Grant for Defishy to create VaR simulation model with Euler Liquidation Health Graph View implemented in eGP1

Archive
1

Title: Grant Request for Defishy.xyz to create VaR simulation model with Euler Liquidation Health Graph View implemented in eGP1

Author(s): Charlie Ambrose, Alex Craig, Luke Xie (Founders of Defishy.xyz)

Submission Date: October 10, 2022

Simple Summary
We are requesting a grant to create Value at Risk simulation modeling with the Euler Liquidation Health Graph View that we implemented in eGP1. Value at risk (VaR) is the potential loss in value of a given portfolio/asset given a specified time period and confidence level. VaR modeling helps users understand the potential worst case scenario based on historical and probabilistic parameters. Our VaR modeling will measure Borrowed Capital at Risk and assess how likely it is that, based on current assets and historical metrics, an account’s liquidation health score will drop below 1 or to a place of liquidation. Liquidations, especially from whales, can trigger liquidation cascades, which can result in massive asset price drops and further liquidations–this occurred in May 2022 and was a primary contributor to the market crash. These new Euler modelings will help to protect Euler and Euler users from a repeat of this liquidation cascade scenario.

Abstract
The Defishy team is composed of three members of Dartmouth Blockchain. Our philosophy is to continually provide value to the blockchain ecosystem through shipping useful products and research. The Defishy team consists of three individuals with development, design, and growth experience. We have extensive experience in DeFi, having contributed at Yeti Finance, NEAR, Temple DAO, and more.

Proven Euler Shippers
The team recently completed eGP1 for Euler, which is currently live on Defishy.xyz. The grant led to previously unforeseen insights for Euler users and the Euler team. eGP1 was started on 8/10/22 and finished within its expected timeline on 8/31/22. The current dashboard allows users to view current liquidation health scores across Euler and helps to prevent further liquidation cascades like the one seen in May 2022.

Motivation
There are a number of risk vectors in borrowing and lending protocols, including smart contract risk, price oracle manipulation risk, governance / pool management risk, and market risk (liquidations + bad debt). As protocols like Euler continue to onboard more users, it is increasingly important to equip users with necessary risk-management analytics for healthy activity. Safeguarding users against risk ensures the long-term sustainability of the protocol, as well as an improved user experience (competitive differentiation with other borrowing and lending protocols).

Specification & Implementation
We are requesting a $20k initial grant (all $20k to be paid upon completion of the grant deliverables–-no payment required up front) and $3.5k recurring on a monthly basis (paid on the first of each subsequent month the Euler community wants the app to be live, starting once the initial grant work is complete) to fund the development of a full data analytics/risk management suite for Euler with the following deliverables:

  • VaR equation development for understanding borrowed capital at risk and likelihood of liquidation health score below 1 (including VaR simulation for user-selected time periods)
  • Updated UI/UX from eGP1 that includes ability for users to adjust time frame for different simulation time periods
  • Integration of current Euler Graph View with VaR simulation modeling
  • Advanced filtering mechanics:
    • Ability to filter both graph and table by asset types used for collateral and borrowed position
    • Ability to filter by size of collateral (with ability to eliminate self-collateralizing positions that will not be liquidated)
  • Pop-up breakdown of a wallet’s composition (health score, borrows, and collateral) when hovering over a wallet in the Graph View

The deliverables will be live at app.defishy.xyz/euler.

How long will development take?

  • We estimate development time to take ~7-9 weeks from the proposal vote going though. Based on feedback from community members, we plan to continue iterating and shipping out new and useful features.
  • We will ship each deliverable/section when it is ready as opposed to waiting 7-9 weeks to ship everything at once. This will allow for users to use the tools sooner as well as allow for more iteration and feedback.

What will the grant payment be used for?

  • Platform Upkeep (Database storage, Server hosting, Ethereum Nodes, etc): $3,500 + running costs covered each month
  • Contributor compensation: $16,500 + $2,500/month
    • This contributor compensation is not retroactive for any updates we make to the current dashboard from eGP1. It is for our team to cover each member’s time cost to work on the grant over the allotted time frame and will cover design, frontend, backend, inspecting smart contracts, etc.
3 Likes
2

Hi 0xBroze!

May I ask u, have u ever previously created a Value at Risk simulation modeling tool ? Maybe for some other protocols? It would be cool to look it a prototype or something similar to understand how it going to look like.

3

These estimates are kind of unprofessional, IMO. Can you please break down the development hours and all the running costs? Do you understand that you ask for $42,000.00 per year + running costs?
What are the running costs? Are they $2,500.00 per month? If so, that would amount to another $30,000.00 per anum.

4

Hey @Raslambek --Thanks for this comment—we have created tooling for interpreting and displaying liquidation health modeling on both Aave and Euler (with Euler also having the Graph View). This work has positioned us to a spot where VaR makes the most sense. I like your comment on the visual idea/prototype, and we will have mock-ups for this as one of the first stages of the grant with the UI/UX adjustments. These will be shown to and approved by the Euler community before they are implemented. Let me know if I can provide any other info :v:

1 Like
5

Hey @NOA - appreciate you reading through. Here is further breakdown of those costs:

3-person team each spending about 20 hours per week on this grant over 8 weeks = 480 hours

$16.5k / 480 hours = $34.38 per hour.

“Running costs” (probably a poor term) allow us to nix a big payment request up front and allow us to be flexible with platform development while requesting a much lower amount. Total annual cost for this grant would be $42k. Hopefully that clears up some of the confusion with all the numbers in the post above.

Monthly payment ensures more value as well as incentivizes assistance from the team if something goes wrong/needs fixing post development.

The $3.5k is not additional to another $2.5k per month—we are only requesting $3.5k/month total (for the reason above).

Let me know if I can help clarify anything further.

6

Hey @0xBroze.Defishy, thank you for the reply.

20k for development; this is understood.
What is the other 22k for?
Could you make a breakdown of all costs, please?

7

For sure—here is a breakdown of the costs:

$16.5k (upon completion) - contributor compensation for initial grant deliverable work

$3.5k (upon completion) - initial storage, database, and app upgrade costs

Total upon completion = $20k

$2.5k per month - team compensation to maintain app, maintain database, make adjustments, take feedback after set development date is complete, etc

~$1k per month - App Upkeep costs

  • Database storage (~$250-475/month)
  • Server hosting (~$120-300/month)
  • Ethereum nodes (~$100-220/month)

Total per month = $3.5k

Please keep in mind that all these costs are including the increased storage we will need for historical data required for VaR modeling.

To clarify: $20k will be paid on completion of the grant and delivery of deliverables (no up front payment requested here). After this, those monthly costs above would make up the rest of the breakdown

8

I think the work previously done by Defishy was great: Defishy

And the current proposal encompasses a very useful tool for the community to assess activity going on on Euler. Let’s see what the community says but I am broadly supportive.

2 Likes
9

can you please provide some data on usage of dashboard produced under eGP1.

views, clicks, recurring views etc.

can you give some colour on user profile for this VaR thing. Also helpful is use scenario.

1 Like
10

Legitimacy is not in question here. Defishy are a reputable group, but the offer is overpriced compared to what the ShippooorDAO guys are building and all the functionalities they are implementing.

11

just spoke to llamas, it seems the other dashboard Defishy built is baiscally what llama liquidation offer (for free) but with a cool bubble graph.

Top wallet, usdc cycling borrow farmer

Second wallet, same usdc + dai cycling borrower

Third wallet, is a wstETH cycling borrower

defillama aggregated all euler sub accounts and filtered out cycling borrows

12

Hey @uhhhh – appreciate you reading through, and I will have those stats for usage shortly. We chatted with users, including Seraphim, after launching eGP1, and this is how he described how the eGP1 dashboard provided new insights for him:

“eGP1 flagged to me big borrows within the wstETH market. It also accurately showed that most wallets at risk within Euler are mining EUL using the mint function and therefore most wallets are actually not at risk of being liquidated due to price moves.” - Seraphim

13

Total upon completion 8 weeks of development: $20k.
Monthly costs after that for maintenance and App Upkeep costs $3,5k.

First 2 months are 20k plus 10 months * 3.5k = $55k for the first year, not $42k as previously stated.

Is this correct?

14

Did llamas provide a link? Would be curious to have a look.

15

10-month framing for recurring makes sense over 12 since we estimate initial development will take about 8 weeks.

Yes:
One-time after ~8 weeks = $20k
Recurring monthly after initial work is complete (approx. 8 weeks after snapshot) = $3.5k

16

This proposal is well-written and will be of great benefit to users of the protocol / the community.
Happy to support it!

2 Likes
17

Hi @uhhhh, thanks for your participation in the discussion.

As far as stats go the Defishy app has about 500 unique users, most of whom are Euler users because the page’s traffic picked up right when we completed eGP1. These users show a healthy amount of page interaction with around a 1 minute average engagement time per page load.

Another notable point besides Defishy’s usage is our team’s history of delivering, as on eGP1 we were able to fulfill all of our grant deliverables within the outset completion date. We have also done work with Aave grants, and have the same history of pushing our deliverables on time. Upholding this reputation of doing quality work on time is a high priority for us since we feel that finding reliable groups who deliver on what they promise is sometimes difficult in the DeFi space.

As far as your second comment about the DefiLlama liquidations dashboard goes, we feel that Defishy successfully fulfills a distinct niche. The DefiLlama dashboard is a good tool for getting an aggregated picture of the liquidation environment across many protocols on a chain. However, it lacks in its filtering abilities making it difficult to get a picture of what a singular protocol looks like.

We created Defishy with the vision of providing the user with a much more granular viewpoint. Compared to DefiLlama’s dashboard, Defishy’s is much better at getting a smaller-scope view of a specific protocol and even specific wallets within that protocol. With this grant, we hope to increase this granularity further by allowing users to filter wallets by borrow and collateral asset types, as well as give information on how much risk individual wallets are in through our VaR simulations.

Hope this helps with getting a clearer picture of Defishy’s vision and value to the community.

18

Thank you for your support! We agree that this project will be a great asset to the Euler community.

1 Like
19

2 things:
“500 unique users” - can i get a time series for this, is usage increasing, are we able to put a CAC on users.

2nd thing.
“a much more granular viewpoint”
“filter wallets by borrow and collateral asset types”
“how much risk individual wallets are in through our VaR simulations”

again, could you walk me through how the user journey would look here. I am not able to form an idea of how this is useful.

an example maybe.

20

Hey @uhhhh for sure:

For your first thing: Usage has remained steady since we shipped eGP1 with about 200 unique users per month. A CAC doesn’t totally apply here because the value we provide is not directly correlated to the number of users we have: it’s correlated to how much value each individual user derives from the platform.

For example, if you have a whale who sees an unfavorable trend in liquidation health scores over a certain time period and ends up saving hundreds of thousands of dollars, this platform was a pretty good investment for him. So even if it’s just one user, this platform will still provide more value than is in the asking price.

For your second thing—yes- here are some descriptions/examples for each of those three quotes:

granular viewpoint - Defishy allows all DeFi users to “get into the weeds” when analyzing borrowing and lending protocols. This data, like all on-chain data, can be collected (as we have), but it is not feasible by the average user. On top of this, most protocols will show this data only on a per-user basis to each individual user. Through Defishy, we made this data available to all, so all can get into the weeds when analyzing where to make investment, borrowing, and lending decisions.

filtering by asset types - Currently on Defishy, users can filter by Total supplied (low to high or vice versa), Total borrowed, and liquidation health. With this next grant, we want to expand that filtering to the ability to filter by asset types for borrowing and collateral. This was requested of us by members of the Euler community, which is why we think it will be beneficial.

Here is an example of what that would look like: I am an stETH holder. To determine if positions with large swaths of stETH deposited as collateral are near liquidation (which could cause a price cascade), I want to see what borrowers have put down stETH as collateral and if they have good liquidation health scores. Currently, there is no easy way for me to check this. With this grant, we want users to be able to filter by an asset type being used as collateral (like stETH) or borrowed, so they can make a judgment similar to the one in that example.

individual wallet risk though VaR - In the example above, the user was trying to determine if his stETH position was going to potentially see a large price drop. Through VaR, we can help the user map out what that price drop would be. This allows him/her to see specifically what the risk is if a two, three, four, five standard deviation event were to occur.

Thank you for questions @uhhhh Hopefully they are helpful to other members of the community as well. Let us know if we can clarify anything further or provide more examples, etc