Redefining Performance: Protocol Update February
“It’s a new dawn, a new day and the banks are failing good”… GM to all the GEARHeads. Turns out our 7th grade Gym teacher was right, a run does bring you a lot of learning. Specially when it’s a bank run. As the banks melted faster than BSC shitcoins, USDC took a hit and lost peg. But all things in crypto universe stabilised and mean reversed as soon as Ivan sold his bags. Between the meltdown and Ivan saving the market, there was pain, ambiguity, uncertainty and what we love the most, volatility. This update brings you comprehensive overview of the protocol’s performance, covering usage metrics, safety and health status, popular strategies, and Credit Account profitability.
These updates are published monthly to track the progress of the Gearbox protocol and keep the DAO and the wider DeFi community informed about all the neato stuff that’s happening at Gearbox.
Talking about stuff that happened at Gearbox, in our effort to become the base layer of leverage in DeFi, we introduced Gearbox V3. It is now under audit. You can read the complete update below.
You can read last month’s update here if you’re living in a time-dilated world where you’re 1 month behind (pro-tip: short USDC on March 9th)
Protocol Performance 🧮
In this update, we’ll start with a high-level overview of the protocol’s performance, including borrowing and lending activities, liquidations, and some key data on how CAs are being utilized. We’ll cover three areas: liquidity pool stats, credit account stats, and integration stats.
You can see all the data here in the Dune dashboard made by apeir99n or here in Google Data Studio made by Amantay or in the analytics section of the Gearbox website, made by Harsh
TVL & Lending Pools 🏊♂️
- TVL: The current TVL is approximately $85m, down $32m since the last update. You probably know why this happened, but in case you’ve been living under a rock or something, it’s attributable to concerns about USDC pegging + the collapse of SVB and other banks. TLDR: it got pretty spooky in DeFi this past couple of weeks, so TVL is down😞.
2. Lending pools: The total supply in lending pools has dropped to $75m. The pool allocations are as follows: ~$15.5m in DAI, ~$20m in USDC, ~$2.75m in FRAX (a newly added asset on the lending side), ~$30m in ETH, and ~$5.4m in WBTC. The introduction of FRAX into the lending pools gives new opportunities for both CA and passive lending users.
3. Credit Account Borrowing: Credit account borrowing has also declined a bit from last month with the following breakdown in assets: ~$12m borrowed in DAI, ~$17m borrowed in USDC, ~$900k borrowed in FRAX, ~$23.5m borrowed in ETH, and ~$300k borrowed in WBTC. The addition of FRAX to the borrowing pool has generated initial interest, and it will be interesting to see how this evolves in the coming months.
4. Utilization Ratios: The overall utilization ratio is at 78.7%, which is in the upper range of the healthy zone. The individual utilization rates are as follows: 76.2% for DAI, 34% for FRAX, 81.8% for USDC, 78.3% for WETH, and 5.9% for WBTC. The newly introduced FRAX asset has lower utilization for the moment, but this is to be expected for a new addition. We anticipate that utilization rates for FRAX will improve over time.
5. APYs: Organic APYs for the assets with significant usage in Gearbox are as follows: DAI at 1.06%, USDC at 1.26%, ETH at 1.81%, WBTC at 0.01%, and FRAX at 0.2%. If we include token incentives, APYs are extremely favorable for passive lenders: DAI at 6.58%, USDC at 7.05%, WETH at 6.84%, WBTC at 3.18%, and FRAX at 8.9%.
Credit Accounts🥷
Let’s dive into the updated credit account statistics, which provide insights into how Gearbox users are interacting with the protocol.
The below analysis is done at an overall level, for further analysis of CAs, a useful approach is to access the data studio page directly and extract insights at a unique CA level behavior to uncover potential alpha. You can find the data studio page here. Make sure you go to Version option > Select Leverage Ninja(V2) for the most relevant data
- Leverage Ninja Stats: As of now, there are 127 CAs open, with 111 unique borrowers (some borrowers have more than one CA). There are currently 6 more CAs open than we had in the last update.
- CA Size Stats: The average (mean) size of CAs is currently at 517k, with an average borrow size of 422k and an average collateral deposited at 90k. These figures have shifted downwards since last month, with CA sizes becoming smaller on average in both collateral and in borrow amounts.
- Leverage Stats: The average leverage in the system is at about 5.65x. The breakdown of leverage by asset is as follows: FRAX users are the most “risk-on” with 6.25x leverage (although this is a small sample size), followed by DAI borrowers at 6.03x, USDC borrowers at 5.86x, and WETH borrowers at 5.53x. WBTC borrowers are the most conservative, with leverage at 4.12x (also a small sample).
- Health Factor Stats: The weighted average health factor for CAs is 1.101, which is slightly healthier than last month and indicates a good overall system health. The health factors for individual assets are as follows: WBTC at 1.141, WETH at 1.105, USDC at 1.100, DAI at 1.084, and FRAX at 1.071. For context, users can be liquidated if this number falls below 1.
5. Liquidations: Since the last update, there has been one liquidation. We’ll take a look at what happened to the liquidated user further down in the article.
CA Strategies/Usage Breakdown
For all the DeFi enthusiasts out there, this section is for you. Here, we’ll discuss some of the most popular strategies that CA users are implementing and provide a quick overview of the most and least profitable CAs that we have identified.
Convex/Curve
Convex continues to be the most commonly used protocol in Gearbox CAs, with approximately $43.3m of assets in Curve via Convex, divided between five main strategies.
Note that many stablecoin strategies currently have elevated MAX ROIs because of the situation with USDC and the knock-on effects — there was heavy trading in almost all stablecoin pools, which temporarily inflated the ROIs on those pools, leading to some of these current max ROI numbers.
Here’s a breakdown of the assets in each strategy:
1. Convex sUSD-3Crv Pool: This remains the most popular strategy in Convex, with about $26.7m in assets. This Convex pool offers the highest max APY of any stablecoin strategy currently at 61.7% APY.
2. Convex stETHCrv Pool: With around $10.4m in assets, this is the second most popular strategy in Convex. The max ROI on this strategy is currently at 18.55% APY.
3. Convex FRAX-3Crv Pool: Holding approximately $2.6m in assets, the max ROI for this pool is 22.19%.
4. Convex gUSD-3Crv Pool: Despite some concerns around gUSD (although maybe these seem quaint at this point), this pool has about $2.6m in assets and offers a maximum ROI of 142.38%.
5. Convex LUSD-3Crv Pool: The LUSD-3crv Convex pool currently has $1m in assets. The current maximum ROI is 158.46%.
Lido Staked ETH
Lido Staked ETH remains the second most popular strategy in Gearbox, with $16m of assets in this strategy. The popularity of this strategy is likely due to several factors, not least the anticipation of staked ETH withdrawals in the upcoming Shanghai update and the healthy yields offered. The strategy currently has a max ROI of 23.5%.
The other two strategies involving Lido’s stETH are the Convex stETHCrv strategy with $10.4m in assets (discussed above) and the Yearn stETHCrv strategy. Both of these strategies involve dividing funds into stETH and WETH and placing them into Curve to earn additional yields via trading fees and CRV/LDO tokens. However, they also forgo half of the potential staking yields since only half the assets are staked in Lido.
The Convex stETH strat has a max ROI of 18.55%. The Yearn strat currently has a negligible amount of funds, so we won’t spend time discussing it.
Yearn
Approximately $2.4m of assets are in various Yearn strategies via Gearbox, with the majority split between the yvWETH vault and the yvDAI vault. The WETH vault has a max ROI of -26.76% with $1.3m in it, while the DAI vault has a healthier max ROI of 3.82%% with $1.1m in it. Despite the lower APYs compared to other strategies in Gearbox that use similar base assets, a not insignificant amount of users still utilize these strategies, presumably due to the perceived safety and security of Yearn. The remaining assets in Yearn strategies are negligible.
Are CA users bringing home the bacon?
Frequently, the most popular strategies aren’t the most profitable ones. So, let’s examine some of our top-performing CAs to discover what valuable insights and lessons we can take away.
Just a note here — https://twitter.com/gearbox_intern does a weekly thread that highlights the most profitable/interesting/noteworthy Credit Accounts users — definitely check some of these threads out if you’re interested in profitable strategies and/or novel ways to use your Gearbox CA. Here’s an example:
Additionally, it’s worth noting that when assessing the profitability of CAs, we adopt a cautious approach by considering potential worst-case scenarios. Our primary objective is to ensure that we can always liquidate an account for its estimated value, even under unfavorable market conditions. Consequently, our reported CA profitability numbers might be somewhat conservative (and losses on CAs appear more significant than they are). For further details, please refer to the relevant section in our documentation.
Top performing CAs
Of the top ten most profitable CAs to date, 7 are simply staked in Lido stETH with between 5x-8x leverage, and two belong to the legendary WBTC trader that continues to hold the #1 and #2 spots for most profitable CAs (more on that in a sec). That’s 9 of the top 10. The one remaining CA in the top 10 is a hybrid farmer-trader who initially deposited all their funds into the FRAX-3crv strategy, but in the chaos of last week did some opportunistic trading by buying some ETH and WBTC on March 12th and March 15th and has since made some strong trading profits on those transactions.
Revisiting the WBTC trader who holds the top two spots on the Gearbox leaderboards month after month… This user sold their WBTC! These two Gearbox CAs are now both entirely in the stables that they are denominated in. We’ll keep an eye peeled for whether or not this user uses these stables to farm or goes back to a long BTC position.
We think it’s pretty noteworthy that this CA user has sold out of both WBTC positions — this is not financial advice, but if you consider yourself a trader or active manager of any kind, this is probably worth noting down and thinking about.
A key takeaway here is the remarkable effects of LSDs (pun intended), specifically Lido’s stETH, when used with leverage. Apart from the 3 CAs that we’ve already covered, the top 10 most profitable CAs are entirely made up of credit accounts that have focused on leveraged farming with stETH. The leverage ratios for these accounts vary (5x to 8x), but they all essentially employ the same strategy. As mentioned last month, it appears that the leverage ratio is starting to hve an effect on returns — the more profitable steth focused CAs tend to have leverage ratios closer to 7x/8x (max unnanualized 10.5% ROI over 4 months) versus ~5x-6x which have produced ROIs closer to 7.5%-8.5%.
“A key takeaway here is the remarkable effects of LSDs“
Worst performing CAs
Same thing as last month — there was one liquidation between the update from last month and this update, but the liquidations didn’t actually happen in February nevertheless we’ll discuss it here rather than wait until next month. As a rule, the “worst performing CAs” must be those that got liquidated, so once again we salute our fallen brethren who got liquidated, and hope that we all can learn from his/her mistakes.
The lessons remain the same as ever. Do not use too much leverage.
This user opened their CA at 9.68x leverage (very high and close to the limit of 10x). They put it all into the sUSD-3crv convex strategy. This occurred on March 7th.
On March 9th, this user was liquidated, losing more than half of what they put in as collateral.
You might be thinking that this user got liquidated during the USDC depeg and the general chaos that occurred around those few days in the markets, particularly with regards to the prices of various stablecoins.
But actually, the deviations that caused this user to be liquidated were pretty small — the bulk of the depegging action didn’t occur until March 10th/11th. Which brings home the point — given the way Gearbox focuses heavily on safety/security, when you combine 9x+ leverage with the LT/health factor calculations, even normal day to day market fluctuations can cause you to be liquidated. We highly encourage people to avoid using 9x+ leverage unless they are extremely confident that they know what they are doing.
“The lessons remain the same as ever. Do not use too much leverage.”
We said this already last month, but we hope that in the coming month, there will be 0 liquidations, which would also allow us to shift our focus back to analyzing the “least profitable” credit accounts instead of discussing those accounts that have unfortunately been forcibly liquidated.
Protocol Health
So, Gearbox is performing reasonably well given the market conditions (even though it’s definitely a down month for TVL). But what about protocol health?
Bad Debt
Since the previous protocol update, there has been a single liquidation. Given the chaos in the market and the USDC depeg + knock-on effects of that, it’s especially worth noting that no bad debt was generated during this liquidation process or as a result of recent market conditions.
Audits
Audits for Gearbox v3.0 are getting underway. Basis the feedback, the devs will be making changes, if any and we can expect an end of Q2 launch date.
Risk Assessments & RiskDAO Reports
RiskDAO doesn’t stop — their reports cover the most interesting things that are happening in DeFi from a risk perspective, once a week, every week. This is a summary of what they discussed this month (yes, given the SVB/USDC drama, next month’s reports will be extremely good reading).
Feb 3rd: Aave and Compound launched v3s on Ethereum Mainnet, neither launch led to unusually high withdrawal volumes from their respective v2s. RAI received attention as its co-founder questioned/expressed potential regrets about the ETH only collateral model that RAI uses. LUSD announced that it is going multi-chain with Stargate Finance. Mango Markets sued Avi Eisenberg for $47m in damages. Re: Gearbox specifically, the RiskDAO team noted that sUSD-3crv pool positions had an outsized share within the collateral pool — not a red flag in itself but something to keep an eye out for.
Feb 10th: MakerDAO announced SparkLend, a protocol owned by Maker governance and built on Aave v3. BonqDAO was exploited for BEUR 100m due to a faulty oracle implementation. Sperax was exploited for $300k USDS but the attack was stopped, the hacker identified and funds will be returned. Everlend shut down, citing missing liquidity on Solana. Avi Eisenberg remained in jail after waiving his bail. Huobi issued $FUD tokens via DebtDAO in an attempt to create a market for FTX user liabilities. RiskDAO again noted that sUSD-3crv pool positions continued to be a high proportion of total collateral in Gearbox.
Feb 17th: Market recovery occurred alongside regulatory turbulence as the SEC clamped down on staking-as-a-service operated by US exchanges. Kraken settled for $30m and shut down its service, while Coinbase prepared to defend its ETH staking in court. Paxos faced issues with the NYDFS and the SEC. A whitehat hacker exploited Arbitrum lending protocol dForce for $3.65m but returned the funds. RiskDAO continued to note the high share of sUSD-3crv as a % of total assets in Gearbox.
Feb 24th: OlympusDAO launched a Lending AMO, and $OHM borrowing markets were enabled on Euler and Silo. Platypus’ USDP depegged due to a flashloan exploit, but a portion of the funds was recovered with BlockSec’s help. Avi Eisenberg sought permission to keep Mango exploit funds. Again, the high proportion of sUSD-3crv was mentioned as not a red flag, but something to keep an eye on.
You can read all RiskDAO reports in the governance forum here, and the Gearbox Risk dashboard can be found here
Protocol Enhancements
There are big changes afoot at Gearbox… We recently announced Gearbox v3 (name possibly subject to change) — which you can read about here.
Gearbox V3
Last month we called it Gearbox v2.1, but now we’re calling it v3, and possibly in the future we’ll call it something else. Nevertheless, the next big update to the Gearbox Protocol is on the way. We won’t belabor the point too much — lots of new things are being added, including (but not limited to)
- Risk segmented pools that allow for higher APYs on passive lending — otherwise known as Alpha Pools
- Total Asset Limits and Asset Quotas per User
- Related to per-user Quotas, Minimum Viable Tokenomics
- Quotas also allow potential deployments to L2s, and integration of smaller DeFi protocols
- Gearbots (mentioned in last update)
- New integrations/assets including Balancer/Aura and others
Read the full v3 announcement article to get all the juicy info!
GIPs and Protocol Changes
A link to the snapshot voting page for Gearbox where you can see all the latest proposals https://snapshot.org/#/gearbox.eth
A batch of new GIPs were voted on and passed by the DAO — here are some of the (non-administrative) interesting ones:
[GIP-43] OBRA DAO Governance Framework
GIP-43 saw the passing of a new governance framework that the DAO will adopt, allowing for a greater level of decentralization and more async form of collaboration, and is the first step towards possibly creating SubDAOs for all functions needed for Gearbox Protocol and the DAO to operate smoothly.
[GIP-45] Support/Maintenance of Risk Reports
GIP-45 saw the approval of the continuation of the maintenance and improvement of the Gearbox risk reports/data studios. Without the data provided by these reports, this protocol update likely wouldn’t exist, so we’re definitely pleased that this GIP passed.
[GIP-46] Add cbETH to Allowed List
GIP-46 is more or less self explanatory — adding cbETH + the cbETH/ETH curve pool allows for more diversification in the ETH staking strategies that Gearbox offers and also allows for both more choices and opportunities for users. The GIP contains explanations of why this might be worth doing and some explorations of potential risks.
[GIP-47] Add MIM, Curve MIM and Convex MIM pools
GIP-47 is also somewhat self explantory — as long the contributors on the Gearbox risk comittee are comfortable with the assets/strategies in question, broadening the number of integrations that Gearbox has is a no brainer. Here it’s proposed that MIM (the token), the curve MIM pool, and the Convex MIM pool are added to the allowedlist. There is quite an indepth explanation of the risks and the philosophy behind the LTs chosen if that kind of thing interests you.
FRAX added as a lending asset
February also saw FRAX added to Gearbox as a deposit asset on the passive lending side and as a borrowable asset on the Credit Account side. So far the lending pool has attracted 2.75m worth of assets, and there is about 950k borrowed. We anticipate that the utilization ratios for FRAX borrow will also go up over time as more users discover the possibility.
And that’s all folks…
Hopefully you found all of this intern’s babbling useful. See you again next month, Gearheads! Meanwhile, stay away from banks, scary things, those.
If you would like to earn passively, remember to Lend into our Lending Pools to earn GEAR. If you want to take leverage as Ninja just ping us on Discord or post on the forum. Otherwise you can always get involved with the DAO — 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:
