The fact that this is still on testnet makes it more interesting to me, not less. If 1, 3, and 5 minute prediction cycles are already running smoothly in a live environment, that says the architecture is doing what it is supposed to do. On Linera, each market runs independently