Summary: Financial spreading turns a borrower’s statements into the lender’s standard template so that every credit decision uses the same ratios calculated the same way. It sounds mechanical, and the mechanical part is exactly where it breaks: layouts change, line items appear, and a template built for last year’s statement silently misreads this year’s. This post explains the work, the failure modes, and what examiners expect, and links to how MightyBot automates it.
What a spread is
A spread is the borrower’s income statement, balance sheet and cash flow restated in the lender’s chart of accounts, usually for three or more periods side by side. From it the analyst calculates leverage, coverage, liquidity and profitability ratios, trends across periods, and global cash flow when the borrower and guarantors are analyzed together.
The inputs are messier than the output. Audited statements, reviewed and compiled statements, management accounts, interim statements, and business and personal tax returns with their schedules all show up, from many accountants, in many layouts. The SBA’s SOP 50 10 8 bases repayment analysis for existing businesses “on the three most recent years of historical financial information” plus an interim statement, and requires lenders to “obtain tax return transcripts and reconcile the Applicant’s financial data against the tax transcripts.”
How the manual version works
The analyst opens each statement, finds each line, decides what it means in the lender’s chart of accounts, and keys it in. Adjustments follow the firm’s conventions: which add-backs are allowed, how owner compensation is treated, how a partial year is annualized, how trailing twelve months is built from interim and annual figures. Then the ratios run.
The judgment in that work is real. So is the transcription, and transcription is where the hours go.
Why templates break
Most spreading tools automate the transcription by mapping a position or a label on the page to a line in the template. That works until the borrower switches accountants, adds a revenue line, sends management accounts in place of audited statements, or files a tax return where last year there was a compiled statement.
Each change forces a remap, and the failure is often silent. A roll-forward keeps running with a new line left out of operating expenses, and the DSCR that reaches committee is wrong by that amount. Nobody notices until a reviewer rebuilds the spread, or an examiner does.
The fix is to read the statement the way an analyst does: work out what each line means in context, place it by meaning, and show a person anything that does not fit.
What examiners expect
The OCC’s Rating Credit Risk handbook is blunt: “There is no substitute for rigorous analysis of a borrower’s financial statements.” It says analysis of revenues, margins, cash flow, leverage, liquidity and capitalization “should be sufficiently detailed to identify trends and anomalies that may affect borrower performance.”
Timeliness counts. The FDIC’s examination manual says “financial information should be updated not less than annually.” The Federal Reserve’s Commercial Bank Examination Manual lists “automated financial statement spreads of borrowers” among the information that “should be readily available and routinely reviewed by management.”
How the automated version works
On MightyBot, agents classify each document first, extract every line with its period and entity, and canonicalize it to the lender’s chart of accounts. Each cell keeps a pointer to the page and position it came from. The firm’s conventions are written as plain-English policies. A line the agent has not seen before is mapped by meaning and flagged for review, so new items surface in front of an analyst rather than disappearing into a formula.
Because the output is canonical data, the same spread feeds ratio calculations, covenant tests and credit policy without re-extraction. The checklist of what to ask a vendor is on the financial spreading page, and a comparison of eight tools is in best financial spreading software.