Why?

Agent-based simulation (ABS) is a computational approach for modeling the actions of agents in order to understand the behavior of a complex system.

ABS is used in industries ranging from self-driving cars to trading. In such applications, it is not easy to intuit how changes to a system may affect outcomes given the participation of heterogenous agents and the endogenous nature of the decisions they make. Further, experimental approaches such as A/B testing are usually infeasible because they take too long and are too costly. In these cases, ABS can deliver statistically rigorous results about a complex system by composing simple models of agents and their environments.

Agent Buttercup was originally developed as a way for our team to explore new pricing mechanisms within the Cozy V2 protocol. We found ABS to be an attractive approach to study DeFi markets, which are driven by a multitude of agents (LPs, traders, liquidators, etc.) optimizing different risk-reward functions. We want to open-source it in order to give DeFi researchers/engineers the tools they need to quantitatively:

• Evaluate the effects of new protocol mechanisms on market outcomes

• Optimize risk settings, model parameters and incentives

• Stress-test the robustness of a protocol against adversarial economic attacks and market environments

The hope is that it enables faster innovation of new DeFi mechanisms and a better grasp on the existing DeFi mechanisms.

Architecture

At a high-level, the core components of Agent Buttercup are as follows:

• State, an object which keeps track of:

  1. EVM state inside a https://github.com/bluealloy/revm instance

  2. Any non-EVM state inside a separate World object

• Agents, who in each step, decide what to do based on the current state and send transactions to update the EVM or World state.

• Time Policy, which determines how block numbers/timestamps progress in each step.

• Summary Generators, which compute summary statistics from state and write them to a file for analysis.

• Manager, the core object responsible for coordinating the state, agents, time policy and summarizers. The manager effectively runs a loop. In each step:

  1. Agents are processed in parallel across multiple threads and asked to send the manager any updates.

  2. Updates are executed against the EVM and world state.

  3. Update results (Did the EVM tx revert?, etc.) are temporarily cached in state.

  4. Agents are given the ability to read state and update any local state they track.

  5. Summary generators are run and statistics are written to the file.

  6. The time policy is called and the EVM block number/timestamp are updated.

Comparison to Alternative DeFi ABS Frameworks

The use of ABS in DeFi is not new.

The DELV team built a Python-based ABS for their protocol on top of Apeworx. To deliver high-performance simulations and lean into the growing ecosystem of Rust-based Ethereum tooling, we decided to build on top of https://github.com/bluealloy/revm. Using revm directly allows us to avoid I/O overhead associated with local node implementations like Anvil, offer maximum customizability, and easily integrate with other revm-based infrastructure like https://github.com/paradigmxyz/reth and https://github.com/foundry-rs/foundry.

Agent Buttercup was originally inspired by the work of the Primitive Finance team on Arbiter. However, the two frameworks are quite different implementation-wise and we assume they will ultimately serve different use cases. The key differences are as follows:

1. Notion of time: The core of Agent Buttercup is a simulation loop, where the block timestamp/number moves forward in each step per a customizable time policy. In Arbiter, there is no built-in notion of time or steps. Time is crucial to the Cozy V2 protocol where market utilization can decay over time. Further, we found the use of steps natural in an ABS framework where agents may need to know they are reading the most up-to-date state before deciding next actions.

2. Notion of state: Agent Buttercup defines a fully customizable definition of state, including EVM state and non-EVM World state. In each step, agents are allowed to read and take actions which update the EVM and/or World. In Arbiter, there is only EVM state and agents cannot read full state. Instead, all agents run completely asynchronously and all actions result from responding to events they receive from each other via a message-passing approach.

Call to Action

Agent Buttercup is now fully open-sourced.

We encourage DeFi researchers/engineers to try it out and let us know what they think. Further, we are committed to improving the framework, making it faster, and integrating it with the rest of the open-source ecosystem. If you are interested in contributing, please get in touch. We welcome collaboration. In the near future, we would like to support forking a chain onto the revm instance and exploring ways to accelerate the simulation loop.

In the Cozy V2 protocol, a market’s cost model sets the costs for purchasing protection and refunds for cancelling protection. Costs paid by protection purchasers are set aside in a fee pool and slowly accrue to protection suppliers as yield.

Each cost model is a contract required to implement a minimal interface, which includes:

• A cost factor function, which outputs a cost as a percentage of the purchase amount.

• A refund factor function, which outputs a refund as a percentage of the market’s fee pool (assets which are still pending accrual to suppliers).

Both of these functions take the market utilization before and after the purchase/cancellation as inputs. So, much like DeFi lending markets where interest rates depend on the pool utilization, cost/refund factors can depend on market utilization.

Exactly how cost/refund factors should depend on utilization is up to set creators using the Cozy V2 protocol. They have flexibility in specifying the cost models used in their sets.

Cozy V2’s jump rate models

A common interest rate model for lending markets, popularized by Compound and Aave, is the jump rate model. A similar model can be used in the Cozy V2 protocol as a cost model.

A jump rate model defines the cost factor as a piecewise, increasing, linear function of utilization. The kink defines a target utilization, after which costs increase rapidly to help keep utilization less than 100%. This ensures that suppliers are able to withdraw assets whenever they want.

Importantly, the jump rate model is static: its parameters are fixed.

Problems with static cost models

Static cost models can sometimes lead to inefficient price discovery and sub-optimal capital efficiency.

As a simple example, consider a jump rate model where the cost factor for the first 10% of protection is 5%. In case the demand-side of the market deems this cost factor too expensive, the market will sit at 0% utilization indefinitely. Suppliers are also unhappy since they are earning no yield from this market. In an ideal world, as time passes and no one purchases protection, the cost factor should come down from 5% to 4%, 3% and so on, until the market clears.

Dynamic cost models can support exactly this type of behavior, automatically adjusting parameters over time in response to supply/demand. Two clear advantages from these adjustments are:

1. As the market’s risk perception evolves over time, they obviate the need for lengthy governance processes to modify parameters or manual configuration updates.

2. Cost models become more resilient in the case of misspecification. When set creators using the Cozy V2 protocol do not have a clear sense of what the fair price of protection should be, they can use a rough approximation and expect the model to eventually move towards the market-clearing price.

Cozy V2’s dynamic level cost models

One solution for set creators using the Cozy V2 protocol is to use the dynamic level cost model, which is a simple extension of the jump rate model.

Here there is an optimal utilization zone,[uLow, uHigh], and uOpt = (uLow + uHigh) / 2. What makes this model dynamic is that when market utilization is above uOpt, the cost factor in the optimal zone automatically increases to make protection more expensive. When market utilization is below uOpt, the cost factor in the optimal zone automatically decreases to make protection less expensive. Visually, the curve between uLow and uHigh (colored in green) can move up or down.

The amount that the curve moves up or down on a given purchase/cancellation is dictated by how long the utilization as deviated from uOpt and the size of the deviation. The more that utilization has exceeded uOpt and the longer it has exceeded uOpt, the more the curve moves up. Similarly, when the curve moves down.

For more details on the inner workings of the dynamic level cost model, check here.

Other dynamic cost models designs

The dynamic level cost model above is just one of many possible ways to make cost models more responsive to supply and demand over time. It is broadly inspired by the idea that many DeFi mechanisms boil down to some form of PID controller.

Many DeFi lending protocols have explored much more complex, PID controller-type interest rate models, including Silo Finance, Euler Finance and Mars Protocol. As mentioned previously, set creators using the Cozy V2 protocol are free to experiment with similar designs in their own cost models.

Problems with dynamic cost models

It is important to point out that dynamic models are not a panacea.

First, dynamic models will tend to be more computationally intensive and consume more gas. In the Cozy V2 protocol, the dynamic level implementation requires updating storage variables on each purchase/cancellation, while the jump rate implementation does not.

Second, in some scenarios, dynamic models can reduce capital efficiency. Chitra, Erins, Kulkarni 2022 analyze lending markets and show that if supply/demand is very inelastic, dynamic interest rate models may permit someone to move rates such that they earn an outsized share of the pool’s yield. Mitigating this outcome would require targeting low pool utilization or implementing supply caps, which makes suppliers worse off. An analogous strategy will tend to be prohibitively expensive in the Cozy V2 protocol given time-locks on protection purchases and set shares.

Using simulations to evaluate cost models

One question that may be of interest to set creators when setting their cost models is:

Should I expect dynamic level cost models to outperform jump rate models when I do no know the market’s fair price of protection?

In a vacuum, it is difficult to quantitatively predict the effects of a particular cost model. A market’s utilization and the yield earned by suppliers is a complex function of the optimizations made by heterogeneous actors interacting with the Cozy V2 protocol.

A useful approach in such cases is agent-based simulation. Cozy’s simulation framework, Agent Buttercup, allows DeFi researchers/engineers to answer exactly these types of questions. Below, the framework is leveraged to extract some insights.

Simulation Set-up

The simulation consists of a single market in the Cozy V2 protocol and does a side-by-side comparison of the jump rate and dynamic level models.

For the exact simulation set-up, please check here.

Cost Models Set-up

In each side-by-side comparison, the cost models:

• Have a target utilization of 75%

  • Jump rate model is set so that kink = 75%.

  • Dynamic level model is set so that uLow = uHigh = 75%.

• Have a minimum cost factor of 0.50%

• Have a maximum cost factor of 10%

• Begin with a cost factor at target utilization of x, where x varies from 0.5% to 8.5%.

Notice that with this set-up the dynamic level and jump rate model start off looking exactly the same, but the optimal zone (which happens to just be a kink since uLow = uHigh) in the dynamic model will move over time.

Agent Definitions

There are three types of agents interacting with the market:

• Passive suppliers, who are price-insensitive and always willing to underwrite protection.

• Passive buyers, who are price-insensitive, have a target protection amount and are always willing to buy protection to reach that target.

• Active buyers, who hold the view that the fair price of protection is 4.5%. They will purchase protection when the cost factor is less than 4.5%. They will sell protection when the refund factor is greater than 4.5%.

Agents are initiated with a random budget of the market’s asset and take actions with random waiting times governed by a Poisson process. Passive buyers are initiated with random target protection amounts.

Simulation Results

The results below show the median utilization and supply APY over the 75-day simulation. The table on the left indicates that the dynamic model always gets the market closer to target utilization, 75%. The table on the right indicates that the dynamic level model usually tends to get suppliers higher yield.

The time series plots below show the simulated cost factors and utilizations for the case where the kink cost factor starts at 8.5%. The dynamic cost factor trends downward, as the model slowly adjust to the fact that active buyers view an 8.5% as too expensive. As a result, dynamic utilization trends upwards. The jump rate model gets stuck with utilization hovering around 40%, as active buyers are unwilling to make purchases which push the price over 4.5%.

Overall, these results support the hypothesis that dynamic level models can outperform jump rate models when set creators do not have a precise estimate of the market’s fair price of protection.

DISCLAIMER: This article is for informational purposes only and is not financial advice. Please do your own research and consult a financial advisor before making any investment decisions.

Creating a market for stETH peg protection

While other protocols were unable to offer markets covering stETH, users of Cozy v2 were able to quickly and easily create a market for stETH, using the permissionless market creation mechanism. The resulting market “stETH Peg” would pay out if the price of stETH diverges from the price of ETH by 50% or more.

To hedge their stETH exposure, the DAO would acquire an amount of protection from that market at a cost that felt fair to the members. If the market triggered, the DAO would be able to redeem their protection for USDC.

Empowering the DAO to buy protection L2

But there was one other problem: the DAO needed a way to buy the protection completely trustlessly, without relying on any privileged third parties. Whereas regular, eligible users can simply go to the app and buy protection, the DAO needed a way to buy protection entirely through transactions executed through governance proposals.

Audited helper contracts enabled the DAO to bridge its assets from L1 to the CozyV2 protocol on Optimism where it was able to place an on-chain bid for the desired protection. These helper contracts are open sourced and can be used by NounsDAO and any other DAO to establish and manage a footprint on L2. To learn more about the helper contracts, visit this proposal.

Placing an onchain bid for protection

NounsDAO used the helper contracts to place an onchain bid for stETH peg protection at a 2% cost rate. This bid presents an on-chain arbitrage that should attract economically rational third parties to fill the order.

Once filled, the DAO will have an active hedge on a portion of their stETH exposure. From there, the DAO can choose to increase the amount of protection by placing another on-chain bid. Or choose to hedge other assets like the newly acquired Rocketpool ETH using a similar mechanism.

A tool for DAO risk management

This proposal is a powerful example of how Cozy V2’s on-chain, permissionless protection can be used not only by individuals but also DAOs to protect funds from protocol risks. If you want to learn more about how you or your project can manage risk with Cozy V2, visit cozy.finance.

In this blog post, discover how Cozy v2 empowers yield seekers and project leaders to develop and access customizable protection markets.

Permissionless and Modular Market Creation

Cozy v2 solves this problem by empowering users to create markets. The protocol provides a step-by-step guide to develop triggers, essential for determining payouts in protection markets. Here's how you can develop triggers using Cozy v2:

Click on "Your Triggers."

Once connected on the Cozy app, click your wallet address dropdown in the top right and select “Your Triggers” to go to the Cozy Admin section.

First, go to the admin section of the Cozy app and select “create a protection trigger.”

Choose the desired trigger type template.

Select a trigger template or create your own.

Chainlink

The Chainlink template compares a chainlink price feed with either another chainlink price feed or a fixed value. This type of trigger is useful for pegged assets like Dai or stETH.

An example of a market using the Chainlink template is the stETH Peg market.

UMA

The UMA template presents a query to the UMA Optimistic Oracle and triggers the market if it returns a YES response. This type of trigger is useful for more general markets.

An example of a market using the UMA template is the Aave v2 market.

Custom

If you want more custom logic, you can create your own trigger. For example, if you want a multisig to make the decision. Or a combination of a programmatic trigger and a multisig.

Configure your trigger

Once you’ve selected your template, configure the trigger by selecting the price feeds, writing the UMA query, or developing your own custom trigger.

Chainlink:

UMA:

Add metadata to your trigger

Provide an image, a title, and a description so that users can most easily use of your trigger.

Review and deploy your trigger

If everything looks good, you can deploy your trigger. It will now be usable by set admins when they are creating and managing their sets.

If you want to create a market with this trigger, you can create your own set and use this trigger to create a market.

Conclusion

You can create a trigger today in the app. Have questions? Join the discord and ask the community.

Disclaimers: Investing in DeFi protocols and assets carries inherent risks. This guide is for informational purposes only, and is not financial, legal, investment or other advice. Please refer to specific market terms and cozy.finance to learn more about coverages and risks. This material and the Cozy protocol are not intended for use by US persons.

In early March of 2023, Euler Finance was hacked for nearly $200M. At the time, Cozy v2 was in early access and had an active Euler Finance market. This market triggered and paid out.

This post guides you through the entire Cozy v2 protection life cycle, using the Euler market as an example. It details each step, from purchasing and pricing to maturation, decay, market triggers, and claiming or selling back protection.

Purchasing and Pricing Protection

First, the user gets protection.

• Select the market

• Enter the protection amount

• Review the price

• Buy the protection

Here’s the transaction where the user bought $1,000 of protection for $11.

The price is based on the cost curve that you can see on the market details page.

Protection Maturation

After purchasing protection, there is a 12-hour maturation period during which the protection is not yet active.

This period is designed to prevent abuse of the system and ensure that users do not purchase protection immediately before a market triggers.

Once the maturation period has passed, your protection is active.

Selling Back Protection

Cozy v2 allows you to sell protection back to the protocol after it has matured.

This option is particularly useful for users who no longer require protection or wish to adjust their protection levels.

Your proceeds from selling back protection may be higher or lower than your initial cost, depending on how much your protection decayed and the relative utilization of the market.

Protection Decay

Protection value in this market decays at a rate of 50% per year. This means that if held for a year, the user’s initial $1000 of protection would still cover $500 at the end of the year.

Market Freeze and Trigger

This Euler market uses the UMA optimistic oracle to determine whether to trigger.

If a yes answer is reported from the following query, the market triggers and users can claim payouts.

Was there a hack, bug, user error, or malfeasance resulting in a loss or lock-up of tokens in Euler (https://app.euler.finance/) at any point after Ethereum Mainnet block number 16175802

As you can see, a yes answer was proposed on March 13th and accepted on March 20th.

When the answer is proposed, the market goes into a frozen state. And when the answer is settled, the market goes into a triggered state.

When a market is frozen, users can no longer buy protection from the market. And protection providers cannot withdraw funds. These restrictions protect against front-running.

When a market triggers, users with matured protection can redeem protection for a payout by claiming.

Step 6: Claiming a Payout or Selling Back Protection

Since the market has triggered, users with matured protection can claim a payout.

You can see the transaction for this here.

This user paid $11 for $1000 of protection. By the time the market triggered, they were able to claim $842.

Conclusion

Cozy v2 protection is simple to get and use. Buy the protection and if the market triggers, claim a payout. Or if you don’t need the protection anymore, sell it back.

If you want a more in depth understanding of the mechanics, evaluate the pricing, maturation, decay, and trigger logic.

And if you have further questions, check out the docs or ask in discord.

Disclaimers: Investing in DeFi protocols and assets carries inherent risks. This guide is for informational purposes only, and is not financial, legal, investment or other advice. Please refer to specific market terms and cozy.finance to learn more about coverages and risks. This material and the Cozy protocol are not intended for use by US persons.

In this post, we walk through the process of buying protection on Cozy v2 step by step.

Step 1: visit Cozy and choose a market

Visit the Cozy v2 app. You'll see a list of available markets, each representing a specific DeFi protocol or asset. Find the one for which you want protection, like Lido staked ETH (stETH).

Step 2: enter an amount of protection

Decide how much protection you need. Enter the amount in the provided field.

Step 3: purchase protection

Click "Buy Protection" and follow the prompts. Connect your wallet and complete the transaction. Once confirmed, you've bought protection.

Step 4a: sell back your protection

If you no longer need your protection, you can sell it back to the protocol at any time. This means that you can keep your protection for as little or as long as you like, with no lock-ups or expirations.

Step 4b: claim your payout if the market triggers

If a hack or vulnerability occurs, you can claim your payout. Visit the Cozy Finance website, go to the relevant market, and click "Claim." Follow the instructions to complete the claim and receive your payout.

Conclusion

Buying protection on Cozy v2 is a simple way to hedge against smart contract risks. Try it today at cozy.finance.

Disclaimers: Investing in DeFi protocols and assets carries inherent risks. This guide is for informational purposes only and not financial, investment or other advice. Please refer to specific market terms to learn more about coverages and risks. Always do your own research and consult a financial professional before making investment decisions. This material and the Cozy protocol are not intended for use by US persons.

Ways to use Cozy

Hedge against risks like smart contract hacks by purchasing protection, or earn yields by providing protection. Don’t see the market you’re looking for? You can use the protocol to create markets yourself. Cozy v2 has the right market, at the right price, at any time.

Buying protection

If you’re a staker, lender, LP, or earning on-chain yield some other way, buying protection on Cozy can help protect your principle from tail risks like the billions of dollars lost in smart contract hacks every year.

Providing protection

Or if you’re a more advanced yield seeker comfortable with underwriting your own risks, providing protection on Cozy can help you earn high risk-adjusted yields.

Creating protection

Other venues for protection suffer from restrictive designs for price, payout, and risk management. Cozy v2 is designed with the philosophy that the best results come from permissionless creation and market forces. Anyone can create a basket of risk with markets that determine price and payout logic as they see fit, increasing the likelihood that a user finds the protection they need.

Key features

Key features include permissionless creation of markets, predictable and automatic payouts, programmatic pricing, and high levels of capital efficiency.

Permissionless market creation

You can create markets that cover anything--whether that’s depegs, hacks, custodian risk, or something else entirely. No gate-keepers here.

Predictable and automatic payouts

Buy protection on Cozy v2 and get paid out transparently and automatically, without an opaque claims process.

Programmatic pricing

Pay a fair price, set by programmatic price discovery.

Capital efficiency

Earn higher yields for providing protection and pay lower prices for buying protection with Cozy v2’s capital efficient design.

Cozy v2 is live

Cozy v2 is live on Optimism in a guarded launch.

Try Cozy v2 today.

Over the next few weeks, we will publish more in depth explorations for each aspect of the protocol. Follow on twitter and subscribe to the blog to be the first to read.

Disclaimers: Investing in DeFi protocols and assets carries inherent risks. This guide is for informational purposes only and not financial, investment or other advice. Please refer to specific market terms to learn more about coverages and risks. Always do your own research and consult a financial professional before making investment decisions. This material and the Cozy protocol are not intended for use by US persons. Any yields and other figures represented in this document are estimates and are not guaranteed.