Why these pairs

One statistical test can be fooled by a lucky sample. Every pair must pass a battery: the same relationship tested in both directions, cross-checked by an independent method, before it goes anywhere near the list.

Two assets can move together and still make a terrible pair. The spread itself must show a real habit of snapping back when stretched — and at a speed that's actually tradeable.

Nobody sits choosing entry and exit levels. Each day the system fits the spread's mean-reversion process and solves for the optimal entry, exit, and stop — then freezes them for 24 hours.

Plans are fitted at 03:30 UTC and never repainted. What you see at breakfast is what was tradeable at breakfast — the live z you watch ticks against that morning's frozen recipe, not a moving target.