Research, not signals.

Calibrated context, daily, for the Indian markets.

A research brief in your inbox before NSE opens, followed by a public audit of yesterday's calls.

View sample briefSee track recordRead our philosophy →

Discipline 01

We do not say buy or sell.

Every research line in every brief we publish describes regime, probability, and context. Never an entry. Never a stop. Never a target.

The discipline is enforced at render time. Any phrase from a forbidden vocabulary list — “buy”, “sell”, “target at”, “stop loss”, and others — raises in our build pipeline before the brief reaches your inbox.

The decision to act on any of our research remains entirely with you, the reader. Read why this matters →

Excerpt — watchlist section, redacted
WATCHLIST — instruments the model flags as in-play today: - LARGE_CAP_BANK_A (BANKING) long bias toward demand pool at ₹[subscriber-only]. P(test today) = 62%; P(test within 60min) = 28%. Confidence: high. - INDEX_NIFTY50 short bias toward supply pool at ₹[subscriber-only]. P(test today) = 47%; P(test within 60min) = 14%. Confidence: moderate.

Discipline 02

Every brief is followed by an audit.

Tomorrow’s brief opens by auditing today’s. Per-confidence-bucket hit rate. Calibration error per prediction type. Drift flags surfaced before they degrade.

When the audit is drawn from a retrospective replay of historical bundles rather than live-collected outcomes, we say so explicitly — and our renderer prepends a disclosure line so you never confuse the two.

No other research service in this space audits itself publicly. See how the audit works →

Excerpt — yesterday audit, real shape
YESTERDAY AUDIT — Note: 35% of resolved predictions were backfilled from historical bundles, not collected live. Treat the numbers below as a directional read, not a live track record. brief=BRIEF_2026_06_03 predictions_made=14 resolved=12 hit rate by confidence bucket: - proximity / very_high: n=4, hit_rate=75%, mean_p=84%, calibration_error=+0.09 - proximity / high: n=5, hit_rate=80%, mean_p=72%, calibration_error=-0.08 - proximity / moderate: n=3, hit_rate=33%, mean_p=58%, calibration_error=+0.25

Discipline 03

Calibration is the product.

A 70% probability we publish should resolve in our favour roughly 70 times out of 100. If it doesn’t, the gap shows up on the dashboard as calibration error — and we flag the head as drifting until it returns to within tolerance.

The whole engine is built around this contract. Every probability you read was calibrated against a hold-out window the model never saw during fitting.

Live calibration dashboard →

Proximity head — calibration error

Daily mean predicted probability minus actual hit rate, last 90 trading days. Closer to zero is better.

Excerpt — per-bucket calibration, last 30d
Last 30 days, proximity head: very_high (n=68): hit_rate=78%, mean_p=84%, error +0.06 high (n=152): hit_rate=71%, mean_p=72%, error +0.01 moderate (n=204): hit_rate=58%, mean_p=57%, error −0.01 low (n=311): hit_rate=29%, mean_p=32%, error +0.03

Pricing

Sober pricing for sober readers.

No freemium games, no urgency promos. Three tiers, monthly or annual. First seven days of the archive are free to browse after you create an account; after that a paid tier is required.

Intraday

Daily Brief, every NSE trading day before open.

₹—
/month (pricing finalising)

Multi-product

Daily Brief + Swing Brief. Monday pre-open + ad-hoc on regime flips.

₹—
/month (pricing finalising)

Diagnosis

One-off audit of your own strategy. Five trading days turnaround.

₹—
per audit (pricing finalising)
Full pricing →

Read one brief before deciding.

The sample brief is a real published artifact, redacted for public view. It will take six minutes to read.

View sample briefTalk to us