Xerberus: Risk Ratings for all Fungible Assets on Cardano

[GENERAL] Name and surname of main applicant

[GENERAL] Email address of main applicant

Additional applicants

Xerberus Labs Ltd.

Noah Detwiler

Matthias Guilbert

[GENERAL] Please specify how many months you expect your project to last (from 2-12 months)

[GENERAL] Please indicate if your proposal has been auto-translated into English from another language.

[GENERAL] Does your project have any dependencies on other organizations, technical or otherwise?

[GENERAL] If YES, please describe what the dependency is and why you believe it is essential for your project’s delivery. If NO, please write “No dependencies.” .

Xerberus depends on external investors who invest in us as referenced in the budget (RESOURCES & VALUE FOR MONEY). The investor is a direct equity investor, and the agreements are signed under the condition that Xerberus is accepted in the Seed Enterprise Investment Scheme (SEIS) of the UK governance. The SEIS scheme has a 72% acceptance ratio, and we are working with a specialist consultant to ensure our acceptance.

[GENERAL] Will your project’s output/s be fully open source?

[GENERAL] If NO, please describe which outputs are not going to be open source. If YES, please write “Project will be fully open source.”

Project will be fully open source.

[METADATA] Category of Proposal

[IMPACT] Please describe your proposed solution.

Summary: We create Risk Ratings for all fungible Assets on Cardano. We base our risk ratings on objective facts, the transactions on the blockchain belonging to a project. These ratings will be available for all community members for free to reduce the amount of ADA lost to poor projects while increasing the amount of ADA available for good projects.

• How we perceive the problem we are solving: The decentralization inherent to blockchains enables wide participation, but it also allows bad actors to join our networks, seeking to exploit our community. The risk associated with bad actors is a major problem in crypto, deterring users and investors and slowing down the mass adoption of these systems. We address this problem by providing access to quantitative metrics of the risk associated with the many Cardano assets in the market. This will keep our community safe from scams and make truly informed investment decisions easier for everyone. 
• This is achieved by providing two critical services. First, Xerberus allows users to explore and compare assets using a risk ranking system that categorizes assets from AAA to D, where D indicates a high probability that the token will default to a price of zero. Through this risk ranking system, users have an immediate, independent, and impartial assessment that they can use to make informed decisions. The second service we provide is an alert system that warns users when the risk profile of a token changes, providing them with a window of opportunity to protect their investments or spot promising opportunities early on as the associated risk increases or decreases, respectively. The risk scores and their changes are triggered by the only objective facts that can be established about all tokens: the transaction patterns connected to them on the blockchain. Our system scans all transactions of a token to capture the underlying pattern, recording, for instance, the type of wallets holding the token and other activities they engage in. When the transaction patterns look unhealthy, i.e., similar to those from known failed projects or even scams, the risk score assigned to the token is adjusted.
• Our reasons for approaching it in the way that we have: Our models are based on all available transactions because they are objective facts that contain relevant information. For instance, a token primarily owned by wallets that tend to buy and hold tokens long-term is less volatile than tokens primarily owned by wallets that tend to hold tokens for only a few weeks. Another example is that a certain group of wallets buying a token could be a good sign, as the holdings in these wallets correlate with successful projects, adding confidence to the likelihood of success. Additionally, certain transaction patterns might be associated with scams because a group of bad actors can be exploiting a particular vulnerability or executing the same strategy repeatedly. These transaction patterns will be recognized by our system and correlated with a high default risk. The application of transaction pattern information in these examples is intuitive. However, transaction patterns can offer much more information. Free markets tend to find fair prices because all available information about an asset is "priced in." Knowledgeable traders can often spot even slightly mispriced assets and open positions to benefit from the spread between the current and the actual perceived valuation. Assuming that these actions reflect traders' beliefs in asset performance, then all aggregated transactions belonging to an asset should contain all the available information about that asset, and we use sophisticated mathematical models to extract this information. You can read about our mathematical models in the white paper (https://docsend.com/view/kw5uizzscsh36sqj).
• Who our project will engage: We will make the risk ranking freely available to everyone, making it easy for the broad community to enjoy the positive impact of our risk scores. Additionally, we create a free API for other builders and power users so they can integrate our scores into their products, such as oracles, portfolio builders, or trading bots. 
• How we will demonstrate or prove our impact: If the people who use Xerberus ranking stay safer and invest better than those who do not use our rankings, our impact is proven. We further describe how to prove our impact in detail in the KPI section. In summary, we will publish key performance indicators that show if the number of ADA invested in rug pulls is decreasing. Having the number of lost ADA decrease and the ADA spent on good projects increase will create a compounding positive interest that all Cardano Community Members will benefit from.

[IMPACT] How does your proposed solution address the challenge and what benefits will this bring to the Cardano ecosystem?

The challenge of Products & Integration aims to fund projects that drive user adoption. One of the biggest challenges we face in mass adoption is that many inexperienced users are reluctant to engage with ecosystem services due to the steep learning curve and the predatory actions of bad actors. Scams are undoubtedly one of the largest drivers of user churn (users leaving the ecosystem), as it is naturally disappointing to save ADA and then lose it to one of many rug pulls. 

Cardano already enjoys the reputation of being the blockchain without hacks. If we now reduce rug pulls and increase support for quality projects, Cardano's position among other blockchains will be further strengthened by the safety its system provides to users. The security of users is an ultimate driver of both retail and institutional adoption, and Xerberus aims to provide exactly this value proposition. 

Security for users and businesses adopting Cardano is critical because our ultimate aim is to provide a decentralized alternative to the centralized banking system currently in power. If we cannot convince the wider public that our systems are more secure from a user perspective than traditional banking, we will fall short of achieving our goals. Therefore, creating a safe user experience will have a tremendous impact on Cardano's broad adoption. 

The goal of this funding proposal is to make Xerberus the "go-to" platform to check the legitimacy of projects. This will allow users an easy and swift understanding of asset risk, thereby helping users make rational, data-driven, and informed decisions, rather than simply acting on their gut feeling or the (questionable?) advice of influencers. This independent and impartial assessment will empower users and help prevent loss of capital through impulsive decisions, such as FOMO. Furthermore, it will also help reduce end users' regrets and increase the efficiency of their investment funds by directing them to safer projects, building trust in the ecosystem, and creating happy and loyal users.

[IMPACT] How do you intend to measure the success of your project?

The successful accomplishment of Xerberus' mission will be quantifiable by the reduction in user funds lost due sto high-risk assets across the entire Cardano ecosystem. We monitor this progress through a key performance indicator (KPI) known as "Accumulated Rugged ADA," with the aim of consistently diminishing its growth rate. The measurable impact of mitigating user fund losses extends beyond numerical value, providing two significant benefits. Firstly, it enhances user safety within the Cardano ecosystem by equipping them with tools to avoid capital loss in high-risk assets. Secondly, it amplifies capital investment in high-quality projects, thereby elevating the overall quality of the ecosystem. While quantifying the effects on the ecosystem presents challenges, the association between Xerberus and these positive downstream effects is robust and can be statistically substantiated. 

At the project level, we assess the trust placed in Xerberus through user surveys and engagement with our product. To gauge the product's success, we employ standard KPIs such as the number of unique monthly users and the frequency of opened email alerts. In addition, we utilize specific KPIs to evaluate the efficacy of our model. This includes the "Error Alert'' metric, which calculates the ratio of false alerts to the total number of alerts issued, enabling us to refine our system's accuracy. Furthermore, we track an inverse KPI called "Alert Missed," which measures the ratio of issued alerts to the number of events that should have triggered an alert upon review. These metrics serve to optimize the sensitivity of our system within the most effective range.

[IMPACT] Please describe your plans to share the outputs and results of your project?

Xerberus, as a project, is diligently working on the forefront of mathematics. Six months ago, we reached a milestone in our research where we surpassed all previously published results in the field we are working on, combining Algebraic Topology, Graph Theory, and Machine Learning.

During the development, we will share knowledge and publish papers: The primary objective of this Catalyst proposal is to secure funding to expand our Research and Development activities and acquire computational power that aligns with the project's requirements, enabling us to reliably transition our model to production. With this progress, we intend to share our findings through papers that we plan to submit to academic journals within the next 12 months, acknowledging the funding provided by Project Catalyst.

After the development, we will share the product: The risk model will be offered to all Cardano users at no cost, providing Risk Scores for all Native Assets. These scores will be easily accessible to end users through a simple interface and to developers via an API that can be freely utilized to incorporate Xerberus' Scores into their own products. In the future, we envision making our risk scores available on the blockchain itself, utilizing a proof system. However, this is beyond the scope of the current proposal.

After the development, we will share our code: Our entire research is based on public blockchain data and mathematical principles that are potentially accessible to anyone with sufficient knowledge and motivation to explore them. As a result, we will open-source the entire codebase developed for our research and model implementation. The models we create serve as foundational components for decentralized risk rating organizations. Ultimately, our organization strives to be verifiable by everyone, which is yet another reason for us to embrace open-source practices. Data scientists and builders across the entire Cardano Ecosystem will benefit from our published codebase.

[CAPABILITY/ FEASIBILITY] What is your capability to deliver your project with high levels of trust and accountability?

A thorough assessment of Xerberus' accountability to the Cardano community must encompass both our technical capabilities as a team and, more importantly, our trustworthiness.

Our technical competence is demonstrated by the fully functional application we have developed while actively engaging with the Cardano community. The "Legacy Xerberus App'' features advanced functionalities, including an advanced wallet and token explorer, as well as a bottom-up wallet labeling system based on statistical properties. While this application initially served as a research tool and a means to build a community, it clearly showcases our commitment and ability to deliver products to end users.

Access the Legacy Xerberus App'' through the following link: https://legacy.xerberus.io/ using the provided credentials:

Email: demo@investors.com

Password: investors@123

Alternatively, visit our documentation: (https://documentation.xerberus.io/legacy-app/wallet-explorer)

The core team behind Xerberus comprises highly accomplished and competent individuals with diverse backgrounds in venture building, mathematics, software development, and more. Their collective knowledge and skills form the foundation of Xerberus' capabilities. Furthermore, our project benefits from the guidance and support of a strong advisory board consisting of experts in quantitative trading and institutional finance, who provide valuable insights, expertise, and an extensive network. Additionally, we have a dedicated group of community supporters who excel in building large databases and developing machine learning models. The combined expertise of our team and their support network positions Xerberus for success in achieving its goals.

The trustworthiness of Xerberus as a company is demonstrated by our formalization as a Limited Liability Company incorporated in London, United Kingdom, under the registration number 14966821 at Companies House (Xerberus Labs LTD). As a registered company in a reputable jurisdiction, we are obligated to publish our financial statements, and the embezzlement of company funds for private expenses is considered a criminal act. We will publish audited financial statements, conducted by an external party, to provide transparency and accountability to the community members who mentor our project's progress. This deliberate setup reflects our commitment to compliance and regulation, positioning us as part of a new wave of crypto companies dedicated to bridging the gap between crypto and traditional finance for widespread adoption. Our commitment to transparency and accountability extends not only to ourselves but also to others within the industry.

Xerberus also has a track record within the Cardano community. We have actively attended several Cardano events and represented Cardano at various conferences throughout the year, including Web3 Berlin and Crypto Cruise Cologne, where we delivered keynote speeches. We are proud to have earned the trust of key members of the broader community. For instance, we serve as official advisors to the Orcfax Oracle Project. Over the past year, we have been actively involved in the community, assisting members by answering questions and providing high-quality information, such as the report on the Token Dumping accusation addressed to the SundaeSwap Team. You can access the report through the following link: (https://docsend.com/view/ntwt6mx75bhfwgdp)

[CAPABILITY/ FEASIBILITY] What are the main goals for the project and how will you validate if your approach is feasible?

The main goal of the proposal is to create an application that produces a risk score for 500 Cardano tokens, updating this score hourly as new transactions are added. Additionally, this application will enable all users to receive a risk score for any submitted policy on demand. These features will be accessible through a website and an API.

We are confident in the feasibility of the risk score system, as there exists a wealth of research in fields such as medicine where similar methodologies have been successfully employed, for instance, as a tool to diagnose cancer.

Below are a few papers by teams from three prestigious institutions that have used the same mathematical approach to address similar issues:

1. Santa Fe Institute: Discrimination of Economic Input-Output Networks Using Persistent Homology (https://www.researchgate.net/publication/305903531_Discrimination_of_economic_Input-Output_networks_using_Persistent_Homology)

1. Department of Mathematics and Statistics from University of Central Oklahoma, among others (https://www.mdpi.com/2297-8747/25/3/58)

1. Cornell University: Persistent Homology of Complex Networks for Dynamic State Detection (https://arxiv.org/abs/1904.07403)

1. Cornell University: Temporal Graph Networks for Deep Learning on Dynamic Graphs (https://arxiv.org/abs/2006.10637)

Beyond academia, we have learned from our advisory board and other knowledgeable contacts that hedge funds also use similar strategies to identify anomalies within markets.

Lastly, and most importantly, we have already developed a working proof of concept. In it, we analyzed a substantial dataset of transactions in the Cardano blockchain using our mathematical models and machine learning modules. While the core application's development is at an advanced stage, it needs to be fine-tuned and paired with a corresponding infrastructure capable of providing a continuous monitoring system.

Please email us to gain access to the Proof of Concept at sp@xerberus.io.

[CAPABILITY/ FEASIBILITY] Please provide a detailed breakdown of your project’s milestones and each of the main tasks or activities to reach the milestone plus the expected timeline for the delivery.

The 12-month implementation period is segmented into four primary phases, each possessing its unique goal and a verifiable milestone to gauge our progress:

1 Extended Exploration Phase (4 months)

Description: We plan to dedicate an additional four months to test different iterations of our Proof of Concept (PoC) and compare results. Our aim is to ascertain which model delivers the highest predictive value for risk score evaluation.

Verifiable Milestone: A functional model accompanied by an in-depth performance review in the form of a comprehensive lab report.

2 Production Phase (4 months)

Description: Upon identifying the best-performing model from the extended exploration phase, we plan to transition into the production phase. Here, our model will be deployed in a scalable server environment to perpetually monitor the selected 500 native Cardano tokens.

Verifiable Milestone: A fully operational product displaying hourly updated risk ratings for the 500 tokens.

3 User Iteration Phase (3 months)

Description: The third phase intends to evaluate user experience using a randomized test group for feedback. We aim to collect data on product usability, additional parameters users would like to see in the interface, and their perception of the model's effectiveness. This valuable feedback will be consolidated and utilized to refine the model and product interface.

Verifiable Milestone: An upgraded version of the working product, taking user feedback into account.

4 Public Access (1 month)

Description: In this final phase, our goal is to offer unrestricted access to the public, making our early warning system accessible to all. We will also provide a functional API to allow other developers to integrate our system into their own products.

Verifiable Milestone: A publicly accessible, fully functional product, along with a working API for integration.

[CAPABILITY/ FEASIBILITY] Please describe the deliverables, outputs and intended outcomes of each milestone.

Lab Reports: Every month, we will publish a detailed lab report summarizing our experiments and findings. These reports will itemize all the computational experiments conducted and compare the alterations made to our model against the established baseline set by the Proof of Concept (PoC) model. These lab reports will culminate in a final report demonstrating the gradual improvement from the baseline model to the model destined for production. Xerberus' commitment to providing fair and neutral risk ratings for all assets must be underpinned by a robust scientific foundation. To ensure the community distances itself from any emotional reactions that may arise from Xerberus Risk Ratings of Cardano native assets, and fully understands and trusts its results, we will persist in publishing monthly lab reports even beyond the Extended Exploration Phase. Although these reports are not critical for Project Catalyst, we believe they are essential in building trust in our project through transparent and evidence-based changes in the risk ratings.

Working Product with 500 Assets: As we finalize the specific model to be used in the production phase, we will transition from monthly lab reports to monthly development reports. These will encapsulate a thorough summary of our progress in front-end and back-end development. At present, we have a simple interface for the existing PoC, but it only displays basic metric outputs from the risk model. To construct the back-end infrastructure needed to continuously monitor 500 tokens and update risk scores every hour, we will need to establish multiple parallel running servers. Regarding the front-end, we aim to enhance the existing interface to allow users to explore the reasons behind risk score changes. We will implement additional features based on their feasibility within our current infrastructure and model.

Improved Product: As we expand our small cohort of test users from tens to hundreds, we will discontinue monthly development reports. This interactive phase is critical, as we depend on user feedback to fine-tune both the risk model and the user interface, with the aim to evolve the initial product into an application that is user-friendly, intuitive, and easy to navigate.

Public Product and API: In the final month, we will dedicate time to refine the interface before launching and publishing the working API along with its documentation.

Public Code Repository: Along with the final product, we will clean our code and write documentation to enable any other data scientist on Cardano to benefit from our work.

[RESOURCES & VALUE FOR MONEY] Please provide a detailed budget breakdown of the proposed work and resources.

The funds requested for our proposal are substantial. This level of funding is necessary to expand the research and development activities of the current team, support future hires, and acquire the computational resources that align with Xerberus' needs. It's important to note that Xerberus follows a funding strategy that combines private investor funding with a Catalyst grant, and our private funding surpasses the amount requested from Project Catalyst.

Here is a comprehensive summary of our total cost planning: Over the 12-month period, our total projected cost will be approximately $250,000 USD. We will raise $150,000 from private investors and combine this with another $100,000 from Project Catalyst in the form of 400,000 ADA. Out of this amount, 84% ($212,100) will be allocated to Research and Development, while the remaining 16% ($34,800) will be directed towards General Administration.

We will spend the $100,000 we expect to receive from Project Catalyst to aid our Research and Development expenses of $212,100:

• $138,000 will be dedicated to the salaries of our three full-time R&D employees, including our Chief Science Officer, backend developer, and computer scientist responsible for optimizing the computational efficiency of the models.
• $72,000 will be exclusively allocated to acquiring the computational resources required to run our risk models, which includes numerous experimental runs during the research phase.
• $2,100 remaining has been spent on miscellaneous expenses such as subscriptions for tools.

Funds received from Catalyst ($100k) will go solely towards R&D expenses. Our particular focus will be on fulfilling our computational needs, amounting to $72,000, while the remaining $28,000 will be allocated to human resources. Due to the significant amount we are requesting from the Cardano Community, we believe that total transparency regarding our funding is appropriate. Below you will find a complete budget for Xerberus Ltd. for a 12-month period, starting in October of this year. You can access the budget through this link: https://docs.google.com/spreadsheets/d/1nrxwwxsEWKDnaQeT9CZFvCtuDJdTbXCBH5sG_QS5I2M/edit?usp=sharing

What happens if ADA falls too much in price? This proposal assumes ADA being valued at 25 cents USD. As of mid-June 2023, ADA is comfortably above our assumed price. However, if we fail to approach the target of 100k USD and our position savings are insufficient to bridge the gap, we will need to expand our equity fundraising efforts. Our current investors are willing to invest 150K under the UK Seed Investment Program, which offers additional securities to the investor. This particular program can be extended to cover up to 250k. While a program extension serves as our backup plan, it comes at a significant opportunity cost in the form of lost equity.

[RESOURCES & VALUE FOR MONEY] How does the cost of the project represent value for money for the Cardano ecosystem?

• Value for Cardano: 400k ADA is indeed a substantial amount to request as a grant, but it is fully justified given that a system like Xerberus can serve as a cornerstone of Cardano's value proposition, positioning it as the "Apple of Blockchains" — a reliable and secure platform. Having a risk model in place for all fungible tokens on Cardano will enable everyone in the community to make better investment decisions. All of these good investment decisions together will compound and pay an ever-increasing interest back to the Cardano Community in the form of excellent projects and fewer rug pulls. 
• Cost of the Project: Over half of the funds invested in this infrastructure come from private financing, which means that the Cardano ecosystem gains $1 dollar of value for less than 50 cents spent. Regarding our expenses, we have allocated $70k specifically for computational power. Each server we plan to utilize costs between $1.8k and $3.8k per month. Our intention is to operate 2-4 such servers continuously for 12 months, possibly scaling down towards the end of the year. Lastly, considering the geographical distribution of our team across London, Berlin, and Jerusalem, the remaining $30k allocated for salaries is quite reasonable.

[IMPORTANT NOTE] The Applicant agreed to Fund10 rules and also that data in the Submission Form and other data provided by the project team during the course of the project will be publicly available.