Can I scan all NSE stocks for a weekday pattern?

Historical Patterns is one ticker per study, not a multi-stock scanner. Use Screening for stock-universe rules; its Classic Patterns Filter is separate.

What window can I choose?

A nonempty 15-minute-aligned window inside 09:15–15:30 IST. The end is exclusive, so 09:15–10:30 contains five complete 15-minute candles.

Can a missing session count as a flat outcome?

No. Missing, partial, source-declared closed and unsupported occurrences are excluded and reported separately. Absence of candles alone does not prove a holiday.

Which price starts the subsequent return?

Separate mode measures the remainder from the first candle's close. Grouped next-session mode measures the next selected session from the condition session's last close. Whole-window return uses a different baseline.

Does rebasing change the actual result prices?

It changes mini-chart presentation. Numerical evidence can be read from original UTC-millisecond OHLCV with includeBars, bounded to 175 candles per occurrence.

Can I save a study permanently?

There is no saved-study library or dedicated patterns.export command. Results are tab-local. You can prepare a separately saved local HTML note from the existing evidence without rerunning it.

Guide

Study NSE Weekday and Opening-Window Behaviour

Use Historical Patterns to study one NSE ticker over completed weekday/time windows: choose 15-minute-aligned IST hours, run explicitly, check coverage and then filter subsequent outcomes. Keep the first candle's condition separate from later performance. This is historical frequency research, not automatic geometric-pattern detection or a trading recommendation.

1. Ask a narrow historical question

An example question is: for this NSE stock, when Monday's first 15-minute candle fell, how often was the rest of 09:15–10:30 positive? Choose the ticker from the supported catalog and confirm the active feed can provide regular-session 15-minute history.

This is not a request to detect triangles or head-and-shoulders formations. Drawing tools annotate those shapes manually; Historical Patterns owns weekday and time-window studies.

2. Configure the condition and outcome

Choose Monday, separate mode, start 09:15 and end 10:30 IST. The end is exclusive: five candles make a complete window.

The first candle measures the condition (first close / first open − 1) × 100. The subsequent outcome is (last window close / first candle close − 1) × 100, using only the four later candles. Whole-window return includes the condition candle and answers a different question.

Choose 1–200 completed candidate weeks. A week count is not a promise of that many eligible sessions. Adding Wednesday in separate mode doubles candidate rows per complete calendar week, not necessarily available market observations.

3. Check coverage before filtering

Start analysis explicitly and wait for completion. Read complete, partial, missing, closed and unsupported counts. A source-declared holiday is different from an empty history response; neither becomes zero return.

Cloud-backed selected windows can be batched. Provider/local reads retain the original source. Do not switch feeds or invent candles to make the sample look complete.

Then select a down-condition filter and a threshold you chose. This is an illustrative research setup, not a recommended signal. Positive, sideways-band and recovery measures have different meanings; record which one was used.

4. Inspect more than the visible page

Statistics cover all filtered eligible occurrences, while the UI displays six cards per page and commands return at most ten rows. Read the actual observation denominator, hit frequency and median before interpreting a result.

Open original candles to verify timestamps and prices. Rebased mini-charts aid visual comparison but are not fill prices. A complete, source-verified occurrence can restore its ticker and dated 15-minute window in the original chart pane without switching the feed.

5. Try an ordered two-session question carefully

For Friday → Monday, choose grouped mode with weekdays [5,1], condition weekday Friday and next-session focus. Order matters. The outcome is Monday's last selected-window close versus Friday's last selected-window close. Grouped mode's default selected-session focus instead studies the condition weekday's first candle and its remainder. A missing Monday is not replaced by Tuesday.

Do not interpret descriptive frequency as probability of profitable execution. The study has no brokerage, slippage or trading account. Translate a hypothesis into explicit rules and test it separately with Backtesting.

Use it with an AI assistant

Use tl-patterns with real commands patterns.getState, patterns.configure, patterns.run, patterns.getResults, patterns.setAnalysis and patterns.openOccurrence. Ask the assistant to retain run identity, coverage and original candle units, and never rerun merely to export existing observations.

See the TradingLead command reference and AI prompt cookbook. A small explanation can be saved with the separate Research reports workflow.

Steps

Open a ready supported NSE chart, then Patterns.
Select one ticker and a weekday such as Monday.
Set a window such as 09:15–10:30 IST and a candidate-week count.
Run and check complete, missing and excluded coverage.
Filter conditions and inspect subsequent outcomes, not whole-window return alone.
Open a complete occurrence and verify original candles.
Document results and assumptions without claiming a future-profit probability.

Limitations

NSE regular-session 15-minute history only.
Candidate weeks are not guaranteed trading sessions or complete years of data.
No automatic chart-pattern detector, cost model or executed strategy.
Filtering introduces selection bias; historical frequencies are not calibrated forecasts.
Opening occurrences requires source verification and does not switch feeds or end replay.
Provider quotas and source errors can affect analysis; no automatic rerun or fallback.

Frequently asked questions

Can I scan all NSE stocks for a weekday pattern?: Historical Patterns is one ticker per study, not a multi-stock scanner. Use Screening for stock-universe rules; its Classic Patterns Filter is separate.
What window can I choose?: A nonempty 15-minute-aligned window inside 09:15–15:30 IST. The end is exclusive, so 09:15–10:30 contains five complete 15-minute candles.
Can a missing session count as a flat outcome?: No. Missing, partial, source-declared closed and unsupported occurrences are excluded and reported separately. Absence of candles alone does not prove a holiday.
Which price starts the subsequent return?: Separate mode measures the remainder from the first candle's close. Grouped next-session mode measures the next selected session from the condition session's last close. Whole-window return uses a different baseline.
Does rebasing change the actual result prices?: It changes mini-chart presentation. Numerical evidence can be read from original UTC-millisecond OHLCV with includeBars, bounded to 175 candles per occurrence.
Can I save a study permanently?: There is no saved-study library or dedicated patterns.export command. Results are tab-local. You can prepare a separately saved local HTML note from the existing evidence without rerunning it.

Last updated: 7 November 2026