A practical guide to manual comparison of multiple evidence records in TEMIRIN using configured records, explicit market identity, current controls, and human review.
August 3, 2026 · 5 min read
A comparison becomes useful only after the market is known. Open the configured watchlist entry, read its exact yes and no outcomes, closing condition, and resolution source, then state the factual question the evidence needs to inform. TEMIRIN creates opportunity records from watchlist markets rather than searching the full venue. This bounded start protects the research from collecting loosely related stories that share keywords but address different thresholds, dates, jurisdictions, or authorities.
Bring an evidence record into the market view only when it carries a supported explicit market identifier. Do not infer a relationship from a person, organization, or topic mentioned in the text. A record without the identifier can stay available in its source context until a user decides whether the configuration should change. Explicit linking may produce fewer rows, but every row has an inspectable reason for appearing beside that specific Polymarket question.
Build the comparison around fields available across source types: publisher or provider, canonical URL or reference, observed time, supplied publication or event time, source configuration, and safe ingestion status. Configured RSS and HTTPS web readers support canonicalization and content deduplication. Provider-configured Telegram and X readers may carry different identifiers. Normalize labels for review, but retain the original source type so a message, official document, and syndicated article do not become interchangeable rows.
Treat duplicate detection as housekeeping rather than corroboration analysis. Two canonical URLs can quote the same primary statement, while one living page can contain a material correction. Open the records and document the attribution chain. If the reader failed, show the operational gap; if the destination alert failed, consult the separate delivery ledger. Collection and transport statuses explain system behavior, not whether the underlying claim is correct or important.
For every linked record, paraphrase the exact claim, identify the evidence form, record its time window, and quote no more than needed in the private reviewer note. Add columns for relevance to the contract clause, preliminary or final status, and known correction. TEMIRIN can assemble deterministic briefs from stored fields. The reviewer owns every interpretation shown in the matrix and records it in plain attributable language.
Use descriptive relationships that expose the reasoning: independent primary confirmation, repetition of the same origin, later revision, different measurement, or unresolved contradiction. Avoid summing sources into a probability or confidence score. More rows do not necessarily mean better evidence, and a single named authority may carry the decisive publication under the market rules. If a conclusion depends on missing material, name that material explicitly and leave the row unresolved until a person checks it.
The opportunity record preserves the first observed trigger price while later synchronization supplies the current price. Show both with their observation times next to the evidence sequence. This arrangement helps a reviewer see what was known before and after repricing, but it does not establish causation. Public bid, ask, spread, and visible depth can reveal whether the displayed probability rests on a narrow or thin book. They still do not guarantee an executable price for any chosen size.
Public-wallet snapshots can add read-only exposure and concentration context when a wallet is configured. Keep the snapshot time visible and do not infer identity, intent, complete cost basis, or why an outside position changed. The evidence matrix, market observations, and wallet view answer different questions. Keep them separately labeled so the reviewer can explain which record informed each conclusion.
A reviewer can save a note, record an in-app approve or reject event, or separately create a paper order. Those are deliberate records with separate timestamps and actors. The configured price-move guard can block an authenticated decision after market movement. A user-entered paper order is checked for per-order and rolling twenty-four-hour notional, per-market exposure, and portfolio concentration. A pass records a permitted simulation under the current policy.
Alert rules may deliver review context through configured email, Telegram, Slack, or outbound webhook channels, with attempts visible in the ledger and manual retry available. The receiving channel cannot approve, reject, acknowledge, or create the paper record. Preserve the evidence comparison and later decision as separate audit entries, and export journal analytics when a team wants to examine its review process. Every real Polymarket action remains the user’s responsibility outside TEMIRIN.
For recurring team reviews, export a bounded set of record identifiers and verify a sample back in the workspace before drawing conclusions. Count items excluded because provenance, market identity, or timestamps were incomplete. Compare reviewers’ written classifications without converting disagreement into a generated truth score. If the process changes, document the new checklist prospectively and leave older matrices intact. This discipline lets the team improve how it compares evidence while preserving an honest view of the information and controls available for each historical market.
Before sharing the comparison, ask a second authorized reviewer to reproduce one conclusion from the linked records alone. Their task is to identify the same contract clause, canonical sources, relevant times, and unresolved assumptions without relying on an oral explanation. Differences become useful review notes. This small reproducibility check tests whether the matrix communicates real provenance and reasoning rather than simply reflecting the first reviewer’s memory.
The market must already be configured, and the stored evidence must carry a supported explicit market identifier. TEMIRIN does not infer the relationship from text similarity.
No. Deterministic briefs organize stored fields. A human reviewer records how claims relate and may leave the relationship unresolved.
Not by default. Canonicalization and deduplication help, but reviewers should inspect attribution and identify shared primary material.
No. Users enter paper orders themselves, and current checks only pass or block the simulation against configured limits and exposure.
