The Macro Twin
We built a digital twin of the world economy on Cognitive Fab's Polytrion platform. It models how rates, currencies, commodities, credit, and growth affect one another. Every week we run popular analysts' calls through it and see which ones hold up. We also run studies of the questions driving markets, and track which macro regime the world economy most resembles.
Under active developmentThis week's video
Model wiring
Resident analysts
Each week we score the latest calls from the analysts we follow against the twin. The bar shows how many of their claims the twin agrees with, half agrees with, or disagrees with, and how many it can't test yet.
The analysts against the model
Guest analysts of the week
Past guests : all guest analysts →
Studies
Each study takes one question about the world economy and runs it through the twin, or tracks one thing the model can't see yet. The newest are below.
AI revenue
What people and companies actually pay for AI products each year, counted once and set against the cost of the compute already running. The twin has no revenue of its own, so this is the reality check on the revenue paths in our studies. Refreshed weekly.
Macro regime
Model calibration
Weekly evaluation method
A market call usually rests on a chain of cause and effect, such as "the Fed cuts, the dollar weakens, gold rallies." The twin lets us check whether that chain holds, starting from the current state of the economy.
Collect
We collect the week's calls from the analysts we follow: rates, the dollar, gold, oil, equities, and recession odds.
Translate
We turn each claim into a scenario for the twin: the drivers the analyst assumes, and the outcome they expect over their stated horizon.
Simulate
The twin starts from this week's market closes and runs each scenario hundreds of times, with and without the shock, on the same random draws. The difference is the effect of the shock alone.
Score
Each claim agrees, half agrees, disagrees, or is not testable yet, with the numbers behind it.