How every number is computed

Every figure on Arthkar is computed from raw data by code we wrote — no editorial opinions dressed up as scores, no numbers copied from other websites. This page explains each computation, where the data comes from, how often it refreshes, and what its limitations are.

Data sources & refresh cadence

• • Daily NAVs — AMFI India's official NAV file, ingested every trading day evening. 13M+ NAV records going back to 2006.
• • Historical NAV — mfapi.in community API for deep backfill, cross-checked against AMFI.
• • Portfolio holdings — monthly disclosures published by AMCs, refreshed after each month-end disclosure cycle.
• • Stock quotes — end-of-day exchange data, snapshotted after market close.
• • Category taxonomy — SEBI's official 2017 categorisation circular.

Honesty Score (1–5)

A rules-based score — not a rating we sell, not pay-to-play. We check a fund on fixed dimensions and each check produces a good / warn / bad signal:

• • Expense ratio vs its SEBI category average — are you overpaying for the same exposure?
• • Sector concentration — top-3 sectors ≥75% is flagged bad, ≥60% warns.
• • Style honesty — a “small cap” fund holding mostly large caps gets called out.
• • Volatility vs what its category promises.

The final 1–5 score is the rounded aggregate of these signals. Every fund page shows the individual signals, so you can disagree with our weighting and still use the underlying facts.

Drawdown stories

We walk the full NAV history maintaining a running peak. Every fall of 10% or more from that peak — and the time it took to recover — becomes a drawdown event. This is the standard running-maximum method used in fund research, computed on actual daily NAVs, not monthly samples.

SIP return pages & the Wealth Time Machine

Every “₹X monthly SIP for Y years” figure replays actual instalments against the real NAV on each SIP date, then reports XIRR. Nothing is projected or assumed — if the market crashed in your window, the number includes the crash. The Time Machine does the same computation month by month and shows it to you as a journey.

The SIP Crash Test

The SIP calculator's Crash Test injects a crash profile approximated from Sensex history (2008: −52% over 14 months, ~20 to recover; 2011: −25%; 2020: −38% over 8 weeks, ~9 months to recover) into your plan, then compares three investors who invest the same rupees: one panic-sells at the bottom (proceeds to a 6% FD), one pauses instalments (cash at 3%), one keeps investing. It is a scenario model — clearly labelled — not a prediction.

Portfolio X-Ray

Your CAS PDF is parsed in-memory and never stored; your PAN is one-way hashed. We roll every fund's disclosed holdings up to stock level to compute true concentration, compute pairwise overlap between your funds (≥40% common holdings gets flagged), and weight expense ratios by your actual allocation to show leakage in ₹/year.

Known limitations — read this

• • Holdings data is monthly and lags by up to ~5 weeks (AMCs disclose after month-end).
• • Deep portfolio enrichment currently covers the ~500 largest funds; the rest show NAV-based analysis until the enrichment cron reaches them.
• • Past returns — even real ones — do not predict future returns. We show history because it is honest, not because it repeats.
• • Stock daily-history features (like the 30-session price chart) are building up since July 2026 and appear as data accumulates.

Found an error in our numbers? Write to hello@arthkar.com — we fix data bugs with the same urgency as site outages.