Define a Swing Hypothesis Before Calling It an Edge
A recognisable pattern is a candidate for testing, not evidence of profitability.
Causal Definition
Write the observed pattern and threshold using data available at the decision time. Declare the candidate universe and rejection conditions before reviewing the sample.
Validation Boundary
A hypothesis earns support only under a stated evaluation method and implementation. Keep explored variants and untouched periods distinct. A textbook name does not create an independently verified edge.
Primary Sources
A testable entry objective before economic scoring
A defined edge needs an explicit data boundary, eligible population and causal claim, not only a chart that later reverted. The authorised entry process trains and ranks on the raw causal boundary-recovery objective. Entry identity is frozen before the selected exit is applied. A swing holding description does not identify that entry objective or independently validate it. The new cash scenarios deliberately make no market hypothesis: they reconcile declared units, price, cash and costs. Their exact results do not become entry candidates, measured model accuracy or economic observations. A more detailed publication can explain the required evidence without pretending that a simple accounting exercise has discovered an edge.
A fictional calendar leak
Suppose an invented candidate excludes earnings using a calendar downloaded after every outcome. It may have a clear written rule and a favourable historical curve while still lacking the calendar state available at each original decision. The missing object is point-in-time event knowledge and its revisions. Adding costs to that curve cannot repair a noncausal entry input. Preserve the experiment version and acquisition times, then use the authorised research process to evaluate the actual instrument if that task is requested. This is a conceptual failure case, not a performed financial backtest, measured vendor defect or proof that a named model fails independent validation.
A negative case stays in the selection history
A real candidate ledger should retain proposals, rejected variants, data exclusions and the predeclared acceptance contract. A complete software test is different from an economic holdout. The seven cash cases include a zero-price scenario and neutral action examples so the arithmetic's branches remain visible; they are not a candidate tournament or a validation distribution. A later exit-adjusted winner share cannot silently become the criterion that selected the original entry. A confidence grade, source-model headline or founder credential needs its own population and cannot replace an identified causal objective. Keep the missing research artifacts specific rather than converting every unresolved question into a generic score.
Acceptance belongs to the real route
A verified independent master requires its actual asset routing, data, periods, warmup, policies and separate checks. This website adds none of those actual sources and makes no promotion claim. No production registry, deployed SHA or subscriber route follows from source code or a research LIVE label. A complete tutorial can be accepted for educational clarity and software correctness while actual economic validation remains absent. Preserve the exact unresolved scope in a review and obtain authentic records before strengthening the claim. The original defined-edge text remains alongside these narrower cases, with no fabricated auditor, user population or source entitlement.