Gearbox V2: Evolution, Evaluation and What Next
Block 15861045, October 30th, 2022. After months of protocol design innovation, engineering grit and creation of a new safety module, the multisig executed the transfer that enabled the V2 of Gearbox Protocol. What ensued was beyond belief even for the most involved. In a matter of 3 days the TVL increased by $100M.
While there was the obvious jubilation within the DAO, a $100M TVL from a larger picture view was only a small part of the plan the protocol intended to execute. This article thus talks you through the evolution of Gearbox to V2 and evaluates how it actually went in the 3 months it has been live. Evaluating the security, the product, the DAO growth and more. This is our self-reflection piece with a recap on the good, the bad and the ugly performance aspects of V2. And of course, how we take this to the next level.
Evolution: Goals and Assumptions for V2
When V1 went live, the goal was largely to test if the protocol design will be safe and efficient enough to deliver what we wanted. V1 succeeded at delivering that and the devs simultaneously worked on the next Gear to complete the Gearbox Mechanism, V2. With V2, the goal was significantly different. First goal being finding initial PMF indicators(Product Market-Fit) in a completely new DeFi category, another was to establish a risk framework that ensured 0 bad debt, third goal was to expand the overall offerings we had. The third was realised with a cohort of leveraged stablecoin farming and LLSD. While, we’ve been fortunate enough as a DAO and as a protocol to see initial signs of potential PMF (product market-fit), and then a few more of those goals, in some areas, the goal now changes to growing it.
You can see how the protocol has been performing using our Protocol Updates. The latest one is linked below!
But before embarking into the future, let’s come back to the past a bit. Let’s look the assumptions that led to product decisions. The key assumptions were:
- Leverage farming as the go-to-market strategy: With the data from V1, albeit not expansive, we decided to go-to-market with Leverage Farming instead of Leveraged Trading. Leverage trading requires you to take on directional risk. Leverage Farming on Gearbox though can be made significantly less risky if the borrow and collateral asset are the same. As you take away the directional risk in leverage, it makes it easier to onboard sticky TVL, easier to not churn users due to liquidations and it also better displays the power of Composable Leverage.
Verdict: Strong Initial Usage! - Leverage Trading won’t have competitive advantage: With worthy competitors like GMX and Perp and our UI being curated to push leverage farming, the initial assumption was that leverage trading might be inferior on Gearbox. Over the last few months we have realised this might not be entirely true. With Gearbox’s low cost of borrow, Leverage Trading with size becomes more efficient on Gearbox. Our Top 2 performing CAs being Leverage Traders who aren’t farming is a confirmation of the same.
Verdict: We’ve underestimated ourselves! - Composability without manual multiple transactions: While Gearbox allows you to deploy your position through multiple protocols to achieve the highest efficiency, it shouldn’t require you to go through multiple protocols. And to deliver on the same, we cam up with Multicall. Over 80% of positions were opened this way, much more than devs anticipated.
Multicall as a way to simplify leverage user interactions: Worked! - Lending pools as a way for lower-risk lenders to engage: Not exactly how we anticipated it since post FTX everything was considered ultra risky, but the idea is settling in slowly. Credit market crunch didn’t affect DeFi negatively per se, but it surely slowed its growth.
Verdict: Partially worked but needs more work!
Remember, these were just assumptions and the go-to-market strategy. Technically speaking, the protocol can handle a whole range of assets and strategies, as well as leverage trading. After all, it’s generalized composable leverage. But there needed to be focus at first. You can’t be solving every problem at once, you need to target some strong use cases initially.
You can read about the prior assumptions in the evolution article we published earlier:
Evaluation: Self-Assessment
While the goals based on the assumptions seem to have satisfactory outcomes, there’s a lot more depth to the operations of Gearbox. And more than the assumptions what matters is looking at what has been accomplished so far. We’ll be doing this through the self-assessment framework Lido made(we’ll try to replicate this to be nicely visually open to everyone in the next versions).
There is much more to say, but you can always ask the details and argue with DAO members in Discord. We are very critical even when we accomplish something good: no stopping the gears ❤. And again, in terms of actual numbers:
1. $0 bad debt. Source: Risk DAO.
2. 0 emergency pauses. Source: verifiable on-chain.
3. 75%+ organic utilization (borrows). Source: Dune.
4. $100M+ TVL across USDC, DAI, and ETH. Source: Dune.
5. Numerous independent liquidators working. Source: Google Studio.
6. Multiple risk monitoring systems for the protocol. Source: Third Eye project.
7. 5 top-tier audits (ChainSecurity, Sigma Prime, Consensus Diligence, MixBytes). Source.
2.1 Community, Marketing, and DAO2DAO
The work never stops: more members, more posts, more articles, more engagement… There is hardly anything in particular that needs to be said here as all of the content across media can be seen on Twitter. Follow it, check out the news, and engage.
There are some things we are super proud of and happy about:
- Yearn Finance: deposited $2.5M from Yearn Treasury into Gearbox passive lending side
- Balancer Boosted Pools: depositing extra idle capital into Gearbox passive lending side
- Frax Finance AMO: $2M+ AMO being deposited into Gearbox passive lending side
- Brahma Finance: built vaults on top of Gearbox leverage Credit Accounts
- Mellow Finance: built vaults on top of Gearbox leverage Credit Accounts
- Incentivized GEAR/ETH pool: Curve, Yearn, and Convex voting for extra $CRV rewards
2.2 Security, Security, Security…
Gearbox devs are so paranoid that they spend 99% of their time on upgrading security. That’s right: not attending to their families, but writing bots, tests, monitoring systems… And that’s why Gearbox has such an expansive approach towards security. We bucket our Security into 3 key categories: Risk Mitigation, Safety Enhancers and Insurance
Risk Mitigation
- Risk DAO: As per our partnership with Risk DAO, they have helped us develop a 3 sided risk strategy that enables the protocol to function seamlessly and recommend bounds to minimise risks and a Risk Dashboard that constantly gauges Bad Debt risk. You can read the details in the article below to know more.
- Audits: There are 5 top-tier audits completed to date by firms including Consensys, Sigma Prime and ChainSecurity and more audits coming up. Details here.
Safety Enhancers
- Bug Bounty: There is an Immunefi bug bounty with multiple payouts having been done to date, with immediate responses upon receiving any notifications, devs always on standby. With V2 launch, there was also a 2X boost in place to ensure more whitehacks dedeicated time to Gearbox codebase.
- Contributor Initiatives: Apart from the Risk Committee, there are initiatives being funded by the DAO that further maintain Risk Reports in order to ensure protocol safety. These were added as a part of the OBRA model
- Third Eye Analytics: There are at least 2 monitoring systems that contributors are aware of, possibly more. These monitoring bots check for price feeds, Health Factor inaccuracies, and any other protocol deviations which could help detect malfunctions or exploits.
- Liquidators: At least 5 external liquidators running bots
Insurance:
- Say you do decide to sell the farm and put it all into passive LP like Michael Saylor did with Bitcoin — it may be wise to get insurance for it. The big brains at Nexus Mutual and Insurace have done a great job offering insurance covers for our users. As of writing this, it costs ~2.5% of the cover amount for a year’s worth of coverage for your LP position.
Nothing is ever fully safe, especially DeFi protocols. But devs and risk researchers do their best to make sure these risks are brought to the absolute minimum possible!
2.3 Product & Integrations.
Based on the assumptions explained in part 1.1, the product was optimized to leverage farming across the most safe protocols. Those were Curve, Convex, and Yearn; including assets like LUSD, GUSD, sUSD, stETH for LLSD narrative, and a few others. The LTs of assets were set up very safely yet allowed for x10 leverage in some cases. You can find the Allowed List of contracts that were integrated for V2 in the docs.
Leverage Ninjas had the choice of farming wherever and however within the Allowed List, no need for pre-made restrictive vaults! Freedom to users…
This is why “product” was marked in bold above: because the protocol is permissionless and is generalized. You could ape into those farms, or you could leverage trade BTC & ETH with leverage. Some users did both. That’s the power of generalized leverage!
Leverage anything, leverage everything.
And to make it possible, Multicall was designed specifically to let Credit Accounts users (Leverage Ninjas as we call them) enter and exit any complex positions easily. For example, entering a FRAX+USDC Convex pool from USDT was as easy as making 1 transaction literally. The combination of multicall + smart router [pathfinder] would made the process smooth.
This flash depositing and/or withdrawing technique made UX much easier and also saved users up a bunch on gas costs.
For example, see how it works in practice:
Check here for more details (ctrl+f for multicall). Find more stats and protocol events here:
Here are some things that could be improved:
- Leverage Trading: as mentioned above, the protocol is generalized, so you can leverage since Dec 2021 already… but the interface and the communications weren’t optimized for that. It’s easily done by improving some safe asset LTs, without the need to go for long-tail risk. This is something to fix in version Codename, see below.
- Interface UX: users have so far found it to be pretty tolerable, but improvements are and will be made. That especially concerns the analytics charts dApp.
- Collateral onboarding: has been slow, mainly because V2 only launched 3 months ago. Nevertheless, cbETH & MIM are appearing in Credit Accounts soon, as well as passive lending FRAX pool. What is needed is speeding up deployment / connecting the pieces, as those so far have been taking abnormally long post voting, which isn’t optimal.
- Interface improvements: have been and will be gradually done to improve user experience. While the passive lender side has not been an issue, there are a few dozen reports on the improvements for the current dApp and the analytics charts!
How and when these and other things will be improved… read in Part 2 below!
2.4 Governance
Transparency on governance speaks for itself:
- 47 Gearbox Improvement Proposals [GIPs] so far with detailed discussions. In some of them, there is a clear pushback from DAO members against developers even, where the former took the lead and “won”. In practice, it shows that governance is not just a meme, neither are quorums — which are hard to get. Check the forum.
- Monthly DAO reports, with every penny spent explained. See Notion.
- IP & licenses being truly in the hands of GEAR holders, see forum for ongoing process.
- Removed any previous core team control and transitioned to the OBRA governance framework, with 3-month review cycles and asynchronous initiatives. See Notion.
Overall, happy with it, but need to prioritize moving to on-chain governance by EOY or as soon as possible. That has been the sentiment among DeFi DAOs too. For now, adding a timelock to already-verifiable guards [executive multisigs] would be a good idea though.
What’s next?
Oh boy, so much! Keep an eye out on the content within the next couple of weeks, as we are ready to present the initial design of the *codename* version. Get ready! And as always, while you waiting for the details, put your idle capital to work on the passive lending side:
Thank you for reading this, the article was written by Ivangbi and has been posted on his behalf. If you would like to join — just get involved on Discord. Discuss, research, lead and share. Call contributors out on their bullshit and collaborate on making things better. Here is how you can follow developments:
- Website: https://gearbox.fi/
- dApp: https://app.gearbox.fi/
- User Docs: https://docs.gearbox.finance/
- Developer Docs: https://dev.gearbox.fi/
- Forum: https://gov.gearbox.fi/t/start-here-forum-rules/
- Blog: https://medium.com/@gearboxprotocol
- Github: https://github.com/Gearbox-protocol
- Twitter: https://twitter.com/GearboxProtocol
- Snapshot page: https://snapshot.org/#/gearbox.eth
- And, of course, the DAO working docs: