Transparent overview of how FinSight v18 calculates your data. All formulas, models and heuristics are described in detail here.
Version: 18 (November 2025) | Status: Current production version
To ensure the most accurate results, we base our calculations on standard Czech accounting statements (CAS). Here is an overview of where to find individual items.
| Item in FinSight | Where to find it (Income Statement / Balance Sheet) | Note |
|---|---|---|
| Sales | Income Statement: I. Sales of products and services + II. Sales of goods | Total company turnover. |
| EBIT | Income Statement: * Operating result | Earnings before interest and taxes. |
| EBITDA | Income Statement: EBIT + E.1. Value adjustments (Depreciation) | Earnings before interest, taxes, depreciation and amortization. |
| Interest expense | Income Statement: N. Interest expense | Cost of debt capital. |
| Cost of goods sold | Income Statement: B. Cost of goods sold | Used as approximation for COGS. |
| Total assets | Balance Sheet: TOTAL ASSETS | |
| Equity | Balance Sheet: A. Equity | |
| Debt | Balance Sheet: B.+C. Debt | Total debt (long-term and short-term). |
| Cash | Balance Sheet: C.IV. Cash and cash equivalents | Cash register + bank accounts. |
We measure the efficiency with which the company generates profit.
The company's ability to pay its obligations on time.
FinSight calculates cash flow using the indirect method from financial statements.
Note: Investing and financing cash flows are estimated from the balance sheet, as they are not directly available from the income statement. For more accurate analysis, we recommend using the official cash flow statement.
We measure capital structure and the company's ability to repay debts.
The break-even point shows at what sales the company will cover all costs.
Fixed and variable costs are estimated by industry (see section "Where We Use Estimates").
FinSight uses Holt's exponential smoothing with trend for predicting future values. The model automatically detects trends in historical data and extrapolates them into the future.
3 scenarios:
Forecasts are indicative and based only on historical data. Actual development depends on many external factors (market, competition, management, macroeconomics).
We use a modified model for private companies (non-manufacturing / private), which is more suitable for Czech private companies (s.r.o.) than the original model for publicly traded companies.
Statistical model that looks for anomalies in accounting (e.g., fictitious sales or cost capitalization). We use the full 8-factor model according to the original research by Prof. Messod Beneish (1999).
Interpretation: If the score is higher than -2.22 (e.g., -1.5), it indicates a probability of earnings manipulation. Values below -2.22 suggest low manipulation risk.
| Component | Calculation | Meaning |
|---|---|---|
| DSRI (Days Sales in Receivables Index) |
(Receivablest / Salest) / (Receivablest-1 / Salest-1) | Growth in receivables-to-sales ratio may indicate fictitious sales. DSRI > 1.2 is a warning signal. |
| GMI (Gross Margin Index) |
((Salest-1 - COGSt-1) / Salest-1) / ((Salest - COGSt) / Salest) | Decline in gross margin may indicate cost manipulation. GMI > 1.1 is a warning signal. |
| AQI (Asset Quality Index) |
(Non-current Assetst / Assetst) / (Non-current Assetst-1 / Assetst-1) | Growth in non-current assets ratio may indicate cost capitalization. AQI > 1.2 is a warning signal. |
| SGI (Sales Growth Index) |
Salest / Salest-1 | Rapid sales growth may be legitimate, but can also mask manipulation. SGI > 1.5 requires attention. |
| DEPI (Depreciation Index) |
(Depreciationt-1 / (Non-current Assetst-1 + Depreciationt-1)) / (Depreciationt / (Non-current Assetst + Depreciationt)) | Decline in depreciation rate may indicate extended asset life to increase profit. DEPI > 1.0 is a warning signal. |
| SGAI (SG&A Expense Index) |
(OPEXt / Salest) / (OPEXt-1 / Salest-1) | Decline in operating expenses-to-sales ratio may indicate hiding expenses. SGAI > 1.1 is a warning signal. |
| TATA (Total Accruals to Total Assets) |
(Net Profit - Depreciation - ΔWorking Capital) / Assets | High accrual items may indicate aggressive accounting. TATA > 0.05 is a warning signal. |
| LVGI (Leverage Index) |
(Debtt / Assetst) / (Debtt-1 / Assetst-1) | Growth in leverage may indicate attempts to hide problems. LVGI > 1.0 is a warning signal. |
⚠️ Important warning: Beneish M-Score is a statistical model, not proof of fraud. High scores may have legitimate reasons (rapid growth, acquisitions, changes in accounting policies). Low scores do not guarantee absence of manipulation. Use as a complementary tool for due diligence, not as the sole source of truth.
Some data is not directly available in basic statements, so we use standardized estimates for indicative analysis purposes. It is important to know about them.
Since we do not know the company's repayment schedule, we estimate annual principal repayment as 15% of total debt. This corresponds to an average loan of approximately 7 years. For companies with short-term financing, this estimate may be inaccurate.
If not provided, we estimate the "Retained Earnings" item as 40% of Equity. For start-ups, this may overestimate the score, for established companies it may underestimate it.
We estimate the ratio of fixed and variable costs based on the industry you choose. Specific ratios by industry:
| Industry | Share of variable costs | Note |
|---|---|---|
| IT/SW | 20% | Minimal variable costs (mainly people = fixed) |
| Services | 30% | Mainly people (fixed), minimal material |
| Healthcare | 40% | Medicines and materials vs. personnel |
| Transport | 55% | Fuel vs. vehicle leasing |
| Manufacturing | 65% | Materials, energy (variable) vs. machinery (fixed) |
| Construction | 70% | Materials and subcontracting |
| Retail | 75% | Purchase of goods is the main cost |
| General (default) | 60% | Average value for unclassified companies |
Reality may differ according to the specific business model. These values are indicative and based on average ratios in the given industries.
We primarily calculate as (Sales - COGS) - EBIT. If COGS is not available, we use a fallback estimate of 15% of sales, which corresponds to an average company with medium cost structure.
If not explicitly provided, we estimate the "Retained Earnings" item as 40% of Equity. This estimate is based on average ratios for Czech private companies (s.r.o.). For start-ups, this may overestimate the score, for established companies it may underestimate it.
If not explicitly provided, we calculate as Current Assets - Current Liabilities. If current assets are also not available, we use the fallback Cash - Current Liabilities, which is less accurate but still usable for indicative analysis.
FinSight is designed with emphasis on privacy and security. All calculations take place directly in your browser, except for PDF export and ARES API calls, which go through server-side proxy.
FinSight uses pure JavaScript (Vanilla JS) without frameworks. All financial indicators are calculated directly in your browser from the data you enter:
All inputs (financial statements) and outputs (calculations, charts) remain only in your browser. Only localStorage is used for:
All calculations run synchronously when entering data. Charts are generated using Chart.js directly in the browser without the need for server communication. Results update in real time.
FinSight v18 supports three languages: Czech, English and German. Translations are loaded from JSON files (i18n/cs.json, i18n/en.json, i18n/de.json) with a cache-busting parameter to always load the latest version. The language is automatically detected from the browser or stored in localStorage according to user preference.
FinSight v16 integrates data from official state registers for more comprehensive company analysis.
FinSight uses ARES REST API (https://ares.gov.cz/) to obtain basic company information:
Data is loaded via PHP proxy script (ares.php), which ensures secure API calls and parsing of JSON responses.
Like VR, insolvency data is also obtained via ARES REST API. The proxy script checks whether the company is in insolvency, reorganization or liquidation. Results are displayed in the "Company Profile" tab with a warning if the company is in trouble.
FinSight calls Financial Administration SOAP API (https://adisrws.mfcr.cz/) for:
SOAP requests are generated in PHP proxy and responses are parsed into JSON format for the frontend.
All API calls go through server-side PHP proxy (ares.php), which ensures:
ARES data is stored in localStorage for faster repeated loading. When the ID number changes, the cache is automatically refreshed. API calls run in parallel using Promise.all() for maximum speed.
Simulation of a bank risk manager's perspective. Evaluates whether the company can "handle" its debts.
Interpretation:
Heuristic for principal repayments: Since we do not know the company's repayment schedule, we estimate annual principal repayment as 15% of total debt. This corresponds to an average loan of approximately 7 years. For companies with short-term financing, this estimate may be inaccurate.
Quick estimate of company market value using EBITDA multiple. Multiples vary by industry, growth and risk.
Industry multiples (indicative):
⚠️ Warning: This is a simplified indicative value, not an official valuation. The actual company value depends on many factors (growth, competition, management, market conditions).
Interactive sensitivity analysis. Move sliders and watch how changes in sales, prices or costs affect profitability and returns.
Calculations in simulator:
The simulator helps plan scenarios and strategic decisions. All calculations run in real time directly in the browser.
Important change in version 10: PDF is now generated server-side using PHP script export.php and Dompdf library. This change enables more professional formatting and better document quality.
PDF is generated on the server using PHP script export.php and Dompdf library (https://github.com/dompdf/dompdf). Data is sent to the server only for PDF generation and immediately deleted after generation.
🔒 Security and privacy: All data is processed only in memory (RAM) and immediately deleted after PDF generation. No data is stored on disk or in a database. Data is received, processed and forgotten within a single HTTP request. PDF is generated and sent back to the browser, where it is downloaded.
Charts from Chart.js are converted to images using native canvas.toDataURL() directly in the browser. Images are sent to the server as base64 data URLs within the JSON payload.
PDF contains complete analysis:
PDF uses professional styling with Dompdf:
PDF export supports all languages (Czech, English, German). Language is automatically detected from user settings or passed explicitly in JSON payload.
FinSight v18 is built on modern web technologies with emphasis on performance, security and transparency.
FinSight does not use any database. All data remains in the user's browser. Only localStorage for ARES data cache, language and theme preferences, and backup of entered data.
FinSight is designed with maximum transparency. All calculations, formulas, models and heuristics are described in detail here. If you have any questions or uncertainties regarding the methodology, please do not hesitate to contact us.
Contact: dvorakh@gmail.com
Documentation version: 18 (November 2025)