As I specified in the first post, I would not consider seasonal algorithms as the main profit type, but a simple support.
I understand that having strategies that operate every day is much more rewarding, but to have a complete algo portfolio, in my opinion, we also need something based on the long term to stabilize profits (also something like pathfinders would be very good, never tested but i saw nice results)
To overcome instability, the idea could be to develop seasonal algorithms on several highly uncorrelated assets (probably commodities that are slightly better as you said) to minimize the risk.
.
.
