How we read

Every prediction is a question with a fork, a horizon date, and three timelines. Timeline A and Timeline B are the two futures the system is choosing between. The quiet path is the one nobody is watching — the peripheral timeline that quietly becomes real when both of the loud sides fail.

A prediction never lives alone. It sits inside a long cycle: the pattern that says what the system can and cannot deliver.

Every live read carries two numbers. Crowd pull is how hard markets, flows, and attention are reaching for a timeline right now. Reality support is what rules, budgets, and physics actually allow — the model’s structural read.

The gap is the difference between the two. When the gap is wide, the crowd is reaching for something the system can’t deliver — or it’s under-reaching for something the system can.

A mover is a specific actor, signal, policy, or event that can bend the odds. Each prediction lists them ranked by leverage: who the model thinks has the most power to push the fork one way or the other.

A pressure moment is a specific threshold whose crossing bends which timeline we end up on. Each moment carries a level (how close), a window (when), and a short list of signals we watch for its approach.

We write a read in five parts: the fork, the crowd pull, the reality support, the movers, and the quiet path. The fork names both timelines in three words or fewer each. The crowd number comes from markets and flows when they exist; the reality number comes from the model.

A prediction resolves at its horizon date, as a hit (A landed), a miss (B landed), or a split (the outcome fell between the two sides we drew). Resolved predictions stay up — the track record shows every one.

A prediction isn’t a bet, a signal, or a courtroom. We don’t sell positions. We don’t edit resolved calls. We don’t pretend the model sees everything — the quiet path is the model’s own admission that we can be wrong from an angle we aren’t looking.