AI-NFT prediction market

ML software to predict future NFT prices based on attribute data scraped from the top tokens on the market.

Comment

Inspiration

  • We took inspiration from existing marketplaces, such as Angur and Polymarket, to build an easy-to-use predictive NFT software. Even in today's crypto-heavy environment, there isn't a marketplace that provides a quick generalization of specific attributes (background, colors, characteristics, etc), which hinders consumer use and overall interest. Our project aimed to tackle this issue by providing a simple generalizing application for top NFTs.

What it does

  • Our project generalizes the different attributes of NFTs to simplify the way our predictive software displays future pricing. Given an NFT, you can get an idea about what attributes play a role in the pricing of that token, making it easy to understand and get involved in the up-and-coming market.

How we built it

  • Frontend:
    • The frontend takes the data processed by the backend, establishing necessary connections between files to organize and relate the information. Subsequently, it displays this data in a table format for user-friendly visualization.
  • Backend:
    • Python scraper to get data from CryptoSlam
    • Once data was processed, we chose the top 100 Token IDs and their attributes
    • We then processed these attributes for each token within different collections to determine future pricing

Challenges we ran into

  • Figuring out how to obtain data from Cryptoslam (data scraping)
  • Ensuring that we use Starknet to complement the entire project, not just one part of it

Accomplishments that we're proud of

  • We spent a lot of time building up the ML logic and there was a lot of data to sort through, which required us to minimize our scope to make the project feasible

What we learned

  • Getting data dependent on attributes from each NFT (CollectionID) was a challenge because our scope was too large, so we had to make changes
  • To obtain the full breadth of data possible for the software, we would need significantly more resources (memory and money) than we have

What's next for the AI-NFT prediction market

  • There is room for enhancement to encompass more than just the top 100 NFT collections

Built With

Share this project:

Updates