OK, so I posted the above, which was based on my overall sense of how Bitcoin behaves when zoomed in to the month/week/day level. But then I thought, is this really true? Is Bitcoin's monthly/weekly/daily performance really random? Well, I'm currently on a long holiday leave, so I have plenty of time on my hands, so why not analyze some real historical price data to test things out?

So I fired up MATLAB and downloaded Bitstamp's BTC/USD historical price data for day open/close prices, from 2016 up to now (3721 days). I then plotted each day's price performance in terms of daily loss (in red) and daily gain (in green). Each dot is one day, and the horizontal axis spans 2 months per row. Note that there may be a shift by a couple of dots per row, as I'm assuming a 366-day year. Here's what came out:



...which confirms and experimentally proves that zooming in to the day level results in a random loss/gain pattern. I postulate that the same holds true to the weekly and monthly levels, hence my original post.

And then I thought: instead of showing loss in red and gain in green, as a 0/1 state, it would be more interesting to show the actual day close - open differences in USD. This should clearly show the 4-year cycles and how the daily fluctuations have played out over the years. And that's what I did. I used a heat map, starting with black (for maximum loss) up to white (for maximum gain), and shades of orange (Bitcoin's trademark color) in between. Sure enough, the 4-year cycles appeared in full glory. Here's the resulting heat map:



I hope you like the results, and I hope they help you change your mind if you were thinking of selling your corn!