MONTHLY TRANSPARENCY REPORT — November 2026
One month, fully measured.
Report-card meets infographic: every section carries one headline stat, one dominant chart, and source + date stamps on each chart. Every number is MEASURED from the live production database (2026-11-20T20:54 IST) — nothing invented. Metrics without measurement history are written plainly as "n/a". Failures are published with equal prominence to passes — no cherry-picking.
1. Prediction accuracy — weekly trend
MONTH GRADED ACCURACY
48.8%
Source: prediction_logs, outcome CORRECT/WRONG, graded next-day close. Week-buckets table niche. 2026-11-20T20:54 IST.
GRADED THIS MONTH
111,044
Source: prediction_logs, timestamp in month window. No claims are made for pending grading.
min 45.5 · max 50.5
Chart: weekly graded accuracy %, November 2026 — data: prediction_logs (SELECT-only, 2026-11-20T20:54 IST).
WEEK GRADED ACCURACY Week of 2026-10-26 11 graded 45.5% Week of 2026-11-02 14,759 graded 50.5% Week of 2026-11-09 50,619 graded 49.1% Week of 2026-11-16 45,655 graded 48.0% Week of 2026-11-23 0 graded grading pending Week of 2026-11-30 0 graded grading pending
2. Gate decisions — system ne kya roka
TOTAL GATE-SKIPS
6,294
Source: autopilot_logs, action=SKIP, user_id NULL (system-level), month window. 2026-11-20T20:54 IST.
GATE / REASON SKIPS VedKuber vote SKIPPED — no direction: macro thesis gave no d 3,315 VedKuber vote SKIPPED — no direction: ML neutral (no BUY/SEL 772 VedKuber vote SKIPPED — entry window closed — new entries on 569 MACRO BUY CALL (CE) thesis (score 3.0), timing not ready: Sh 488 MACRO BUY CALL (CE) thesis (score 0.88), timing not ready: S 367 VedKuber vote SKIPPED — entry window closed — new entries on 340 VedKuber vote SKIPPED — entry window closed — new entries on 226 VedKuber vote SKIPPED — entry window closed — new entries on 217
Source: autopilot_logs SELECT, month window, 2026-11-20T20:54 IST.
3. Users & school-leads
NEW USERS THIS MONTH
9
Source: users.created_at in month. Total 59, paying 6 (live subscription). 2026-11-20T20:54 IST.
NEW LEADS
7
Source: leads.created_at in month; of which school leads 4 (school_name filled). Total leads 7.
SIGNALS CREATED
2,894
Source: signals.created_at in month. 2026-11-20T20:54 IST.
4. Model health
DEAD MODELS (5-MIN FAMILY)
193 / 224
Source: models/5m_*_model.meta.json — dead = cv_acc_mean ≤ majority_baseline_acc. VM current state, 2026-11-20T20:54 IST.
CALIBRATION ECE
raw ECE 0.3449 → calibrated 0.013 (fit 2026-11-17)
Source: models/confidence_isotonic.meta.json. n/a (ECE history not yet stored — trend begins next month)
Models promoted/dropped: n/a (promote/drop events not measured before 2026-11). Bottom voters PUBLISHED (no cherry-picking — bottom-decile rule):
TOP VOTERS (30d forward acc) ACC VOTES Multi-Timeframe 72.5% 5603 OI Analysis 71.4% 63 Market Sentiment 64.2% 9576 Technical Indicators 61.6% 8516 ML Prediction 55.4% 11218
BOTTOM VOTERS (30d forward acc) ACC VOTES ML Prediction 55.4% 11218 Guardian AI 54.7% 11818 Macro Impact 49.2% 9162 Risk Manager 45.0% 11818 Support/Resistance 16.6% 5056
Source: voter_performance, window_days=30, votes_cast ≥ 20. Backward-looking — predictive nahi. Full rating system: /ratings .
5. Galtiyan — incidents, verbatim
AUDIT VERDICTS THIS MONTH
READY 0 · MARGINAL 0 · NOT_READY 4
Source: weekly_audit_reports (user_id NULL), month window. NOT_READY reasons neeche VERBATIM — editing nahi.
"System scored 0.0/100. Performance does not meet the bar for live trading. Win rate 0.0% and/or P&L ₹-16,436 need improvement."— system audit, week of 2026-10-26, readiness 0.0 (verbatim, weekly_audit_reports)
"System scored 0.0/100. Performance does not meet the bar for live trading. Win rate 14.8% and/or P&L ₹-3,385 need improvement."— system audit, week of 2026-11-02, readiness 0.0 (verbatim, weekly_audit_reports)
"System scored 0.0/100. Performance does not meet the bar for live trading. Win rate 41.4% and/or P&L ₹-20,570 need improvement."— system audit, week of 2026-11-09, readiness 0.0 (verbatim, weekly_audit_reports)
"System scored 20.0/100. Performance does not meet the bar for live trading. Win rate 42.4% and/or P&L ₹13,648 need improvement."— system audit, week of 2026-11-16, readiness 20.0 (verbatim, weekly_audit_reports)
6. Coverage
PREDICTIONS THIS MONTH
129,740
Source: prediction_logs count, month window. Distinct symbols graded: 287.
LIVE WORLD SYMBOLS (7d)
243
Source: world_prices distinct symbols, last 7 days (hourly yfinance capture). Country cards + scores: /world-economy .
Honest-rules applied on this page (methodology: /ratings#methodology ):
Every rating's denominator is public — rating with hidden math is a marketing claim, not a rating.
Backward-looking disclaimer: past performance is not predictive.
NR is honest — no star/score from thin data; "n/a" beats an invented number.
No cherry-picking — bottom voters and NOT_READY audits published with same prominence.
Source + date stamp on every figure.
Pichhli editions: Daily Public Report · Weekly Audit Report · Track Record · Rating System