-
statistics of real world data, based on model's smart contracts predictions
video - https://www.loom.com/share/e2b85c2ec31e41aaa16da7cd7a75e09d?sid=c32bea7c-2840-4f79-af9d-e159f027e7db
you can come to see a live demo
Inspiration
one of our inspirations for this idea is morpho. which is a peer to peer lending protocol. We mostly got inspired by Xinyuan Sun.
What it does
Our team has engineered a unique peer-to-peer lending protocol that utilizes AI to generate personalized smart contracts to enhance trade outcomes for lenders and borrowers. This direct connection between lender and borrower eliminates inefficiencies typically caused by an imbalance between the number of suppliers and borrowers found in traditional pools, thereby improving overall capital efficiency.
The issue with liquidity pools is the socialization of yields, causing numerous lenders to divide the interest paid by an insufficient pool of borrowers. This results in lowered earnings for suppliers and heightened payments by borrowers to compensate for under-utilization, thus culminating in a significant spread.
To address this, we utilize a dynamic algorithm that effectively locates relevant lenders and borrowers, optimizing this search process. We achieve this by analyzing on-chain blocks and modeling user behavior.
Our innovative approach has fostered the creation of a matching engine wherein traders can pinpoint possible deals devoid of intermediary interference. Thanks to our intelligent algorithm, we can shape a smart contract that ensures effective, seamless peer-to-peer trading.
How we built it
the project is divided to 3: 1.model - for the model we used an RNN base model using huggingface and pytorch.The model predicts the action each user will do in the next step, our model gets an F1-score of 85% 2.peer to peer matching algorithm - Drawing on the predictions of the model, we engineered an algorithm capable of dynamically matching lenders and borrowers in a peer-to-peer setting.The algorithm is purpose-built to optimize the outcomes for all traders involved. 3.generate smart contract -The output of the peer-to-peer algorithm is used to generate a unique smart contract for every individual transaction.
Challenges we ran into
The process of discerning the crucial inputs for predicting the user-optimized parameters of the smart contract presented significant challenges. This is vital for crafting the precise smart contract that encapsulates the user's intentions and furnishes them with a mirrored digital contract representing their needs.
Accomplishments that we're proud of
We take pride in the achievement of constructing a dynamic algorithm in the brisk span of just one night. Our journey was marked with numerous impediments that, at times, dampened our spirits, making success seem like a distant dream. However, our unfaltering dedication enabled us to overcome these obstacles. Furthermore, our accurate market predictions, boasting an impressive F1 score of 85%, only heighten our sense of accomplishment and fuel our pursuit of excellence.
What's next for NeuroContract-
Looking ahead, we aim to implement our agile smart contract generator to diversify the contract types created from the same insights. We are broadening our insight to create contracts that go beyond the standard lending and borrowing protocols. We are looking forward to leveraging Artificial Intelligence for a diversified protocol design generation like dex. This initiative forms part of our progressive journey towards the incremental differentiation of responsibilities and role allocation within the blockchain industry.
Log in or sign up for Devpost to join the conversation.