Methodology

How India Risk Monitor sources, validates, scores, and refreshes its 60 metrics. Plain English.

Data sources

IRM pulls data from ~30 distinct sources in three categories:

Government direct: RBI (rates, FX, reserves) · MoSPI (CPI, IIP, WPI) · PIB ministry releases · DPIIT · PPAC (petroleum) · eaindustry.nic.in (Eight Core Industries)
Exchange / regulator: NSE (indices, F&O, FII/DII flows) · BSE · NSDL · SEBI
Industry / aggregator: FADA (auto sales) · CRISIL (credit spread) · JPC Steel · Drewry (WCI) · Baltic Exchange · Trading Economics · DBnomics · data.gov.in

Refresh cadence

Each metric has its own natural cadence. IRM refreshes when source publishes:

Cadence	Count	Examples
Live (~15 min)	7	Nifty, Bank Nifty, VIX, Brent, Gold, DXY, INR/USD
Daily	20	FII/DII flows, banking liquidity, G-sec yield, WACR spread
Weekly	4	FX reserves, reservoir levels, Drewry WCI, credit spread
Fortnightly	1	Credit-deposit growth
Monthly	25	CPI, WPI, IIP, GST, UPI, auto sales, PMI, trade deficit
Quarterly	2	CAD %GDP, port dwell time

How "stale" is shown

When a metric isn't refreshed within its expected window, a STALE pill appears next to it on the dashboard. The badge shows the age (e.g. STALE 5d) and the source's expected cadence. No silent staleness.

Resilience · why data keeps flowing

Each metric has a tier_chain of fallback sources tried in priority order:

Manual override · drop a JSON file to instantly fix any stuck metric
Primary source · government or exchange direct
Tier B/C backup · DBnomics, news aggregators, Wikipedia, industry RSS
LLM extraction · free Groq/Gemini parse article body if source format changed
Local content cache · 14-day gzipped HTML mirror falls back if upstream is down
Wayback Machine · archive.org snapshot as ultimate net

Plus: India self-hosted runner for IP-restricted government sites; Cloudflare Worker proxy as failover; daily content mirror; cross-source verification.

Data integrity checks

Plausibility bands per metric (e.g. Brent must be $30-$200/bbl) · rejects out-of-band parses
Anomaly detection (z-score) · flags values >4σ from 12-month sparkline
Shadow verification · suspicious values get cross-checked with secondary tier
Source attribution · every history row has its source URL stored
Snapshot store · 14-day archive of every successful fetch's raw HTML for audit

India Risk Score

The headline score (0-100) is a weighted average of 6 driver composites:

Driver	Weight	Feeders
Oil physical	3	Brent · India basket · Hormuz throughput · VLCC tanker rates
Freight	2	WCI · BDI · VLCC · port dwell
Institutional flows	3	FII equity · FPI debt · DII · absorption · F&O OI · blocks
India macro	3	INR · CPI · WPI · IIP · fiscal · CAD
Real economy	2	GST · PMI · IIP · steel · auto · rail freight
Sector breadth	2	VIX · PE · sector tier counts

Scoring scale: 0-40 calm · 40-60 neutral · 60-75 elevated · 75-85 high · >85 shock.

What this is NOT

NOT trading advice. No buy/sell signals. Analytical tool for research, not execution.
NOT real-time tick data. Live metrics refresh roughly every 15 min, not millisecond.
NOT individual stock data. IRM is macro + flows. For stock screens use Screener.in.
NOT comprehensive. 60 metrics cover the broad picture but miss niche signals (commodity futures spreads, options Greeks, etc).

License + attribution

IRM data outputs (CSVs and JSON in /data/) are released under CC-BY 4.0. The dashboard code is private but the data and methodology are open. Attribute as: "Source: India Risk Monitor (india-risk-monitor.pages.dev), aggregated from RBI/MoSPI/NSE/etc."

Built by

Aniket Kulkarni · SEBI Registered Research Analyst INH000009843 · Pune. Independent. No conflicts. No paywall. No data sale.

Last updated 2026-05-12.