Vol Analyzer

This project allows a user to view the effects of a company's past social media price in the context of their historical token price.

Using Uniswap as a by the minute price oracle, a user is able to see how much impact a social media post had on token price in spans of 10 minutes, 1 hour, 1 day and 1 week after post was posted. A user is also able to view the top most impactful posts across all posts. "Impactful" is defined as %change in price, with price measured in Eth to filter out BTC/ETH beta.

Each post is also tagged using AI, so a user can potentially see what types of posts generate the most price volatility. The available tags are "Misc", "Partnerships", "In Person Events", "Hack/Exploits", "New Features", "Initial Hack/Exploit Announcement", and "Tokenomics".

This project was built on a react frontend with the data sources being sourced from the uniswap v3 subgraph, twitter, discord, and further processing done with gpt-3's davinci-003 LLM model.

Uniswap was instrumental as it was the only feasible way to obtain a by the minute accurate historical price feed for almost any token. Specifically, swap data from liquidity pools was used to calculate prices in USD and ETH.

TheGraph was used as a tool to obtain data from uniswap as it was already neatly indexed with USD prices already included in the schema. There were issues encountered trying to programatically download over ten thousand swap objects so to work around this, we wrote a script to generate incremental curl requests to retrieve the swaps 1000 at a time.

Non partner technologies used included gpt3, react, ruby, python, html/css.