Costs 6 min read Updated Aug 26, 2026

Reconcile prop-firm bank charges without double-counting

Suppose your own cost log shows $378 of firm charges for the month and the bank-linked view identifies $248. Neither total is wrong. The identified number counts only settled, same-currency charges whose bank descriptor names a covered firm, so it is a floor by construction. Reconciliation is explaining the $130 gap row by row, not forcing the totals to agree.

In this guide

The short answer

Start here

Keep Manual Bank, the cost log you type yourself, as the record of what the attempt has cost, and treat identified bank charges as a separate evidence floor. Compare the two totals, then resolve each differing row against the bank statement and the firm's own purchase history: add a missing charge once, delete a duplicate, or mark the row explained by a feed limit. Never add the linked total to the manual total, and never treat the gap itself as proven missing spend.

What matters

  • Log every firm charge in Manual Bank exactly once, then use the identified bank total only as the floor that log has to clear.
  • Change the manual log only after the bank statement row and the firm's purchase history agree on the charge.
  • Keep the identified total gross: record a refund as its own dated credit only after tying it to the original charge.

1

Four filters decide what the identified total counts

The linked lens, the bank-connected view of your spending, admits a charge into its identified total only when the row passes four checks: it has settled, meaning cleared and no longer pending; it is an outflow, money leaving the account; it carries your exact base currency, because no row is ever converted; and its bank descriptor contains an unambiguous token for a firm NANO covers. A row that fails any one check stays out of the total even when it is real spend.

That construction makes the identified number a floor. A charge older than the history the sync requests (it asks each institution for up to 365 days) never enters the total, and neither does one paid from a card you never linked or one billed under a descriptor the matcher cannot pin to exactly one firm. The matcher refuses to guess because a wrong attribution corrupts your per-firm costs, while a missing one just leaves the floor lower. What an identified charge does prove is narrow: the linked account paid that merchant that amount on that date. It does not say which evaluation, reset or funded account the payment bought. Your own record supplies that link.

2

Work the $130 difference one row at a time

Put the two totals side by side: $378 in Manual Bank against $248 identified across two settled charges. The $130 gap has a short list of possible causes, and each calls for a different fix: a real charge the feed cannot see (an unlinked card, a date older than the 365-day history request, an unmatched descriptor), a manual row entered twice, a manual amount typed wrong, or a charge paid in another currency that the lens surfaced and excluded.

Resolve each candidate against two documents: the bank statement itself and the firm's purchase history in your account portal. A manual row with no matching statement line on any of your accounts is the row to challenge first. Make one correction per confirmed finding (add the missing charge once with its statement date, delete the duplicate, or mark the row explained by a feed limit) and stop when every dollar of the $130 carries one of those labels.

3

Keep the two totals separate, and refunds out of both

The expensive wrong move is adding the $248 identified to the $378 manual and treating $626 as spend. The two baskets overlap on every charge you logged correctly, so the sum double-counts most rows. The linked view is a cross-check against the manual record: when a confirmed charge is missing from Manual Bank it enters once, and when it is already represented the manual total does not move.

Refunds follow the same one-entry discipline. The identified total is gross, computed before refunds, because the lens drops every incoming credit by design, so a $100 refund never shrinks the $248. Record the credit as its own dated manual entry only after you can tie it to the original charge. A credit you cannot tie to a charge is another reconciliation item; silently netting it into either total hides it.

Worked example

Assumptions, not a forecast

A $130 gap and what it proves

  • Hypothetical month: Manual Bank holds $378 of firm costs, an evaluation purchase, a reset and a data fee.
  • The linked lens identifies two settled, same-currency charges totaling $248.
  • No receipt, statement row or feed-limit note has yet explained the $378 − $248 = $130 difference.

Exactly two claims are defensible: $378 is logged with supporting records, and at least $248 of firm spend is bank-evidenced. The $130 is a work queue of unmatched rows and nothing more until each row is resolved against the statement and the firm's purchase history.

Matching totals would not prove completeness either: two charges missing from both records, or a refund offsetting a typo, can make two wrong totals agree.

Tool tutorial

Run the reconciliation

10, 15 minutes per statement month
  1. 1

    Bring Manual Bank current: every purchase, reset, subscription and fee you can support with a receipt or statement row. This total is the number the reconciliation defends.

  2. 2

    Read the linked lens header before touching rows: sync date, institution status and any excluded-currency note. A stale sync or a missing institution already explains part of any gap.

  3. 3

    Compute the difference, here $378 − $248 = $130, then list the manual rows without a matching bank charge and the bank charges without a manual row.

  4. 4

    Give each unmatched row one of three outcomes: add a missing manual entry once with its statement date, delete a duplicate, or mark it explained by a feed limit. Stop when every dollar of the difference carries a label; an unlabeled dollar means the reconciliation is not finished.

Bank spend lens Pro tool

Use bank data as a cross-check, not a replacement

The Pro bank-assisted lens can surface identified charges from supported U.S. institutions while keeping them outside Manual Bank totals.

Pro feature. Institution support, access and sync state are shown at the destination. Manual Bank remains the operating record.

Evidence boundary

What NANO can and cannot know

NANO cannot guarantee full transaction history, identify every merchant, assign a bank charge to a specific account or evaluation, net refunds into a verified spend result, or convert currencies; rows in another currency are surfaced and excluded from the total without conversion. A failed or stale bank read leaves the identified total incomplete; never read it as zero spend.

Common questions

Why is my linked bank spend lower than what I logged?
Because the identified total counts only settled, same-currency charges with a firm-named descriptor from institutions you actually linked, pending rows, other currencies, unlinked cards and unmatched merchants all stay out. Treat the gap as a review list, and check the sync date first: a stale read explains a low number before any row does.
Should a refund reduce the identified spend total?
No. The lens keeps the identified total gross and drops incoming credits by design, because silently netting a refund hides it instead of explaining it. Record the refund as its own dated credit in Manual Bank once you have tied it to the original charge.
The feed missed a charge I know I paid, what do I do?
Add it to Manual Bank yourself, once, with the statement date and amount. The feed misses charges older than the 365 days of history it requests, charges paid from a card you never linked, and charges billed under a descriptor the matcher cannot pin to one firm, the manual log exists precisely because the feed is a floor.