Models

Every model we run, and what its record actually is

0 models are named in this product — written a signal, been ruled on, or both. 0 of them have settled results to measure. Not one has a yield that can be told apart from zero — the test is a clustered z of ±2 over settled fixtures, the same one /performance withholds the headline yield on. 0 publish today. Every figure below is measured from settled results on each load; the ruling quoted beneath it is prose, and where the two disagree the card says so.

The standard for promoting a model is paper_trade_gate(): 150 settled fixtures, no-vig closing-line value above 0.5%, and a CLV z above 2, with the scoring anchor required to be independent of the selection anchor. It has never been applied to anything — the paper-trade ledger it reads is empty — so no architecture below has passed it, including the two that publish.