Methodology & Limitations

How the fundamentals score is built

Each security is scored out of 100 across five categories. Every factor is scored by clamped piecewise-linear interpolation between published control points — there is no machine-learned weighting, no hidden tuning, and the same thresholds apply to every stock in every market.

When a data point is missing, that factor receives 40% of its weight rather than zero — absence of data is not evidence of weakness. If fewer than 35% of factors have data, the security is marked “Insufficient Data” rather than given a misleading score. The exact control points live insrc/lib/scoring.ts.

Where the data comes from

• • Quotes, fundamentals, price history: Yahoo Finance via the unofficial yahoo-finance2 client. Typically delayed 15–20 minutes. Yahoo can and does change or withdraw fields without notice.
• • Indian mutual fund NAVs: AMFI data via MFAPI.in, usually one day in arrears.
• • FD rates: a manually researched, dated snapshot from public rate trackers — not a live feed. Always confirm on the bank’s own site.
• • Tax slabs: India FY 2026-27, excluding surcharge, HRA and many deductions.

What this platform deliberately does not do

• • No price predictions or targets. No credible method exists for forecasting individual stock prices, and a number generated by a language model is not a forecast — it is a guess wearing a suit.
• • No personalised buy/sell/hold calls. Producing those for an individual is regulated activity in India (SEBI Investment Adviser / Research Analyst) and in the US (Investment Advisers Act). Kautika holds no such registration.
• • No fabricated track record. The ledger on the Signals page starts empty and only records signals forward in time, stamped before outcomes are known.
• • No fundamentals backtest. Our data source provides only current fundamentals, not point-in-time history. Backtesting a P/E screen with today’s P/E would use information that didn’t exist then — the most common way published backtests are quietly faked. We only backtest price-based rules.

Known weaknesses you should hold against us

• • Survivorship bias. The screener universe is ~175 companies that are large and liquid today. Firms that failed or were delisted are absent, so any historical statistic computed over this set is optimistic.
• • Accounting is not reality. Reported figures can be aggressive, restated, or simply late. Scores inherit every flaw in the filings.
• • Banks and financials score oddly. Debt/equity and current ratio mean something different for a lender. Treat those scores with extra scepticism.
• • Sparse coverage outside large caps. Smaller and foreign-listed names often lack the fields the score depends on.
• • Technical readings are lagging by construction. Moving averages and RSI describe what already happened; they whipsaw in sideways markets.
• • Backtests exclude costs and taxes. Brokerage, spread, slippage, STT and capital gains are not modelled and would reduce every result shown.