Forensic Framework
Used for monitoring, not prediction

How Flagium
Measures Structural Equity Risk

Flagium AI applies structured forensic analytics to public company filings to detect early signs of financial deterioration — before concerns are broadly recognized.

What the score means

Structural financial stress, measured 0–100

What data is used

Quarterly & annual public filings, XBRL

How often updated

Synced each earnings season

Lookback window

12 quarters (3 years) of filed data

01. Framework Overview

Six Forensic Pillars

The risk score is constructed by evaluating each company across six independent forensic pillars. Each pillar monitors a distinct dimension of financial health. Signals from all pillars are synthesised into a single composite score.

📊
01

Earnings Quality

Examines whether reported profits are backed by actual cash generation. Flags divergence between operating cash flow and stated net income, which can indicate accounting-driven earnings unsupported by business fundamentals.

Monitors: Cash conversion deficit, profit collapse, margin erosion
💧
02

Liquidity & Coverage

Assesses the company's ability to service its obligations in the near term. Monitors free cash flow generation and interest coverage to identify companies that are running on borrowed time.

Monitors: Negative FCF, interest coverage stress, working capital expansion
🏛️
03

Solvency & Leverage

Evaluates the long-term structural balance between revenue generation and debt accumulation. Identifies companies where debt is growing disproportionately to operational output.

Monitors: Revenue-to-debt divergence, capex inefficiency, operating leverage stress
📉
04

Trend Deterioration

Tracks the rate of change in financial health indicators across multiple reporting periods. A single bad quarter is different from a multi-quarter deterioration sequence. This pillar captures the trajectory, not just the snapshot.

Monitors: Multi-quarter signal persistence, acceleration of stress indicators
🔭
05

Relative Sector Risk

Benchmarks each company's stress indicators against its sector peers. Stress that is systemic across a sector is treated differently from stress that is company-specific, enabling more precise risk attribution.

Monitors: Sector-relative margin compression, growth weakness vs. peers
⚖️
06

Governance Signals

Monitors non-financial indicators that historically precede financial deterioration. Auditor changes, regulatory penalties, and promoter-level actions are tracked as leading indicators of structural risk.

Monitors: Auditor resignations, material weaknesses, promoter pledge changes

Business Model-Adaptive Frameworks

The forensic engine adapts its signal set to the underlying business model of each company — not just its industry label. A Pharma company with inventory is evaluated differently from a SaaS company with zero inventory, even if both are listed in the same broad sector.

Applied to All Companies

F1–F5

Core Financial Health

  • Cash conversion deficit
  • Free Cash Flow stress
  • Revenue-to-debt divergence
  • Interest coverage ratio
  • Profit collapse
G1–G2

Governance Overlay

  • Promoter integrity signals
  • Auditor changes & material weaknesses
  • Regulatory penalties
  • Board-level compliance

Specialised by Business Model

F6–F10

Banking & NBFC

Banks, NBFCs, HFCs

  • NIM compression
  • GNPA spike
  • Provision coverage decay
  • Cost of funds divergence
  • Capital adequacy (CAR)
M1–M3

Manufacturing & Physical

Auto, Steel, Pharma, FMCG, Capital Goods

  • Inventory stress & bloat
  • Working capital expansion
  • Capex efficiency decay
I1–I3

IT & Asset-Light Services

IT Services, Consulting, SaaS, Platforms

  • EBIT margin compression
  • Revenue concentration risk
  • Relative growth weakness vs. peers
A1

Asset Management

Mutual Fund AMCs, Wealth Managers

  • AUM yield compression
  • Fee margin erosion
  • AUM growth vs. sector peers

02. Data Sources

What We Analyse

Every risk score is derived exclusively from publicly available, officially disclosed financial data. We do not use estimates, analyst consensus, or market-derived signals.

📋

Quarterly Filings

Q1–Q4 financial results published by companies on BSE/NSE regulatory portals.

📁

Annual Reports

Audited full-year consolidated and standalone financial statements.

🗂️

XBRL Disclosures

Structured financial data submitted in XBRL format to Indian stock exchanges.

📐

Standardized Financials

Normalized balance sheets, P&L, and cash flow statements reformatted for forensic comparison.

⚖️

Audit & Governance Records

Auditor change filings, regulatory penalty notices, and board-level disclosures.

🔭

Sector Intelligence

Aggregated peer-group metrics used to benchmark individual company performance.

Coverage: 1,500+ NSE/BSE-listed companies across Corporate, BFSI, and AMC sectors.

03. Scoring Logic

How Signals Become a Score

Signals are not simply added together. Each signal is evaluated across five dimensions before contributing to the final score. For example, the combination of rising debt, weak OCF, and recurring margin stress can accelerate a score materially compared to isolated events.

🎯

Severity

Each signal carries a base impact weight reflecting the structural severity of that financial event. Survival-level risks (solvency failure, capital decay) are weighted more heavily than temporary margin pressures.

🔁

Persistence

A signal that recurs across multiple consecutive quarters carries more weight than a one-off event. The engine distinguishes between isolated anomalies and sustained deterioration patterns.

📈

Trend Acceleration

The rate of change matters. A company whose stress is rapidly accelerating quarter-over-quarter is treated as higher risk than one with stable, static stress indicators.

🔭

Sector Context

Each signal is benchmarked against sector peers. Company-specific stress is weighted more heavily than sector-wide deterioration, which may reflect macro conditions outside the company's control.

⏱️

Recency & Time Decay

Not all historical data carries equal weight. Signals from recent reporting periods are given significantly higher analytical weight than older ones. This ensures the risk score reflects the current operating environment, not a company's state from several years ago. As data ages, its contribution to the score decays systematically — preventing stale signals from masking genuine improvements or deteriorations in the present period.

Weighting Philosophy

Signal weights are calibrated against historical corporate distress cycles specific to Indian listed markets — not imported from global models. They are tuned to maximise the correlation between early warning scores and subsequent credit or operational deterioration events. The logic and variables the engine weighs are disclosed here in full; the exact calibration coefficients are not published.

04. Score Interpretation

Reading the Risk Score

Scores are designed for relative comparison across a portfolio — not as absolute thresholds. A score of 65 in an industry with median scores of 70 carries a different meaning than the same score in an industry with a median of 20.

0 – 20
Low Risk

No material stress signals detected. Financial structure appears stable across monitored pillars.

21 – 40
Moderate

Early or isolated signals present. Warrants monitoring but not immediate concern.

41 – 60
Elevated

Multiple signals active. Structural stress is building across one or more pillars.

61 – 80
High Risk

Persistent, multi-pillar deterioration. Requires close scrutiny and position review.

81 – 100
Severe

Severe structural breakdown across multiple dimensions. High probability of continued deterioration.

05. Limitations & Disclaimer

What Flagium AI Does Not Do

Understanding the boundaries of any analytical tool is as important as understanding its capabilities.

Not a Buy/Sell Signal

Risk scores indicate structural financial stress — not price direction. A high-risk score does not mean the stock will fall, and a low score is not a buy recommendation.

Based on Public Filings

The engine only analyses what companies disclose in official filings. It cannot detect fraud or misstatements before they are publicly revealed.

Backward-Looking Data

Scores are built from historical reported financials. They reflect what has been reported, not real-time operating conditions.

Not Investment Advice

Flagium AI is a research and monitoring tool. All investment decisions should be made in conjunction with licensed financial advisors.

Flagium AI is a structural financial risk monitoring system. It is not a trading system, price prediction engine, or investment advisory service. All outputs should be interpreted in conjunction with independent research and qualified financial advice.

Advanced Methodology Inquiries

For analysts, advisors, and professional investors seeking deeper framework discussions, please reach out.

Contact Us

Methodology in Action

KIRLOSENG: Anatomy of a Forensic Score

See how our 6-Pillar Framework and 12-quarter trajectory mapping identify structural stress in a real-world manufacturing business model.

Explore the Walkthrough