V2 Protocol Update: November 2022. Are Leverage Ninjas winning?!
It’s been about a month since Gearbox v2 went live in Leverage Ninja mode and with the upsurge in TVL and usage that was brought along, we believe it’s time to kick off a deeper look into the protocol. And for the same, let us bring to you the first edition of the Protocol Update.
The protocol update covers all things related to the protocol: Protocol performance, health. enhancements. And then some pure alpha on what the best performers were doing and where the most money was made.
This will be different from the monthly DAO update which largely focuses on DAO operations, finances & activities. You can read this month’s DAO update below:
This month’s protocol update is in particular an interesting one with so many significant events happening in the industry over November leading to changes in APYs, TVLs and user sentiment as well as the introduction of V2 leading to a lot of curious questions from the new users regarding APY fluctuations. A great way to resolve a lot of those queries is by going through the leverage bible, for anything else read below and dig in!
Protocol Performance 💹
First things first, let’s look at the bigger picture. How the various aspects of the protocol have performed. Want to know how many ninjas are there? How much is being borrowed and how much is available for lending? How many people have been liquidated 💀? Where are these people making their money? To understand all these, let’s look at it from 3 sides:
- TVL liquidity pools stats
- Credit Accounts stats
- Integration stats
You can find all the data here in dune made by apeir99n or here in google data studio made by amantay. And some more in main charts by harsh.
TVL & Liquidity Pools 🏊♂️
- TVL: Rose from 12m$ to 115m$ since the launch, aided by the GEAR LM program to attract LP, you can read more about that here. These APYs are reflected in the dApp are claimable right now in there as well for the first month of earning. Connect your wallet and see your $GEAR!
- Lending pools: Cumulative value has gone up to $101m, deposited by 4173 unique LPers (in-and-out). USDC pool has the highest size at $40m, WETH at $31m, DAI at $19m, WBTC at $10m and wstETH at $1.5m.
3. Borrowings: USDC has the most usage for borrows at the moment — there is $24m in USDC borrowed on Gearbox, compared to $16m in DAI and $18.5m in WETH. (WETH was probably the most borrowed at the time of v2 launch, when ETH was at ~$1550 … thanks SBF!) WETH is the most popular with 55 v2 Credit Accounts borrowing it. There is only $100k of WBTC being borrowed in Gearbox by 1 lonely Ninja’s credit account at the moment. This is probably down to the lack of WBTC strategies with decent APYs
4. Utilization and APY: With the borrowings being sizeable, the utilisations have been fairly good for the larger pools. A 70–85% utilisation on the stable pools have meant an average borrow rate of around ~2% for USDC and DAI. WETH has been able to maintain a 70–85% utilisation rate as well leading to a ±3% level. WstETH and WBTC have had significantly lower utilisations and in turn lower APYs. Keep in mind that 85% is the U which makes rates kaboom steep up, so 85% is quite literally the max.
With utilisation based organic APYs and added GEAR rewards, passively earning on Gearbox can be a highly profitable strategy. Not the Avi style, hope not.
5. LM program: With utilizations changing as drastically as they did post V2 launch, the exepected trends for LM program changed wildly too. To ensure that the allocations across pools represented the new behaviour, the GEAR rewards were reallocated post a DAO vote a while back.
Credit Accounts 🥷
- Leverage Ninja Accounts: Over the month of November, a total of 272 CAs have been opened by our Ninjas, while 126 are still active. This is despite the limits on minimum borrowing and request based access for security reasons, which makes up for a very positive outlook. The numbers don’t exactly imply “146 have left” as some of them have re-opened or changed the underlying debt assets. Cumulatively, it has grown!
- Average Account Size: The average account size is $587k. The average collateral is $103k while the average borrow is $484k. This leads a total of $13m of collateral put up by our Ninjas.
- Average Leverage: The overall average leverage in the system is 5.69X which is a “healthy level” to be at. The most degen borrowers are those who borrow DAI — they are on average using 6.24x leverage. USDC is slightly lower, at 5.89x, and WETH borrowers are the safest at 5.12x leverage on average. WBTC at 7.82x but that’s because only 1 account is active with 7.82x leverage.
4. Health Factor: The average health factor of CAs is 1.09, which again is a moderately healthy score to have given the correlated-debt-to-position. AddCollateral option has been used 1.5 Times per account on average by our Ninjas to ensure health factors are in check:
5. Liquidations: Despite the leverage, we have only had 3 liquidations. This is due to Gearbox’s unique feature of enabling users to leverage up an asset without having to have directional exposure to the borrowed assets. If you have WETH collateral and you borrow WETH against it to stake in yearn, the volatility in ETHs value won’t affect your HF almost at all. This is because you are supposed to return the WETH you have borrowed and not it’s USD equivalent, effectively turning it into a WETH/WETH position.
Where does leverage capital get deployed to?
OK, lets cut to the chase. This is really what you’re here for, right? What are the strategies that are most popular in Gearbox? Where’s all the smart leverage ninja money going? In other words… Where’s the alpha?
Well, here’s the alpha and the key trends on what people that make money do right as well the ones who lose money, what they do wrong.
1. Convex (+ Curve). Convex has been leading in terms of attracting liquidity from the very beginning and it continues to do so. At the moment, there’s a total of $48M+ of liquidity flowing to Convex through Gearbox CAs. Where’s it going? The answer lies in 3 key pools, all based on stablecoins.
- Convex sUSD-3CRV Pool: sUSD pool at the moment is $20.3M or almost a 3rd of all outflow from Gearbox CAs, making it biggest strategy in the protocol. At avg leverage, this equals 20.11% APY. Sorry what is bear?
- Convex LUSD-3CRV Pool: At $9m, this is the 3rd biggest Gearbox strategy. With 11.5% APY at avg leverage, understandably so.
- Convex gUSD: Interestingly, at 20.5%, gUSD has the highest APY of all strats but at $6.9m is significantly lower than the top 3. This may be due to (fairly or unfairly) fears around GUSD and depeg worries.
2. LIDO: At $10.3m, leveraged staking with LIDO seems to be the second biggest use case apart from stablecoin farming. Another interesting thing to note is that LIDO is currently substantially more popular than both options for LPing stETH with ETH — either the Convex/Curve stETH-ETH pool or the Yearn stETHCrv strategy. This is likely because pure staking APYs right now are a bit higher than the combined APY from staking + LPing in Curve. Keep in mind that when you LP stETH/ETH in Curve, you are at least holding some portion of vanilla ETH, and you are not earning staking rewards on that portion.
3. Yearn: At $10m+ of liquidity being captured, yearn’s value is derived from 2 key vaults. Yearn DAI vaults at 10.9% APY hold $6.1m and Yearn WETH vaults at 6.8% APY holds $4.7m. This is interesting because despite the lower APYs the perceived safety of these vanilla asset vaults in Yearn might be allowing them to attract users.
4. FRAX: At $6.5m, the liquidity is currently split 4.4m in the Convex-Frax-3Crv pool and 2.1m in the Convex-Frax-USDC pool, which is interesting because at the moment this article is being written, the latter strategy has slightly higher APYs than the former.
While strats on integrations can give you hints of where to take your money to, below is a breakdown of the methods adopted to be the most profitable as well as the least profitable Credit Account.
Are Credit Accounts winning? — Yes, if done right!
Keep in mind that when using in-house Gearbox numbers, everything is viewed through a pessimistic lens. This is because Gearbox’s primary concern is, in a sense, “what is the minimum these assets could be worth?”. This attitude is important to ensure that we can always successfully liquidate an account for the amount we think it’s worth. In this sense, ‘profitability’ is somewhat understated and losses are somewhat overstated. See this part of our docs for more details.
There are rules to CA profitability, which hold true even w/o leverage!
1. You have to be active with Credit Accounts usage — DeFi yields change all the time. Passive gains are on the other side.
The truth is, there doesn’t exist a “most profitable strategy” in Gearbox v2, at least not so far. Gearbox doesn’t provide you with end solutions — it’s the protocols built on top OR your skills and alpha that do that. You need to stay active if you are a Ninja, yields in DeFi change almost every week!
The top few performing CAs, both in terms of total profit earned of ROI in % terms, are basically all using different strategies! From the 5 CAs with the highest ROI in % terms right now, one borrowed USDC, one borrowed DAI, and 3 borrowed WETH. Of these 5, two are WETH accounts that are fully in stETH. However, these two CAs are using very different levels of leverage — one of them is using 6x leverage (moderately risky) and one is using 3.7x (relatively safe). The remaining WETH CA is fully in the Yearn WETH vault (with 7.1x leverage).
An aside about borrow costs: it’s very important for all CA users to remember that your net APY will always be affected by the cost of borrow that you’re paying. Because of this, shifts in the borrow cost for each borrowable asset in Gearbox can change the profitability of your strategy, and in some cases even turn a winning strategy into a strategy with negative yield. If you are degen farming with high leverage (e.g 7x or more), we highly recommend that you keep an eye on the utilization ratios and interest rates for the assets you have borrowed so that you are aware if/when your strategies become unprofitable.
2. Timing farming entry points and pegs is important. Not all farming is truly passive, especially with LUSD & stETH.
This shows that depending on specifically when you entered your positions, your performance would have altered drastically — there was a couple of weeks where Yearn WETH was at ~10% and was outperforming stETH. Granted, a couple of weeks doesn’t sound like a lot, but remember, Gearbox v2 only went live at the end of October! The same holds true for stETH farming: if you are entering a 7x position when it’s 0.97 is much better than when it’s hovering around 0.99. Instant gains / loss.
3. APYs vary per protocol, check for their methodologies. Sometimes it takes weeks to hit the average, either lower or higher.
The DAI denominated CA is similarly leverage farming the Yearn DAI vault with (!) 9x leverage… and this CA has thus far received a good ROI. However, due to high DAI utilization right now (~83%), currently the Yearn DAI vault is almost making negative APY returns if you are using leverage.
Some protocols use 7d average, some 30d. As such, random spikes can influence the APYs drastically, so you need to be aware how often harvests happen for whatever you are sitting in. Slippage also affects this, the higher the slippage you encountered the more you need to farm into net-zero. This is all normal even with no leverage, leverage just amplifies the sensitivity.
I’ve saved the best (or at least the most interesting) for last…
The most profitable Credit Account in Gearbox borrowed USDC (drumroll please)… and bought WBTC. Chad trader calling bottoms? — They bought some WBTC on the 28th of November and some on the 29th. They are using (pretty comfy) 3x leverage, and BTC has gone up about 3% since then, so this Credit Account is up 10% in ~10 days.
While we often emphasise that Gearbox is *not* primarily a protocol for trading, but for capital efficiency. It’s possible that the cost of borrow on Gearbox might be lower than the annualized cost of funding rates on trading platforms, often at 15% — so if you’re only planning to use low amounts of leverage anyways (up to 10x) and don’t need 20x, then Gearbox is a reasonable venue for your moderately degenerate margin trading.
Outside of this god-tier WBTC trader, the ROIs on other strategies range from ~1.9%-3.4% per 1x. These numbers may sound low to you, but remember, Gearbox has only been live for a little more than a month. Now, it is not really kosher to simply annualize figures, because obviously market conditions can change, strategy APYs can change, and borrow costs can change, but just note that 1.9%-3.4% annualized (based on the amount of time v2 has been live) is approximately 17%-31%… Which, if you’re looking for yields, is not too bad at all.
Again, study the Leverage Bible!
And check the in-depth dashboard:
Least profitable strategies
Similarly to the most profitable strategies, it doesn’t seem that there is any specific strategy or asset that has lead to the largest losses in percentage terms for v2 CA users. But there is a significant difference between those that profited and those that didn’t, read below:
Trade frequency: A common pattern in all of the worst performing accounts are moving their money between strategies pretty frequently. These CA users seem to be chasing higher yields on a weekly (or more) basis.
- Every time you change strategies, you are incurring some non trivial amount of slippage — and,
- This slippage is magnified by the fact that you are using leverage.
It’s possible that these users are used to being small/medium sized farmers. However, when you are using leverage, suddenly your medium sized stack becomes moderately large (e.g 200k → 1m). Which potentially means much greater slippage, especially in cases where you are entering and existing assets with imbalanced liquidity.
So, it’s better to wait in your strategy and earn an APY 1–2% lower than the top one instead of losing money to slippage. “BUT, but if there’s no clear winning or losing strategy, where do I park my money?” — You can always refer to our allstars at the moment at 7x leverage, at least for now:
Protocol Health
With V2 having so many technical advancements, managing protocol safety and health becomes a crucial part. And below is a smol overview of how we have been doing.
Bug Bounty
We boosted our bug bounty program in order promote whitehack efforts to find bugs on 1st November. Through immunefi, we received and corrected the one report we received. The whitehack was paid $20,000.
Bad debt
0 bad debt in the protocol, this is due to a high efficiency of the liquidation mechanism that we have in place as well as partnership with Risk DAO.
RiskDAO
RiskDAO, as per our partnership, has been helping us improve our safety parameters. Their ideology is based off of constant monitoring and improvement, same can be read below:
RiskDAO has been providing Gearbox with weekly risk reports in the governance forums — you can see these here. It further makes recommendations for ongoing market conditions.
So far, RiskDAO reports have covered:
- The fallout from FTX (didn’t affect Gearbox protocol health at all, this is why DeFi must win)
- Genesis troubles, issues with Gemini Earn, possibility of a GUSD depeg and the potential effects it could have on Gearbox (TLDR we are safu)
- CRV shenanigans on Aave that lead to bad debt and why such attempted attacks would not work against Gearbox (TLDR: we only count 25% of CRV’s price as margin in CAs, so it could 75% during liquidation and still be fine — also, virtually nobody has CRV in their CAs in the first place.
- WBTC risks stemming from concerns about their multisig.
And for all the reasons above, Gearbox has renewed our partnership for another 3 months. You can see the passed vote in this topic.
Protocol Enhancements
Let’s just put it this way, devs are chads who after launching V2 have already made improvements and started planning *redacted*…
Gearbox V2.1
Actually let’s leave it for a different day…
PS: There’s alpha on this in one of our twitter spaces, just saying,
Oracle Enhancements
One underappreciated aspect of DeFi is how important well designed oracles are to the ecosystem and how oracle manipulation is one of the key ways that DeFi protocols can be attacked or exploited. In the short time since v2 launched, we’ve already had three oracle updates, all of which either improve protocol security or user experience.
NOTE: None of these things were “exploitable” — these changes to oracles were all proactive ways to improve the protocol, this was not a case of updates being needed to patch vulnerabilities.
- LUSD Oracle Update
The very first recommendation that RiskDAO made to us was to introduce an upper bound for the LUSD Oracle price at $1.1. This recommendation was made because A) LUSD has a history of trading above $1, and B) there is occasionally an imbalance in the LUSD curve pool where there is less LUSD and more of other stables.
This leads to a situation where a relatively small amount of funds can move the price of sharply LUSD upwards. Any time oracle prices of assets can be moved with a small amount of funds, you are opening up the possibility of economic exploits (AKA ‘highly profitable trading strategies’). According to RiskDAO, Gearbox was totally safe from such attacks thanks to the limits currently in place on Credit Accounts, but in a hypothetical future where those limits didn’t exist, there was a risk factor (for example, if someone pumped LUSD price to $1.5 whilst holding a large amount of LUSD in their CA).
This is why RiskDAO recommended that we introduce an upper bound on the LUSD price oracle, such that Gearbox never accepts a price for LUSD above $1.1, even if its technically trading higher than that. This recommendation was put forward and was approved by governance in GIP-27 and implemented. See this governance thread for the full discussion.
2. stETH Oracle Update
One additional change that we made for protocol health reasons was to update the stETH oracle. This change was made in order to improve the user experience.
The stETH oracle change was related to our first ever liquidation Gearbox v2. The user had a 9x leveraged position where they were borrowing ETH and holding stETH in their CA (leverage farming on stETH at truly degenerate levels). Which was partly due to the chainlink oracles deviation between stETH/USD and ETH/USD. You can find details in this thread.
In the end, Gearbox DAO approved a proposal (GIP-28) to update the stETH oracle — instead of using ETH/USD and stETH/USD to imply stETH/ETH, we created a composite oracle that takes ETH/USD and stETH/ETH and implies stETH/USD.
3. WBTC Oracle Update
As a consequence of the worries outlined above about WBTC, RiskDAO proposed that we update the WBTC oracles. Previously, Gearbox was using BTC/USD oracles, which are considerably more accurate on average because they are updated much more frequently. However, using a BTC/USD oracle would not capture any potential depeg event.
The proposal suggests a new oracle that combines BTC/USD and WBTC/BTC oracles into a WBTC/USD oracle. On average, this oracle would track the price of BTC somewhat less accurately, but in any major depeg event would allow liquidations to proceed in an orderly way.
This proposal was accepted by Governance in GIP-33 and to be deployed.
dApp Enchancements
The current main dApp is also getting some improvements, to make positions and farming more transparent and safer for the Ninjas:
- Your HF is modelled to show changes before you make any trade, borrow more, add collateral, or any other operations. Helps you figure out how good or bad things might turn before you press confirm.
- The borrow rate and rewards are visible on the main CA page.
- Extra rewards in $GEAR have started and are also visible in CAs.
- The prices are now dual (in terms of the interface) where you can switch between Spot and Chainlink. This is important as often spot gives a realistic view of the positions, while often Chainlink gives a less biased yet overly negative scenario. More power to users to decide!
5. An extra red notification is shown when your HF is truly too dangerous.
6. Gas optimizations (not new, just re-iterating the point) are so amazing even for super complex positions! Just check out for yourself.
7. More helping texts on ? marks to give users better information.
Keeps on improving! Suggest what you’d want ❤
Get GEARed up, have a cider! If you would like to earn passively, remember to LP 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:
