It's interesting to see how the SPY ETF, which represents a significant portion of the US stock market, can be used to measure market trends and volatility. It would be even more useful to have a longer database available for analysis, perhaps going back to 2000, to include multiple market crashes for comparison. I am from Finland and am more insightful about that market. In fact, comparing the performance of Helsinki stock exchange the to that of the US market could provide useful insights and potentially even opportunities. It's also worth noting that Helsinki's stock market, while smaller in size, has seen its own share of ups and downs over the years and could potentially be analyzed in a similar manner. Have you had a chance to follow any other country indices?

I practiced cardiology for forty years and would expect failure of medical devices to be a memory-less process with numbers of events over fixed intervals to have exponential tails. If you can show that this is not true with a quantile plot of your tail data versus the ExponentialDistribution[1], and if there is a good linear relationship in the quantile plot the Log[tail data] to the ExponentialDistribution[1], that would be evidence for power tail scaling which might indicate a fractal process.

Yes, this is minute data, which comes from the Schawb StreetSmart trading application. It allows download of six days of data, which I have been doing every Saturday for the past 14 months, and updating the data base. The data seems to be of very good quality. Unfortunately I didn't start before the COVID crash. If you have academic connections you may know somebody who has access to WRDS. Unfortunately they won't share with mere mortals. The problem with the minute data is that trading causes high volumes in the morning and at the close, so sampling over represents the lightest trading with lowest volatility. I have some transaction by transaction data which I plan to post in a few days, when I get it ready, which will show a power tail exponent of about 3.0 which is more in line with the final daily output of the SPY. The tick data are very difficult to work with because most of the sequential prices show no change, nevertheless it is the tick data that we really need in the Wolfram Data Repository for any serious research..