Crypto parameter trials need a complete independent master
Retain failed variants and source limits; calculator tests, candidate screens and adapters are not full strategy completion.
Every inspected variant belongs in the ledger
A strategy selected after comparing many histories carries a selection decision. Preserve variants, inspected periods, data revisions, exclusions and reasons for rejection. Do not report only the final favourable chart. The desk does not invent a search count or empirical correction for unknown trials. Its twelve added investigations are different educational questions, not twelve model experiments or independent samples. An authentic crypto study requires eligible licensed data, correct asset identity and a declared causal objective before its first economic result can be interpreted.
Entry and exit research remain separate
Rank entry models on raw causal boundary recovery and freeze their identity before applying one predeclared exit. Later return, profit factor, win rate or holding drift may assess that frozen contract; they cannot choose its entry ranking. Do not reopen an exit tournament after looking at validation or forward outcomes. Native minute execution, whole-position ownership, maximum holding and established best hypothetical fill window remain their own rules. The settlement lens implements none of them. It is a full-position authored arithmetic contract, not an entry selector, native exit system or instrument adapter.
A missing asset integration is actual work
If the asset route is not supported, complete and certify the required integration rather than silently applying equity conventions to crypto. Reuse compatible verified artifacts without erasing identity guards, resource limits, periods or warmup. Component tests and candidate screening preserve useful progress but do not complete the canonical research factory or independent master. A verified no-model master can be a completed finding; untested candidates or missing data are not. This publication performed no such asset task and therefore makes no new independent financial conclusion.
Separate software evidence from financial evidence
Python Decimal and decimal.js agree on the invented settlement equations, while browser checks exercise controls, downloads and invalid inputs. That establishes software consistency within the declared contract. It supplies no authentic source observations, calibrated probabilities, economic validation samples or subscriber routes. Production promotion is separately authorised and cloud-only, with durable evidence. Neither code presence nor a research LIVE label proves deployed state. Keep those evidence classes visible when summarising results. The fictional trial register below illustrates transparent incompletion without claiming that a real crypto market failed the complete system.
Fictional crypto evidence case
An invented register has variant A inspected, variant B screened and variant C missing its eligible data. No independently verified master exists. The result is incomplete research, not a three-variant profitable model or a completed no-model market. The register contains no real strategy, financial period, returns or production promotion.
| Object | Record | Bounded finding |
|---|---|---|
| A | Inspected fiction | Not independent master |
| B | Component screen | Not full completion |
| C | Data absent | Not verified no-model |

Primary references and scope
Source review: 14 October 2026. Official mechanism descriptions are not this desk's paid-account test, independently captured fills or vendor-return audit.