Feature
Historical Patterns for NSE Weekday Research
TradingLead Historical Patterns studies one NSE ticker across completed historical weekday and 15-minute time windows. Compare separate sessions or ordered weekday groups, filter observed outcomes and inspect the original candles. This is descriptive time-window research, not automatic detection of triangles, candlestick formations or head-and-shoulders patterns.
What question does it answer?
Historical Patterns asks questions such as: after a stock's first Monday candle falls, what happened during the rest of the chosen morning window? It retrieves dated windows for one NSE ticker, preserves their source prices and reports outcomes only where coverage is complete.
This is different from the stock screener, which checks many stocks against rules, and Chart Grid, which displays multiple ticker cards. It is also different from manually drawing a geometric pattern.
Separate sessions versus ordered groups
| Mode | Condition and outcome |
|---|---|
| Separate | The first 15-minute candle is the condition; remaining candles in the selected session window are the subsequent outcome |
| Grouped, selected-session focus | The chosen weekday's first candle is the condition; its remaining window is the outcome |
| Grouped, next-session focus | The whole selected condition session is compared with the following selected session in the ordered group |
For example, Monday 09:15–10:30 IST, end exclusive, has five candles in a complete window. The first candle explains the condition; four later candles form the outcome. A one-candle window has no subsequent outcome.
Friday → Monday groups cross the weekend but stay within one seven-day cycle. Choose next-session focus to compare Friday with Monday; grouping alone does not select that outcome. The order is meaningful. A missing Monday is not silently changed to Tuesday, and a source-declared holiday is not a flat trading observation.
Settings and result measures
- Choose ISO weekdays Monday 1 through Sunday 7 and 1–200 candidate weeks. Regular weekday/session eligibility still applies.
- Windows must be 15-minute-aligned inside 09:15–15:30 IST.
- Filter by condition direction (all, down or up) and a 0–20% threshold.
- Measure positive outcomes, sideways outcomes within a 0–20% band, or recovery relative to the condition's first open.
- Sort newest/oldest, show excluded occurrences, choose chart/statistics/table views and optionally rebase chart prices.
Separate-session outcome return is (last window close / first candle close − 1) × 100. Grouped next-session return uses the condition session's last close as its baseline. Whole-window return instead compares the final displayed close with the first displayed open. These formulas are not interchangeable.
Filtered statistics report eligible observation counts, hits/frequency and median measured outcome; coverage also separates complete, partial, missing, closed and unsupported rows. Statistics cover all filtered eligible rows, not just the six-card UI page.
Coverage before conclusions
The source must declare the selected NSE ticker, Asia/Kolkata timezone, regular session and 15-minute support. A ready chart does not guarantee the requested years of history. Cloud-backed windows can use a batched candle request; provider/local paths use their own bounded history without a cloud fallback.
Inspect excluded counts before interpreting a frequency. A 60% positive frequency among a small filtered sample is a historical description, not a calibrated 60% chance of profit. Fees, order timing and position sizing require a separate backtest.
Original candle reads retain UTC milliseconds. Rebased mini-chart prices are visual transformations, not execution prices. Open a complete occurrence to navigate its original pane while retaining source guards; analysis does not switch feeds or exit replay to conceal failure.
Use it with an AI assistant
Use tl-patterns. Real commands include patterns.getState, patterns.configure, patterns.run, patterns.getResults, patterns.setAnalysis and patterns.openOccurrence. Ask the assistant to state candidate weeks, window formulas and exclusions, wait for completion and inspect original candles before summarizing a pattern.
See the TradingLead command reference, AI prompt cookbook and NSE weekday study guide. Discovery is not proof of historical-data availability.
Steps
- Open a ready NSE chart and the Patterns sidebar.
- Choose the ticker, weekdays and a 15-minute-aligned IST window.
- Choose 1–200 candidate weeks and separate or ordered grouped mode.
- Run explicitly and wait for completion.
- Review coverage before filtering statistics.
- Inspect original candles and open a complete dated occurrence.
Limitations
- One NSE ticker and regular-session 15-minute windows only.
- Candidate weeks are calendar selections, not guaranteed source coverage.
- Missing/partial occurrences are excluded, not treated as zero-return days or inferred holidays.
- Regular-session support does not establish Muhurat or other special-session analysis.
- Studies and analysis state are account-scoped and tab-local, without a saved-study library or dedicated export command.
- Source readiness and installed command support must be discovered; analysis can consume provider quota.
- Descriptive observations only, not pattern-recognition signals, order execution or future-return probabilities.
Frequently asked questions
- Does TradingLead automatically detect head-and-shoulders or triangles?
- Not through Historical Patterns. It studies weekday and intraday-window outcomes for one ticker. Geometric drawing tools let you annotate supplied points manually; the scanner's Classic Patterns Filter is a separate rule, not this sidebar.
- Which markets and intervals can I study?
- One explicit NSE ticker at a time, using 15-minute candles within the regular 09:15–15:30 Asia/Kolkata session. Ratios, other exchanges, unsupported sessions and unavailable source history are rejected.
- Does 200 mean 200 trading days?
- No. It means up to 200 completed candidate weeks. In separate mode, selecting Monday and Wednesday produces two candidate rows per week; holidays and missing data can reduce eligible observations.
- Can I compare Friday with the following Monday?
- Yes. Grouped mode preserves an ordered sequence such as Friday then Monday, within one seven-day cycle. The chosen condition weekday must have a following selected weekday. A holiday is not replaced with another observed trading day.
- Can I inspect actual prices rather than rebased mini-charts?
- Yes. Results can expose original UTC-millisecond OHLCV candles, bounded to 175 per grouped occurrence. Rebasing changes display only. Complete source-verified occurrences can open in the original chart pane.
- Are pattern statistics a tested trading strategy?
- No. They describe filtered historical observations without execution costs or a trading account. Coverage gaps, selection bias and repeated filtering matter; test explicit rules separately in Backtesting.
Related
Last updated: