They ran a pretty solid analysis of the subject. Not only do they look at correlation, but they regress out the trend line and show a correlation in the noise. Correlation of noise typically does imply a causal connection. They also look at correlation between GPUs that are used for mining and ones that are not and show that the correlation only holds for GPUs used in mining, so the case seems pretty tight. If you are just trying to poke holes in the analysis I would focus more on the possibility of there being other contributing factors that are more significant (i.g. supply chain issues). In any case the author was considerate enough to include the dataset and the python notebook so you can run subsequent analyses on your own. Some quick things worth trying: 1. )Weight both the price of Cryptocurrencies and the price of GPUs for Market Size. 2.) Add in other explanatory variables to try to find the relative contributions of different sources.