Content authored by practitioners with experience at Amazon India, Intuit QuickBooks, and the Tata Group. Meet the team →
A fabricated bank statement with plausible-looking individual amounts passes visual inspection because reviewers check individual transactions, not the statistical properties of the full distribution. Fabricated amounts constructed by humans deviate from the digit distributions that genuine financial data produces.
Run two complementary forensic checks on the full set of transaction amounts: (1) Compare the distribution of leading digits against the expected frequency distribution for naturally occurring financial data — significant over-representation of any digit range is a fabrication signal. (2) Analyse the transaction amount sequence for unnatural rhythms, fixed-step progressions, or suspicious clustering that would not arise in independent real-world spending events.
Apply to accounts with 50 or more transactions for reliable statistical power. Calibrate for account type — salary accounts, fixed-EMI-heavy accounts, and ATM-dominated accounts require adjusted baselines before digit distribution comparison is meaningful.
Two flags: a digit-distribution anomaly flag (clean / flagged / insufficient data) and a sequence pattern flag (organic / patterned / review), both presented in the fraud signals section of the analysis report with the specific deviation metrics that triggered the classification.
A fraudster constructing transaction amounts for a fabricated bank statement faces a problem they may not be aware of: human-generated numbers do not look like financial data. The amounts a person invents when trying to appear realistic — scattered values, varied amounts, nothing too round, nothing too uniform — produce statistical patterns that genuine spending does not. Digit-pattern analysis is the forensic technique that makes this visible, and it is one of the oldest tools in the forensic accounting toolkit precisely because it catches what looks fine to the human eye.
What Digit-Pattern Analysis Is
Digit-pattern analysis examines the distribution of digits in a set of financial amounts to identify deviations from the patterns that genuine data produces. In large sets of naturally occurring numbers, the leading digit is not uniformly distributed — lower digits appear as the first digit disproportionately often compared to higher digits. This asymmetry appears consistently in data generated by independent real-world processes: salaries, invoices, utility bills, market prices.
Genuine bank statements from active accounts with diverse transaction types tend to follow this distribution because the amounts are determined by external events the account holder does not fully control. Fabricated statements — where a person manually constructs amounts intended to look realistic — tend to produce flatter distributions, because humans do not intuitively approximate logarithmic digit frequencies.
How It Works — What Is Checked
Digit Distribution Check
The first check compares the distribution of leading digits across all transaction amounts against the expected frequency pattern for financial data. A statement where the digit 5 appears as the leading digit at twice the expected rate, or where round thousands appear at a rate inconsistent with organic spending, registers a deviation.
The output is not a verdict. It is a signal indicating whether the digit distribution looks consistent with genuine financial activity or shows the flatter, more artificial pattern associated with constructed amounts. The flag prompts a human reviewer to investigate — it does not automatically reject the application.
Sequence Pattern Analysis
The second check examines the sequence of transaction amounts for rhythms that real spending does not produce. Genuine transactions are independent events determined by external factors: a grocery bill is whatever was purchased; an electricity bill is whatever was consumed. Fabricated sequences often show subtle structure — amounts that increment by a fixed value, clusters of the same amount appearing at suspiciously even intervals, or variety that looks engineered rather than organic.
This is distinct from the digit distribution check. A fabricated statement could pass digit distribution analysis if the fraudster generated amounts with statistical care, but still fail sequence analysis if the amounts show unnatural temporal patterns. The two checks are complementary.
Forensic Signal Reference Table
| Signal Type | What It Looks For | What It Does Not Catch |
|---|---|---|
| Digit distribution analysis | Over- or under-representation of leading digits vs expected frequency | Fabrications carefully constructed with correct digit distributions |
| Sequence pattern analysis | Arithmetic progressions, fixed-step increments, unnatural amount clustering | Fabrications that mimic genuine transactional randomness effectively |
| Round-number clustering | Excess concentration of round thousands and lakhs | Legitimate ATM-heavy accounts (calibration required) |
| Salary consistency check | Fixed-amount monthly credits that appear and disappear | Fabricated salaries set at the correct level and consistent across months |
India-Specific Context
Indian MSME credit is the segment where fabricated bank statement detection matters most. A significant share of India’s approximately 63 million registered MSMEs do not maintain audited financial statements. For these applicants, the bank statement is the primary income and financial health document. NBFCs processing high volumes of MSME loan applications — 100 to 500 files per month is common for mid-size lenders — cannot manually apply forensic checks to every file.
Microfinance applicants present a different pattern. Statements from microfinance accounts often have very low transaction counts and amounts clustered in small ranges, which limits the statistical power of digit distribution analysis. Sequence pattern analysis and balance chain verification are more diagnostic for thin statements.
Forensic CAs applying the Institute of Chartered Accountants of India’s guidance on fraud risk assessment routinely use digit-distribution checks as a first-pass filter when reviewing large document sets — precisely because the check is scalable and does not require expert judgment on each individual amount.
The bank statement analysis platform runs both digit distribution and sequence pattern checks automatically on every uploaded statement, presenting the output as reviewable signals rather than binary verdicts — consistent with the principle that these checks identify what warrants review, not what is proven fraudulent.
Combining digit-pattern flags with balance chain verification and metadata inspection produces a layered forensic view. The bank statement fraud detection capability applies all three checks in parallel, so credit teams get a consolidated fraud signal summary rather than isolated outputs from individual tests.