A practical guide to manual category review of portfolio and paper records in TEMIRIN using configured records, explicit market identity, current controls, and human review.
August 3, 2026 · 4 min read
Category PnL attribution begins by separating record types. Public-wallet positions come from read-only external data and saved snapshots. Paper orders are user-entered simulations with their own requested price, notional, time, and later mark. Trigger-price references belong to watchlist-derived research records and are not positions at all. Combining these three sources in one profit column creates a precise-looking number that cannot be reconciled.
Decide whether the analysis covers public-wallet realized values, unrealized marks, paper mark-to-market changes, or decision context. State the valuation timestamp, provider coverage, and cost-basis limitations. TEMIRIN cannot infer external hedges, transfers, other wallets, or why a venue position exists. When public data is incomplete, keep the affected value incomplete rather than moving it to a convenient category or assuming a zero basis.
A category should come from the stored market or workspace record. Use a controlled label set with clear definitions, and preserve an unknown or uncategorized state. If a market appears under the wrong label, correct the configuration prospectively with an actor and timestamp. Similar market titles should remain separate until a reviewer verifies their exact contract wording and deliberately records the grouping used for analysis.
Decide how multi-theme markets will be handled before calculating totals. A simple primary category is often more auditable than fractional allocations invented after seeing results. Keep the explicit market ID as the join key, and confirm that closed, renamed, or duplicated display titles do not split or combine rows unexpectedly. Category is an analytical dimension and policy field; it is not a proof that outcomes are independent.
For a public wallet, distinguish closed-position values from open-position marks and show the snapshot time. Verify whether the available provider fields support realized and unrealized separation before publishing it. Transfers and missing cost basis can make a computed return misleading. Reconcile a sample against the public venue view and retain a plain-language reason for values that cannot be reproduced from the available record.
For paper orders, calculate later mark-to-market change from the saved simulated entry and current mark under a stated convention. Keep it labeled paper-only and separate from the watchlist trigger-price comparison. The trigger may have preceded the user's paper entry and is not a fill. TEMIRIN does not submit orders, reserve capital, or create realized venue returns, so category summaries must retain that simulation boundary.
Journal records can explain which evidence, price context, policy result, and reviewer note accompanied a paper order or no-trade decision. Link them only through explicit record identifiers. A favorable category total does not prove that its sources caused the outcome, that the same process will repeat, or that every reviewed record became a position. Include rejected and no-trade counts so activity and selectivity are visible beside simulated results.
Compare categories over a fixed period with counts, exposure, incomplete rows, and concentration. One large market can dominate a small category, while several markets tied to the same event can create hidden correlation. Reviewers should inspect these relationships manually and record any dependency used in the analysis. Keep observed historical totals separate from the user-entered notionals and policy results attached to later paper orders.
A useful category export includes record type, market ID, outcome, category, source ledger, entry or observation time, valuation time, notional where applicable, status, PnL field definition, and missing-data reason. Include the public address reference only when needed and authorized. Never export connector secrets, unsupported wallet credentials, or a blended total that cannot be traced back to its component rows.
Re-run the same report with a fixed cutoff and confirm that historical paper entries and decision notes have not changed. A later public-wallet snapshot may legitimately update the current view, so retain its new timestamp instead of replacing the older observation. Good attribution is a transparent accounting of supported records under declared rules. It preserves what the workspace observed without converting it into advice or a claim that TEMIRIN controlled any real venue transaction.
Before sharing the report, select one category and rebuild its total from the exported rows. Confirm that category labels, ledger types, signs, marks, and excluded incomplete values match the written methodology. This reconciliation catches duplicated market IDs and accidental wallet-paper blending more reliably than a polished chart. Save the check date and analyst note so a later reader knows which snapshot was verified.
They should be reported separately because one comes from read-only external position data and the other from simulated workspace records with different valuation and coverage rules.
Use the explicit stored category, preserve unknown states, and record deliberate category corrections prospectively with an actor and timestamp.
No. It is the first stored observation for a watchlist-derived record. A paper order has a separate user-entered request and simulated entry context.
