DISCUSSION: Gearbox on L2s, Redefining Scale


L2s have become the rage and potentially the next source of growth for DeFi. Deploying on L2s is a sure shot way of increasing user base and TVL both but opportunities are also dependent on viability. For Gearbox, L2 deployment hadn’t been talked about significantly, while they were always in consideration, for 2 reasons. First, Gearbox is a 2nd degree DeFi protocol, that means it builds on top of existing DeFi protocols and thus has direct dependency on their deployment. Meanwhile, leverage in itself is a high liquidity requiring application on chain. This means it’s not just about the presence of existing AllowedList and more assets and integrations on an L2, but their growth and liquidity that dictate when deploying on L2s becomes safe for Gearbox. While L2s have grown, their size in comparison to mainnet is still not significant. Liquidity further poses risks if any size were to be attained. Our existing integrations are then further limited in terms of reach on L2s. Native L2 integrations meanwhile again pose liquidity risks, this meant scaling to L2s was unlikely…

Or so was the case till V3 was introduced.

V3 Quotas: Limiting protocol exposure to reduce liquidity risk

With V3, we now have Quotas. Quotas are effectively limits on the overall exposure that the protocol will have to a certain asset or integration. This means that based on the liquidity available on L2s we can limit the use of a certain asset to the point where we are liquidating the said asset won’t cause bad debt. Hence, ensuring that even if we deploy to L2s, we simply need to increase the exposure as L2s scale. This gives us speed to scale, a higher scope of scale and more avenues to which this scale applies to.

You can read more about Quotas in the V3 introduction article.


And given Gearbox now has the capability to do so, it’s time we start a stronger discussion around the topic. This pre-GIP focuses on discussing these different facets of various L2s to understand where Gearbox should pilot its initial L2 deployment.

One major advantage of deploying Gearbox on L2 blockchains is the potential for cheaper transactions. Although Gearbox smart-contracts are quite gas efficient, leverage strategies require a lot of inter-protocol interactions and it’s not really viable to borrow less than $20-$30k. It’s also vital to make the liquidation process profitable, otherwise liquidators won’t be interested to be a part of the ecosystem. L2 are able to substantially decrease “entry ticket” to attract small fish users and at the same time cheaper transactions might allow more sophisticated strategies for whales.

While the above makes L2 deployment a good value add, we need to evaluate the various L2s on factors apart from EVM compatibility. These include reputable top-tier trade/farm protocols, Chainlink price feeds, and deep DEX liquidity for various assets.


Based on these factors, we will compare the top 3 Layer 2 Ethereum networks: Polygon, Arbitrum and Optimism. The comparisons will not just include the growth aspects but also the risk aspects of the scaling. Basis the report on the same: Risk committee and Risk DAO will be giving their inputs, BD and VIBES will be giving growth inputs and Devs will be evaluating the possibility of deployment as well as the “Wen?”. The community is encouraged to add feedback and question the responses as well as add any additional points they feel have been missed out for the deployment. Further, we would like to invite specific L2 teams to add anything they feel we are missing out in the evaluation.

The goal is to give the community the time to discuss thoroughly the deployment on L2 and then take it forward basis the final GIP results.


1. Protocol Presence

Here is the table showing deployments of the protocols that are already integrated into Gearbox because of their reputation, deep liquidity, and farming opportunities:

Protocol Polygon Arbitrum Optimism
Uniswap + + +
Curve + + +
Yearn - + +
Convex + + -
Lido + - -
Balancer + + -

As one might see, Arbitrum and Polygon have 5 out of 6, while Optimism has 3 out of 6 with convex being the key difference. Convex is Gearbox’s biggest integration, making up for almost 60% of the utilisation on the mainnet.

2. Chainlink Price Feeds

Chainlink has price feeds on Polygon, Arbitrum, and Optimism. Gearbox takes risks very seriously and allows only top-tier assets and the majority of them have price feeds available on all of the three chains. All the available feeds are listed here Price Feed Contract Addresses | Chainlink Documentation

3. Liquidity

Thank god we have llamas and their amazing dashboards!

Arbitrum(1.98B) is clear of Polygon (1.07B TVL) and Optimism (0.98B TVL),With a strong trend in terms of TVL increase in the favor of arbitrum as well. While Arbitrum clearly leads the pack, it’s TVL is still only 6.7% of ETH mainnet. This is where the liquidity issue springs up.

Let’s take a look further at the relevant protocols’ liquidity:

As one can see, the liquidity on L2 is 20x-100x lower than on Ethereum.

For instance Uniswap USDC/WETH pool on Polygon has $7.7M (the largest relevant pool), on Arbitrum ETH/USDC holds $23.4M, on Optimism WETH/USDC is $6M TVL.

The largest Curve pool on Polygon is DAI/USDC/USDT worth $18.6M, on Arbitrum USDC/USDT with $27M, on Optimism DAI/USDC/USDT/sUSD worth $32M. The daily volumes on those pools are around $0.2-$0.6M.

To understand scale, Convex’s TVL through Gearbox is bigger than Convex’s TVL on arbitrum and polygon combined.

In conclusion, although deploying Gearbox on L2 blockchains like Polygon, Arbitrum, and Optimism offers the potential for wider adoption and from the infrastructure point of view everything is rather good (trustworthy protocols and Chainlink price feeds are in place), the lack of deep on-chain liquidity and the potential for reduced security remain the major blocker. This effectively limits the quotas on L2s severely.

4. Bridges

Low liquidity on DEXes can lead to various price manipulations, making it difficult for liquidations to work properly. Accordingly, it is necessary to consider how quickly such manipulation can be stopped. To answer this question it is important to look at the work of bridges. Below are some estimations to get 1M USDC from Mainnet to this L2s (this data has been got as is and ofc could change depending on different market conditions)

Arbitrum BSC Optimism Polygon
Across 4min N/A 6min 3min
Anyswap 19min 23min 17min 26min
Arbitrum bridge 19min N/A N/A N/A
cBridge 4min 5min 3min 9min
Hop 1min N/A 1min 24min
Optimism bridge N/A N/A 11min N/A
Polygon bridge N/A N/A N/A 25min
Stargate 2min 2min 2min 2min

Also it’s good to mention that the largest CEX Binance (which owns 50%+ of CEX trading volume) supports deposits/withdrawals for all these L2s/Chains.

As you can see, the bridge landscape is quite competitive now and it’s almost always possible to have an arb-ing cycle between Mainnet and L2 in several minutes. The problem could arise for cases when big volatility is on the markets - in such cases Bridges can also face a lack of liquidity…

5. Network Congestions

And last but by no means least important are network stops. We will not give specific examples here - I think everyone can find them if they wish. Unfortunately this happens for all early stage L2s/Chains. And this is a big problem for any lending systems - since liquidations at such moments are impossible in principle. Any price volatility in such moments leads to bad debt.


To Summarise, L2s can become a source of DeFi growth with cheap and fast transactions but atm have liquidity issues. While Gearbox V2 couldn’t enable safe L2 presence with the available liquidity, V3 has that capability with Quotas. Quotas enable that by limiting the exposure of the protocol to an asset in order to ensure it’s safe to liquidate without creating bad debt. With this possibility, the Pre-GIP aims at evaluating the key L2s.

Arbitrum, Polygon and Optimism are the L2s under consideration. While they all have chainlink oracles, Arbitrum has the highest number of relevant protocols present. The size of these protocols in relation with mainnet and current Gearbox size is minuscule, Arbitrum is 49% of the TVL share amongst the three. The liquidity across is low though Arbitrum atm has the highest liquidity of the 3.

While the growth numbers mean Quotas won’t be too high, it can help increase speed of scaling as and when L2s grow. But, the bridges still aren’t extremely robust and can lead to issues during high volatility. Network congestions which we have previously seen are also a possible issue yet to be tackled and these problems are common across the 3 L2s.

It’s now onto the DAO to discuss and decide on the steps ahead.


cool!!!!!I support!!!!!!!!!!!!!!!!!!!!!!!!!!!

I like the idea to deploy Gearbox on ARB/OP with opportune limits, but only if devs can develops new adapters to emerging protocols.
Some examples


I do think a more L2 native approach is required for this expansion to live up to its potential. Many already integrated protocols that may dominate on mainnet are often small players in the L2 ecosystems and tend not to have the deepest liquidity or best farming opportunities.

Some examples. Velodrome has more than 2x the combined TVL of Uniswap and Curve on Optimism. While GMX has more TVL than Uniswap, Curve and Balancer on Arbitrum. Even Camelot on Arbitrum has more TVL than Curve and Balancer. Not to mention Yearn which has less than 300k combined TVL on Optimism and Arbitrum.

The launch strategy for each of these chains needs to be tailored to the specific ecosystems and strengths of the L2.

1 Like

First, congrats team for being one of the growing and steady protocols during turbulent times. Looking on the market trends (i.e. Arb airdrop, zkSync & Polygon zkEVM mainnets) definitely L2s are the way to go - great potential for growing the userbase.
RedStone would be happy to assist with the expansion, especially in areas such as:

  • Additional security mechanisms preventing bad debt, for example a 2 dimensional price (what would the actual price for liquidating be of a specific position, at current liquidity conditions).
  • Delivering new collateral types on new chains, together with backtesting analysis, assessing risks and helping to choose the secure & demanded ones
  • Custom Price Feeds, including data aggregation methods minimising risks coming from lower liquidity of the underlying assets like LWAP (Liquidity Weighted Average Price)

Also, happy to work shoulder-to-shoulder on other types of custom implementations, i.e. effective liquidation flows :saluting_face:

1 Like