equity-service · scan of 17 October 2026

Trading Book

Generated 2026-10-17 15:12 IST

Market breadth

The share of the universe closing above its own 20-day average, against the mean of the prior five sessions. The entry gate opens when the first is above the second. This is participation, not price: it is how a rally carried by a few heavyweights is told apart from one the market joined. The column below is the breadth comparison alone; a buy also needs the follow-through gate and a market status that could be read at all.

The last ten readings

Sector breadth

The share of each sector's names closing above their own 20-day average, over the last 30 sessions, on the same definition the entry gate uses for the market as a whole. Blue is the share above, red the share below. A sector needs at least 8 priced names to be drawn at all -- below that one stock moving is larger than the move the gate turns on.

TIS setups

The four TIS scans — daily and weekly, stocks and ETFs. Runs: SWING 2026-10-16 17:30 IST; WEEKLY 2026-10-16 17:33 IST; ETF 2026-10-16 17:51 IST; ETF_WEEKLY 2026-10-16 17:51 IST

SWING — 879 scanned, 391 with a stop derived, 10 at ≥7 of 8 gates, 1 actionable, actionable plus the ≥7-gate watch rows — blocked by G3 x9, G5 x7. sizing: Rs 20,000 risk / Rs 500,000 cap (defaults - 2 tenants configured)

WEEKLY — 879 scanned, 242 with a stop derived, 5 at ≥7 of 8 gates, 0 actionable, actionable plus the ≥7-gate watch rows — blocked by G4 x5, G5 x4. sizing: Rs 20,000 risk / Rs 500,000 cap (defaults - 2 tenants configured)

ETF — 23 scanned, 19 with a stop derived, 0 at ≥7 of 8 gates, 0 actionable, actionable plus the ≥7-gate watch rows. sizing: Rs 20,000 risk / Rs 500,000 cap (defaults - 2 tenants configured)

ETF: nothing actionable, and nothing within one gate of it (19 names with a stop derived, none at ≥7 of 7).

ETF_WEEKLY — 23 scanned, 20 with a stop derived, 0 at ≥7 of 8 gates, 0 actionable, actionable plus the ≥7-gate watch rows. sizing: Rs 20,000 risk / Rs 500,000 cap (defaults - 2 tenants configured)

ETF_WEEKLY: nothing actionable, and nothing within one gate of it (20 names with a stop derived, none at ≥7 of 7).

Rows are the scan's own record, not a recomputation. 'blocked by' lists every failing gate, not the first.

A *candidate* is any name the scanner could derive a pivot and a stop for, before a single gate is judged; the gates are then scored on every one of them so that a six-of-seven is visible the morning after. The shortlist a reader means by 'passed the gates' is the ≥6 count.

'to entry' is how far the buy trigger sits above the last price the scan saw — not an overnight gap. For the daily scanners the trigger is 0.2% over the last close, so a larger figure means the quote has since moved; negative means the market is already through it.

Other scanners

The two that are not TIS. Both are in the nightly scan, but only the four TIS scanners feed the agent: STRIKE is intraday and everything placed here is CNC, so these rows are read by hand. STRIKE is SWING with a 2R target; REVERSAL buys a washout and enters at the market, so its rows are judged on R1-R3 rather than on the eight gates. Runs: STRIKE 2026-10-16 17:47 IST; REVERSAL 2026-10-16 17:50 IST

STRIKE — 879 scanned, 391 with a stop derived, 10 at ≥7 of 8 gates, 1 actionable, actionable plus the ≥7-gate watch rows — blocked by G3 x9, G5 x7. sizing: Rs 20,000 risk / Rs 500,000 cap (defaults - 2 tenants configured)

REVERSAL — 588 scanned, 393 with a stop derived, 0 at ≥7 of 8 gates, 0 actionable, actionable plus the ≥7-gate watch rows. sizing: Rs 20,000 risk / Rs 500,000 cap (defaults - 2 tenants configured)

REVERSAL: nothing actionable, and nothing within one gate of it (393 names with a stop derived, none at ≥7 of 7).

A *candidate* is any name the scanner could derive a pivot and a stop for, before a single gate is judged; the gates are then scored on every one of them so that a six-of-seven is visible the morning after. The shortlist a reader means by 'passed the gates' is the ≥6 count.

What ran

The session before the last one, so each move has a day of follow-through beside it. `next` is that following session — a run given straight back is a different event from one that held.

20 names moved 3% or more; showing 15.

Themes: sectors with two or more of the 20 movers. This is what the list is read for — which group is working, so the next setup taken is in it. 'held' is how many kept the move the next session; a theme LAGGING on the RRG while leading the tape is a disagreement to resolve before sizing in it.

The 'why' column is the top headline a news search returned for that name, coloured by the same heuristic the News section explains. A blank means nothing was written, which for a name up 15% is itself worth noticing: a move with no story behind it is a move to distrust.

Group moves among the names shown — Chemicals-Specialty x3; Retail/Whlsle-Jewelry x2. The themes table below groups *all* the movers, not just these.

Regime by index

Each index judged on its own by the same rules as the market status — distribution days, rally attempt, follow-through. Reading only: the gate reads the market status, which is computed from the broad indices alone.

Broad indices

Sectors

A sector's distribution count is weaker evidence than the market's, and knowingly so. The bars carry total cash-market volume rather than the instrument's own, so the 'on higher volume' half of a distribution day is a fact about the whole market and only the price half belongs to the sector. Read it as 'this sector fell on days the market traded heavily'.

The proxies are ETFs, not the indices. An ETF can trade at a premium and carries its own liquidity; on a thin day its close is a worse reading of the sector than the index's own would be. It is what the bhavcopy gives us.

A blank MarketSmith column means their mail did not name that index, which is not a disagreement. They publish one headline status — which belongs to the Nifty and is shown on that row — and a sector block beneath it. There is no published line for smallcap, midcap, Next 50, PSU bank or healthcare.

The Nifty row can read Rally Attempt here while the gate stays shut, and that is not a contradiction. This column is the reporting map, which calls a held attempt what MarketSmith calls it; the gate reads the market status, which leaves a correction on a follow-through day and nothing else.

What the gate refused

Every setup the market filter turned down in the last two trading weeks, followed forward bar by bar with the entry and stop the scanner actually derived.

never triggered 70 · stopped 7 · open 33 · no bars yet 20

130 distinct setups over 10 scan days, 2026-10-05 to 2026-10-16, across 130 symbols. Refused for: breadth 387, computed status missing or downtrend 504 (before de-duplication; a setup can fail both).

Of the 40 that would have filled, 18% stopped out. Across all of them, stopped and open together, the gate cost -0.35% per setup. That is the number to watch: one name that ran is a tail, and this is the distribution it came from.

A setup that never triggered was refused nothing — the buy would not have filled either way, so those are not a cost. Only the filled ones are.

Colour is from the gate's point of view. A refusal that went on to stop out is the gate working; one still open and in profit is the gate costing money.

What is holding up

Every name's return since the index's last peak close on 2026-08-06, 49 session(s) ago, grouped by industry. Two benchmarks over the same window: NIFTYBEES -5.95% and SMALLCAP -1.31%. A sector above the line is outperforming even where it is still negative — in a drawdown that is what leadership looks like. The comparison that counts is against the smallcap peer: this universe is broad and measuring it against fifty large caps flatters almost all of it.

Holding up — best median first. 38 of 69 sectors are beating the smallcap peer.

Taking the damage — worst median first.

Median, not mean: one name up 80% on a takeover bid should not make its whole sector look like leadership. A sector needs at least 3 names in the universe to appear at all.

The anchor is the index's own highest close in the last 60 sessions, taken from price rather than from our recorded regime — so it does not move when MarketSmith relabels its status, and anyone can check it against a chart.

No reason is given, because the tape cannot give one. A group down 9% on crude and a group down 9% on one bad result look identical here; the News and Exchange filings sections below are where the difference lives.

News

Headlines for the market as a whole, newest first, from a Google News search. Company headlines sit on the setup cards and beside the names that ran.

Green and red are a **headline heuristic**: a few dozen words — 'bags order', 'wins contract' against 'probe', 'downgrade', 'profit falls'. It reads the headline and nothing else — not the article, not the exposure, not whether the move is already priced. It is there so the eye stops at the right row, and it is not a signal.

Open trades

The book, as this service records it — priced off the same series the charts are drawn from.

1 open, 0 with a stop that has moved. 'locked' is where the stop sits relative to entry in units of the original risk — positive means the trade can no longer lose.

How the market opened

No quote for 2026-10-17 — the newest quote is from 2026-10-16. Before the bell that is an outage in the pre-open mail; after it, the question has been answered by the tape instead.

MarketSmith

The emails the gate is read from, newest first.

MarketSmith's Buy Watchlist. Our universe holds tickers and the mail publishes company names, so a blank 'in our universe' is an unmatched name, not an absent one — match by eye against the scan.

Model portfolio moves

'parsed as' is the subject-line classifier, which recognises only status-change alerts — 'unrecognised' on a daily commentary is by design, not a failure. 'body says' is the Market Pulse read out of the mail itself; see `show_marketsmith` for the whole mail.

Newest mail (2026-10-16 17:23 IST): Confirmed Uptrend.

IPOs — mainboard

A separate instrument and a separate decision. No gate here reads a chart.

GMP is an unregulated over-the-counter quote with no print behind it. Measured over 295 listings it correlates 0.87 with the realised listing gain and overstates it by about 2.6pp; a zero GMP has averaged -6.6% and was positive 29% of the time. The subscription columns are the exchange's own and are the half that can be verified.

SME is excluded. Run backtest_gmp to re-measure any of the above.

Exchange filings

What the companies told the exchange, for names we follow.

Most filings are housekeeping — meetings, trading windows, newspaper copies. Only the categories that can move a price are flagged. The exchange's own category is kept verbatim rather than mapped onto our vocabulary, so a new one shows up unfamiliar instead of silently becoming 'other'.

This is the source a news search is not: a filing is dated, typed, keyed on the NSE symbol, and legally required. Searching news for RRKABEL the same day returned a festival marketing story twice and a two-month-old AGM notice.

Broker coverage

Context only. Nothing reads this as a gate.

A broker target is published after the move, by a firm with a relationship, and revised toward the price rather than ahead of it. A Hold with a target below the market is the useful case.