Ticker data is expensive! Many of these models have their results from live data, but some are just backtests. Relatively inexpensive datasets for backtesting at the ticker level start at 2008 (unfortunately, obviously, given that considering a flashpoint like that is important for rigorously confirming a model works).

As soon as feasible, I want to purchase the larger datasets that often go as far back as 1985 for a more comprehensive picture.

It would be nice. My first impression was that the selected models seems very risky and would realy depend on the last decade which was probably the best in 70 years. I would like to see how they would have performed during a crisis (and bonus point for using a log scale on long duration to avoid minimizing the past)

Related to that, do you have any recommendations on good quality ticker data sources you'd buy?