A result image is presentation. Evidence is the chain behind it.
Trading-signal result evidence should connect the original signal timestamp to an admitted entry, lifecycle, terminal state, real market path, selected result endpoint, and final artifact without silently repairing missing facts.
This is the distinction Curren uses when reasoning about public result artifacts. A chart may look internally consistent while still hiding hindsight, ambiguous event ordering, substituted candles, or a percentage calculated from a different point than the one marked visually.
The safest architecture therefore treats the renderer as the final communication layer. It receives authority that has already survived an evidence gate; it does not create authority by drawing a convincing chart.
What a screenshot cannot prove on its own.
A screenshot freezes pixels, not provenance. Without the underlying signal contract and market evidence, a viewer cannot reliably tell whether the displayed entry existed when claimed, whether the terminal state was finalized, or whether the result was selected using information that only became available later.
| Visible claim | Evidence question |
|---|---|
| Signal entry | Was this the persisted entry authority for the original signal? |
| Market path | Are the displayed bars real observations within the admitted causal window? |
| Result marker | Does its price and time resolve to the same endpoint as the headline move? |
| Result metric | Was it calculated from the same proven endpoint represented on the chart? |
| Artifact | Can the image be tied back to the contract and evidence that produced it? |
If those questions cannot be answered, the screenshot may still be useful as an illustration. It should not be promoted to authenticated result evidence merely because the image is polished.
The evidence frame must respect the signal-time cutoff.
Point-in-time reasoning separates what was knowable at the signal timestamp from what happened afterward. Historical context can extend backward, but future market information must not leak into the original decision frame simply because the later outcome is now known.
Rule: establish the causal cutoff before evaluating the later result. The outcome can be measured afterward; it must not rewrite what the original signal appeared to know.
This matters for research as much as rendering. Once hindsight is allowed to alter the feature context, visible candle selection, or decision explanation, it becomes difficult to distinguish predictive information from a retrospective narrative.
OHLC bars do not always prove which intrabar event happened first.
A one-minute candle can report its open, high, low, and close without revealing the exact sequence in which the high and low were reached. That creates a real evidence problem when a single bar appears to contain both a favorable extreme and a return to a boundary such as break-even.
If the public market bar cannot prove whether the favorable extreme happened before or after the boundary touch, choosing the more flattering interpretation would be unsupported. Curren's reviewed result methodology resolves this by excluding an ambiguous boundary-touch candle from the favorable-endpoint selection when ordering cannot be established from public OHLC evidence.
The larger principle is more general: absence of ordering evidence is not permission to assume favorable ordering.
A favorable endpoint needs a rule for when the measurement window closes.
An open-ended maximum can keep improving long after the trade logic has stopped being meaningfully connected to the original signal. A fixed arbitrary time cap has the opposite problem: it can cut off a still-active runner simply because a clock expired.
Curren's approved X-result path uses a market-state boundary instead. After the result has reached the required runner anchor, fully closed market bars can continue to improve the favorable endpoint until the path first returns to the persisted entry price. That break-even return closes the measurement window.
For a long result, the favorable endpoint is the highest admitted high before the break-even return. For a short result, it is the lowest admitted low before that return. The current open candle is not endpoint authority.
This makes the endpoint rule inspectable: it is neither an unlimited hindsight maximum nor an arbitrary number of hours chosen after seeing the chart.
Readability can adapt without inventing market history.
A result path may be too sparse on the original trigger timeframe to communicate the entry-to-endpoint path clearly. The display timeframe can therefore become finer when needed for readability, provided the bars remain real market observations and the causal and endpoint anchors remain intact.
This is an important distinction between adapting the view and rewriting the evidence. The renderer can change density, crop, labels, and visual hierarchy. It should not smooth, interpolate, or synthesize candles simply to create a prettier trajectory.
The endpoint should survive all the way to the pixels.
Endpoint consistency is easy to lose when multiple systems compute different pieces of a result. A chart marker might use one price while the percentage, R multiple, timestamp, or annotation uses another. The image then tells several subtly different stories at once.
Curren's result contract instead binds the endpoint badge, connector, marker, wick relationship, move percentage, R multiple, endpoint price, and endpoint time to the same proven event. Render metadata and hashes preserve the connection between the evidence, contract, and artifact.
Provenance does not prove that a trading idea was good. It proves something narrower and valuable: the public result can be traced back to the evidence that was admitted to support it.
The website preview is deliberately not a performance record.
The interactive result card on the Curren landing page uses synthetic values to demonstrate interface behavior and the visual grammar of the approved result layout. It does not claim that those example moves occurred, and it does not perform production evidence admission in the browser.
Authenticated result authority remains server-side. That separation lets the public website explain the evidence model without turning illustrative UI into an unsupported market-performance claim.