MCom Forensic Accounting & Investigation

Faculty of Commerce, Law and Management · CDU-BUS-FOR-900 · MM · v1.0.0

The first-pass analytics of a fraud examination: first-digit conformity to Benford’s law with Nigrini’s MAD bands, the Beneish M-score for earnings manipulation, the net worth method of proving unexplained income, and the number-duplication test on payments.

Lineage

Conquest Designer University, 2026 — authored from the brief and the ACFE body of knowledge; no prior institution is claimed

Final research project

Screening before suspicion: digit, ratio, net-worth and duplication tests as a reproducible first pass on a ledger

Syllabus

Digital analysis: Benford’s first-digit law, mean absolute deviation, chi-square, conformity bands
Financial statement fraud: the eight Beneish indices and the M-score
Indirect methods of proof: net worth, expenditures, unexplained funds
Payment analytics: the number-duplication test, repeated amounts and their rows

Reference

Benford: P(first digit d) = log10(1 + 1/d)
MAD (first digit): mean |observed − expected| over d = 1..9; ≤ 0.006 close, ≤ 0.012 acceptable, ≤ 0.015 marginal, above: nonconformity (Nigrini 2012)
Beneish M: −4.84 + 0.920·DSRI + 0.528·GMI + 0.404·AQI + 0.892·SGI + 0.115·DEPI − 0.172·SGAI + 4.679·TATA − 0.327·LVGI; > −1.78 flags a likely manipulator
Net worth method: unexplained = (closing − opening net worth) + living expenses − known income

Units

First-digit test (Benford) · Beneish M-score · Net worth method: unexplained funds · Number-duplication test

Role when narrated: You are a certified fraud examiner reporting a screening result to counsel. Say what the test measures, show the numbers, and state plainly that a flag directs further work and proves nothing on its own.

First-digit test (Benford) · benford_first_digit

Try with the example

Do the leading digits of these amounts conform to Benford’s law, and how far do they deviate?

Reference: Nigrini (2012) Benford’s Law, ch. 5 (first-digit test, MAD conformity bands 0.006 / 0.012 / 0.015); amounts below 10 excluded

Returns

counted: Amounts tested (≥ 10)
digitCounts: Observed count of each first digit 1..9
mad: Mean absolute deviation from Benford
chiSquare: Σ (O − E)² ÷ E over the nine digits (8 df)
conformity: close, acceptable, marginal or nonconformity

The example

{
  "amounts": [
    1120,
    1380,
    1460,
    1975,
    1042,
    1210,
    1833,
    1598,
    1099,
    1305,
    1712,
    2150,
    2890,
    2431,
    2077,
    2666,
    3120,
    3490,
    3005,
    4210,
    4890,
    5120,
    5775,
    6340,
    7010,
    8450,
    9320,
    1188,
    1523,
    2260,
    3675,
    1407,
    2555,
    1066,
    4402,
    1960,
    2780,
    6150,
    1254,
    1839
  ]
}

Agents: POST /faculty/mcom-forensic-accounting--investigation/benford_first_digit with JSON {pan, …inputs} and the x-402-payment header. Manifest.

Beneish M-score · beneish_m_score

Try with the example

Do the eight Beneish indices put this company above the threshold for a likely earnings manipulator?

Reference: Beneish (1999) Financial Analysts Journal 55(5) 24–36: the eight-variable probit model and the −1.78 threshold

Returns

mScore: The M-score
likelyManipulator: M > −1.78

The example

{
  "dsri": 1.465,
  "gmi": 1.193,
  "aqi": 1.254,
  "sgi": 1.607,
  "depi": 1.077,
  "sgai": 1.041,
  "tata": 0.031,
  "lvgi": 1.111
}

Agents: POST /faculty/mcom-forensic-accounting--investigation/beneish_m_score with JSON {pan, …inputs} and the x-402-payment header. Manifest.

Net worth method: unexplained funds · net_worth_method

Try with the example

How much of the increase in a subject’s net worth, plus what they spent, is not explained by known income?

Reference: IRS Internal Revenue Manual 9.5.9 (net worth method of proof); ACFE Fraud Examiners Manual, Financial Transactions — indirect methods of proving income

Returns

openingNetWorth: Opening assets − liabilities
closingNetWorth: Closing assets − liabilities
netWorthIncrease: Closing − opening net worth
fundsApplied: Increase + living expenses
unexplainedFunds: Funds applied − known income

The example

{
  "openingAssets": 500000,
  "openingLiabilities": 200000,
  "closingAssets": 900000,
  "closingLiabilities": 250000,
  "livingExpenses": 120000,
  "knownIncome": 260000
}

Agents: POST /faculty/mcom-forensic-accounting--investigation/net_worth_method with JSON {pan, …inputs} and the x-402-payment header. Manifest.

Number-duplication test · duplicate_amounts

Try with the example

Which payment amounts recur, how often, and how many rows do the repeats account for?

Reference: Nigrini, M. J. (2011) Forensic Analytics, Wiley, ch. 8 (the number-duplication test); amounts compared to the cent

Returns

rows: Amounts tested
distinctAmounts: Different amounts
duplicatedAmounts: Amounts that occur more than once
duplicateRows: Rows carrying a duplicated amount
mostFrequentAmount: The amount that recurs most (the smallest such, on a tie)
mostFrequentCount: How often it occurs

The example

{
  "amounts": [
    1250,
    980.5,
    1250,
    4400,
    75,
    980.5,
    1250,
    310,
    5000,
    4400,
    18.99,
    6200
  ]
}

Agents: POST /faculty/mcom-forensic-accounting--investigation/duplicate_amounts with JSON {pan, …inputs} and the x-402-payment header. Manifest.