Methodology
A public, high-level description of how this laboratory produces valuation estimates and financial quality indicators for research use.
1. Quantitative valuation
The valuation engine uses machine-learning ensembles trained on publicly disclosed fundamentals to produce a model-based market-cap estimate. From that estimate the platform derives an implied price and a relative valuation multiple (MVP).
MVP is estimated market cap divided by current market cap. The model price estimate is MVP times the current share price. MVP above 1.8 may indicate valuation bias or risks not fully reflected in the estimate. Further review is needed.
2. Earnings and risk estimates
Future ROE is estimated as the next four quarters of attributable profit divided by current positive attributable equity. It is an earnings estimate, not an overall financial health rating. Reported period ROE uses a different period and denominator basis.
Annual audit risk estimates the chance of a non-standard audit opinion; the target year is listed in the company details. ST risk estimates entry into special-treatment status over the next year. N/A means unavailable or inapplicable, not zero risk.
3. Data dates and updates
Market data refreshes daily; valuation and financial estimates change with model releases, not daily retraining. Reporting-period end, quote date and model information cutoff describe different dates.
Scores for past reports are recomputed with the current model. They are not predictions issued at those historical dates or evidence of out-of-sample accuracy. The website does not currently provide a daily valuation history.
4. Coverage and taxonomy
Company pages and industry comparisons cover non-financial A-shares with published valuations, not every listed company. Industry filters use the Shenwan taxonomy with Chinese names. Missing values are excluded from each metric's statistics, not treated as zero.
Use a full six-digit ticker or an unambiguous Chinese company name. Industry tables show up to 30 companies with the sort order stated explicitly; this is not a recommendation ranking.
Limitations
Models are imperfect approximations. Estimates can be sensitive to accounting quality, industry structure, and regime changes. Historical fit does not imply future performance. Outputs should be interpreted as research signals, not trading instructions.
Disclaimer
All content on this site is for quantitative analysis and research purposes only. It does not constitute investment advice, a recommendation, or an offer to buy or sell any security. Users remain responsible for their own judgments and risk.