Reproducible verdicts
Three fixed proposals. One returns APPROVE, one REVIEW, one BLOCK. Run them yourself and watch deterministic policy code decide, not a model.
LAST UPDATED POLICY KYVRANE-V101 / THE THREE VERDICTS
Each scenario is a fixed assessment input. The verdict beside it is not a label chosen by hand: it is what the policy engine returns, and the test suite asserts it on every commit. Change one number in the proposal and the verdict changes with it.
A 2.40% expected return, priced against real SoDEX order-book depth. Slippage, fees, and market impact are subtracted, and what survives still clears the 1.00% policy minimum. The trade is authorized.
The same 2.40% expectation, five times the size. The order eats deeper into the book so it pays visibly more slippage than scenario 01, and at a $3,000 reference it is $15,000 of notional, past the $10,000 manual-review threshold. Policy gates it for a human rather than rejecting it.
A 0.20% expected return looks positive until execution is priced in. Once order-book slippage, fees, and market impact are subtracted the net executable edge goes negative. The trade loses money before it is placed, and capital is blocked.
02 / WHAT THE HASH PROVES
Every receipt carries a SHA-256 hash over its canonical content, including the moment it was issued. That seal reveals any later change to the receipt. Because the issue time is inside the seal, running the same scenario twice produces two different hashes, and that is correct: they are two distinct receipts of two distinct decisions taken at two distinct times.
The reproducibility guarantee lives in the verdict, not the hash. The same input always yields the same APPROVE, REVIEW, or BLOCK with the same reason codes. The hash proves the receipt of that decision has not been tampered with since it was issued. It does not prove a market outcome and it does not guarantee profit.
04 / YOUR TURN
The workspace takes any pair, side, quantity, and expected return, prices execution against it, and returns a receipt on the same deterministic path these three scenarios use.