There seems to be a gap between the “Random Start Years, Random Restart” and “Random Years, Non Chronological” methodologies that would seemingly be a low effort value add.

The “Random Years, Non-Chronological” option adds variety in the conditions tested, but loses the value of following historical sequences that every other historical methodology mode offers. Generally there is value in following historical trends in how one year might flow into the next. However, to follow that through all available history (or just more recent history with the Random Start Years varieties) limits the combinations that will come up and overweights recent performance in the results (since every trial will include all recent years).

I would propose to add an option in at least both of the “Random Restart” modes to specify a % chance of an early restart (before hitting the end of the available chronological sequence). Then each year, the engine would test against that percentage and either continue its chronological progression as it does today, or trigger a restart as it does when hitting the last year today. This would add greatly to the variety of conditions that would be tested, while still following historical trends of multi year returns while avoiding overweighting of recent performance.