I scraped early Solana token lifecycles into a structured dataset (140 charts)

I've traded, well, gambled, Solana memecoins for almost 3 years now, and I've began to realize the amount of factors at play in determining if a coin is worth buying. I've mostly dabbled in low market cap coins while keeping the vast majority of my crypto assets in high market cap coins, Bitcoin for example. After watching so many new coins with great narratives go straight to 0, I decided to start approaching this emotional game logically.

After a while searching, I couldn't find a dataset that provides the non-obvious features I was seeking. I ended up building a web scraper that detects new Solana coins, capturing snapshots every ~10 seconds, while simultaneously querying API data for socials, rugcheck data, token metadata, and a bunch of additional information. With this ingested data, I built a clean dataset for analyzing this large number of new features the scraper had extracted.

Each token snapshot includes tons of features such as:

- market cap - volume - holders - top 10 holder % - bot holding estimates - dev wallet behavior - social links - website analysis (title, HTML, text snippets, reputation, etc.) - rugcheck scores + risk - and plenty of other tokenomic-based fields

In total, I scraped thousands of early token charts, and picked out 140+ clean charts, each with nearly 300 datapoints on average.

Even with just a quick exploratory analysis, I started noticing small patterns, such as the correlation between the presence of social links and market cap ATH. I'm a data engineer, not a data scientist (yet), and I'm positive those with stronger ML backgrounds could find much deeper patterns and predictive signals than I can.

For the full dataset description/structure/schema, the Hugging Face Dataset Card can be found in the attached post URL.

I'm more than happy to answer any project-related questions about the scraper, the data ingested, or really anything else :)