Noticed relative volume isn't something that most crypto APIs offer, so I made a relative volume analysis in a python notebook. The thing with crypto is that it's pretty insular. You buy/sell crypto using other crypto most of the time, ETH and BTC being the main exceptions.
But what if you could tell where volume is leaving, and where abnormal relative volume is appearing?
top is volume, bottom is relative volume. This uses (currently) the last 400 days of data, training on 95% and testing on the last 5%. Relative volume is a good way to compare volumes between other securities with ratios.
For reference, relative volume = (Current volume / Average volume). Now, you'll notice that DOGE had a relative volume several hundred % higher than bitcoin. Eventually I'd like to get some models training on this to notice minutiae I might be missing and track potential breakouts.
Fun thought experiment, good resume project I suppose