Disclosure: This is my own dataset. Access is gated.<\/strong><\/p>\n Hey everyone,<\/p>\n I’ve been working on a dataset since September, and finally published it on Hugging Face<\/a>.<\/p>\n I’ve traded (well.. gambled<\/em>) with Solana memecoins for almost 3 years now, and discovered an incredible amount of factors at play when trying to determine if a coin was worth buying.<\/p>\n I’d dabble mostly in low market cap coins, while keeping the vast majority of my crypto assets in mid-high cap coins, Bitcoin for example. It was upsetting seeing new narratives with high price potential go straight to 0, and finally decided to start approaching this emotional game logically.<\/p>\n I ended up building a web scraper to both constantly scrape new coin data as they were deployed, and make API calls to a coin’s social data, rugcheck data, and tons of other tokenomics at the same time.<\/p>\n The dataset includes large amount of features per token snapshot (every max 10 second pulse), such as:<\/p>\n In total I collected thousands<\/em> of coin’s chart histories, and filtered this number down to 140+<\/strong> clean charts, each with nearly 300<\/strong> data points on average.<\/p>\n With some quick exploratory analysis, I was able to spot smaller patterns, such as how the presence of social links could correlate with a higher market cap ATH. I’m a data engineer, not a data scientist yet, I’m sure those with formal ML backgrounds could find much deeper patterns and predictive signals from this dataset than I can.<\/p>\n For the full dataset description\/structure\/charts\/and examples, see the Hugging Face Dataset Card<\/a>.<\/p>\n<\/div>\n submitted by \/u\/wtfmase <\/a> \n
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
