield for one year! It’s time that POOL holders can win prizes and that way we can make it easily possible! In the future if technically possible we could even maybe think about redirecting the “buyback and burn POOL” WETH funds how it currently is planned into that POOL Prize Vault instead! That way instead of buying back and burning POOL we give all POOL holders a chance to win prizes!

[Tjark]

Thanks for sharing these, Alex!

@Alx

Top chains people wanted to launch on were clear: Arbitrum and Base were tied with 35% of the vote.

While Base seemingly has top-tier developer support with a very vocal community with lots of traction on socials, Arbitrum dominates in terms of yield sources and Defi TVL.

Adding images of the other results and my personal interpretations:

By no surprise, most participants can agree on (w)ETH and USDC. That’s hard money on Ethereum. Interestingly, POOL comes in third. We can also see a trend for liquid (re-)staking tokens, followed by alternative stablecoins.

My theory around the POOL result would be, that people are looking for token utility to have a reason to hold, instead of selling their stack. With $WETH as the new prize token, the demand might decrease. The $POOL token narrative needs to be redefined and we need more education about the token’s utility.

Yield sources in ranked order including the chains they are live on:

1. Aave V3: Ethereum, Optimism, Arbitrum, Base, Scroll, +7 others
2. Yearn V3: Ethereum, Polygon PoS, Arbitrum launching now
3. Lido: Ethereum, Polygon PoS (wstETH yield on Arbitrum via Idle, Notional, Pendle, Beefy; on Optimsim via CIAN or Pendle)
4. Beefy (listed twice): Ethereum, Optimism, Arbitrum, Base, zkSync, +15 others
   5.1. Maker: Ethereum (or Spark on Gnosis & Ethereum)
   5.2. Curve: Ethereum, Optimism, Arbitrum, Base, Polygon PoS, +8 others

Additional yield sources that were suggested: Synthetix, Exactly, Summer, Kamino, Marginfi, Meteora, JLP, and Origin.

Remarks by the community:

• “This goes to the Security Standards questions, it’s not obvious what the difference is between Vault Audits and Vault Wrapper Audits, perhaps a confounding factor in survey results.”

• “Intersection of safe, high yield, additional rewards, airdrop potential”

In terms of security standards, the majority of participants agree that vaults should be audited and the audits should be performed by well-known and trusted auditors.

Half of the participants don’t see contract upgradability and ownership as crucial.

Since most of the survey participants seem to be core community members (re the high amount of $POOL replies to the token question) an interpretation could be that they choose to trust the protocol and its developers. Interesting results nevertheless!

[Tjark]

@Alx

MY take is we should be adding some information to the vault details page of cabana (perhaps into the vault lists?) highlighting key security considerations for each vault

Agree. More info = more good
I could imagine something like a security checklist once you pull up additional info for a vault:
Vault audited
Non-upgradable
Ownership removed

[Alx]

@underthesea

Mainnet is not something we can back-burner.

Agree with you on this, the depth of TVL/Liquidity alone on mainnet could outweigh the downside of higher transaction costs. This could be a good reason to prioritize Mainnet after the Arbitrum launch.

@underthesea

An important one I do not see listed here is Compound.

I think this could make sense as well, it seems like @trmid has been detailing some of the technical challenges of this on the PT Discord, but it does seem like this could be worth getting the compound V3 wrapper audited.

[Alx]

@Tjark

Agree. More info = more good
I could imagine something like a security checklist once you pull up additional info for a vault:
Vault audited
Non-upgradable
Ownership removed

I really like this! Could go right onto the vault details page. I guess we would just need some way to get that data, ideally programmatically. But at worst, it could be manually added to vault list or added as a field in Cabana Factory maybe?