Stocks
100% current| Daily | 494/494 |
| Hourly | 494/494 |
| 15 min | 494/494 |
| Weekly | 494/494 |
Control room
Scheduler active · last heartbeat 20 Oct, 04:31 ISTBefore any research runs, the platform proves its data is current. Each group is checked against a target date, not just "downloaded".
Main indices & sectors
100% current| Daily | 14/14 |
| Hourly | 14/14 |
| 15 min | 14/14 |
Additional sectors
100% current| Daily | 13/13 |
F&O activity buckets
100% processed| Bucket 1 · lower | 70 |
| Bucket 2 · middle | 71 |
| Bucket 3 · higher | 71 |
F&O reference needs refresh · flagged on every page
The research flow
Long candidate screen · Step 1
Session 16 Oct 2026 · completedUses stored, verified data only. No downloads and no trades happen here.
Close > 50-day average+20-session return > 0
NIFTY 500
22,270
20 sessions −1.70% · Mixed
NIFTY 50
23,146
20 sessions −0.53% · Mixed
Breadth
76 / 212
F&O stocks above 20-day average
Coverage
100%
0 missing · 0 insufficient data
Funnel · Bucket 3 (higher F&O activity, the pilot bucket)
F&O universeAll buckets
212
Bucket 3 evaluatedHigher activity
71
Pass the screenResearch candidates, not buy signals
19
Do not passOne or both conditions not met
52
Long priorities · Step 2
19 candidates · Bucket 3Ranks each candidate by its 20-session return relative to its daily price swings. Two views: against the whole bucket, or against its own sector.
| Stock | Rank | Ratio | 20-session return | 2-session return | Sector context |
|---|
Long setups · Step 3
Signal 16 Oct · entry 19 Oct · exit by 22 OctA ranked stock still needs the right moment: first a pullback (P), then a recovery (Q). Only then is the hypothetical plan triggered. Click a status to filter, and a row to see its plan.
| Stock | Base · Rel | P · Q | Status | ATR |
|---|
Short side, same session
Mirror logic: first a rally, then a breakdown. Bucket 3: 47 candidates · 24 awaiting setup · 19 awaiting confirmation · 4 confirmed.
ONGCJIOFINBELPFC
Test & follow · Step 4
Research only · no ordersTwo separate questions. Do the rules work on history? And what happened to each saved plan in the next three sessions?
Historical window
25 May → 9 Oct 2026
Full declared universe, not today's survivors
Cost assumption
0.05% per side
Re-run at 0.10% for sensitivity
Smoke check
Completed
3 / 3 sessions · not performance evidence
Primary result
Not completed
No result is claimed until it runs
Saved plans · cumulative hypothetical return from entry
| Stock | Side | Ranks | Day 1 · 19 Oct | Day 2 · 21 Oct | Day 3 · 22 Oct |
|---|---|---|---|---|---|
| LAURUSLABSTrigger confirmed · timely | Long | Base 8 · Rel 1 | Unavailable | Unavailable | Unavailable |
| ONGCTrigger confirmed · timely | Short | Base 17 · Rel 20 | Not checked | Not checked | Not checked |
| JIOFINTrigger confirmed · timely | Short | Base 19 · Rel 7 | Not checked | Not checked | Not checked |
| BELTrigger confirmed · timely | Short | Base 20 · Rel 22 | Not checked | Not checked | Not checked |
| PFCTrigger confirmed · timely | Short | Base 30 · Rel 11 | Not checked | Not checked | Not checked |
Saved run history · long side
| Session | Confirmed | Candidates | Note |
|---|---|---|---|
| 16 Oct | 2 | 43 | All buckets |
| 14 Oct | 2 | 36 | |
| 13 Oct | 0 | 28 | No setup confirmed; nothing forced |
| 12 Oct | 7 | 24 |
Guardrails
Mode · forward shadowReadiness and permission are separate switches. A green data check never grants the right to trade.
Paper trading armedFalse
Execution authorisedFalse
Shadow scheduleOn
ReadinessUnavailable
Risk limitsProposed · need sign-off
Paper ledgerNot started
Kill switch. One control stops every new entry across the platform and is written to the audit log. Try it. In this demo it only changes the page.
State: entries allowed in shadow modeProduct decisions
- Stopped version 1 before anyone relied on itAn audit found data and model bugs, so I rebuilt from the data layer up instead of patching the model.
- Data proves itself firstEvery run starts with checks against a target date. A stale input is flagged on every page that uses it.
- Unknown is not zeroMissing results show as unavailable, with the reason. Nothing is filled in to look complete.
- Research and execution are separate"Trigger confirmed" never means buy. Paper trading and execution each need their own explicit approval.
- Test the whole universeHistorical tests replay every stock that was eligible then, avoiding survivorship bias from today's shortlist.
- Frozen dates, real costsEach plan is frozen when published, and returns are net of trading costs, with a higher-cost re-run.