AI Briefings
a professional morning note, written by the machine from the live tape
BRIEF #142 · GENERATED 13:07 BDT · 90-SEC READ · SOURCES: LIVE TAPE · FILINGS · NEWS CLUSTERS
DSEX up 0.00% at 5924.6 — 0 advancing vs 0 declining
- NEW · flow leadership moved Food → Pharma
- NEW · breadth flipped positive (yesterday 21 declined)
- UNCHANGED · foreign desk a buyer 5 sessions running
Buyers have the tape — Pharma holds the flow lead and ACMELAB is the session's engine.
Breadth is split — 0 advancing against 0 declining. This is a stock-picker's tape rather than a trend day.
Turnover is concentrating in Pharma and Bank — together 40% of traded value. Textile is being left behind.
Leaders: ACMELAB +0.0%, BATBC +0.0%, BRACBANK +0.0%. Laggards: WALTONHIL +0.0%, TITASGAS +0.0%, SUMITPOWER +0.0%.
Shield watch: DHAKAFIBRE sits at 97/100 — outsized moves there deserve scepticism.
A green index doesn't mean every stock follows. Check breadth before reading anything into your own names.
Production cadence: 9:30 pre-market · every 30 minutes live · post-close wrap · Sunday week-ahead. Written from simulated market data — this is not investment advice, and nothing here is a recommendation to buy or sell.
What Moved the Index
Who actually did the work behind the index move.
Nothing is dragging the index — every name carrying material turnover contributed positively.
Contribution = price move × turnover weight. Leaders and drags are shown separately, and an index move that three names explain is narrower than it looks.
Days Like Today
What happened after the sessions that looked like this one.
Across the 51 matching sessions the next day was a mild upward drift — base rates over a year of this market, not a forecast for tomorrow.
Flow Intelligence
Who is buying, and whether they agree with each other.
- Foreign deskBUYINGnet +4.5 crore, 5 sessions on this side, concentrated in ACMELAB + BATBC
- Institutional blocksTWO-WAYbuy-side pressure 50% of turnover-weighted flow — neither side is pressing
- Retail flowSTEADY0% of this tick's velocity is in circuit-heavy and Z-category listings — in line with a calm session
The desks and the crowd are on the same side of this tape, which is the case that carries the least information. This build carries no history of desk positioning, so neither side of that is scored.
Desk and block flow are simulated. The base rate quoted is drawn from this market's own history, not from observed trading.
Under the Radar
Quiet signals the tape has not moved on yet.
- ROBIEPS +29% year on year against +5.2% of price momentum — the widest earnings-to-price gap in the Universe.
- ISLAMIBANK4.0% of its free float changes hands daily with no story attached to it on today's tape.
- SUMITPOWERA 8.6% yield reached by price drift rather than by a raised dividend — it crossed into income-screen territory without an announcement.
Divergences between fundamentals and price close slowly. These are things to look at, not signals to act on.
What This Bears On
What the note above bears on, depending on why you are here.
LONG-TERM
Nothing dated this week changes a multi-year thesis. Across this market's history, single red sessions inside an intact trend were retraced within 5 sessions in 68% of the 66 cases counted.
INCOME
A policy-rate hold leaves deposit-beating yields where they are. Two dated items bear on income this week: SQURPHARMA's record date on Sep 2, and the yield screen's membership, which the session's drift changes without any announcement.
TRADER
On this tape the index has carried 40% of traded value in two sectors, with Pharma leading — dispersion has sat between sectors rather than in the index. Two dated events fall inside the week: the CPI print and the excise session.
Each of these states what the material above bears on for that horizon. None of them is a recommendation, and none tells you what to hold.
The Honest Ledger
The strongest argument each way, and what would break it.
STRONGEST BULL POINTS
- ▸ The two strongest supporting analogs on today's tape are industrial energy squeeze easing after lng cargo lands (75% hit rate over 4 cases) and bsec drafts margin-loan reform; floor-price era formally closed (72%).
- ▸ The foreign desk is 5 sessions into a net-buy run concentrated in index heavyweights — patient money rather than hot money.
- ▸ Breadth is confirming the move: 0 names are advancing, so participation is real rather than three tickers doing the work.
MIND-CHANGER ▸ breadth turns negative for two straight sessions, or the foreign desk flips to the sell side
STRONGEST BEAR POINTS
- ▸ The strongest opposing analog is tobacco excise risk returns in supplementary-budget talk: 5 prior cases, -3.2% average move over the following week.
- ▸ Turnover leadership is narrow — two sectors carry 40% of traded value, and narrow tapes have historically broken with less warning than broad ones.
- ▸ Retail velocity into circuit-heavy listings is running at 0% of the tick, against the 9% a calm session runs at.
MIND-CHANGER ▸ the flow leadership broadens past two sectors, or the reform cluster's analog is confirmed with turnover behind it
Both cases are argued from simulated stories and this market's own history. The panel exists to state the case against whichever way you already lean — it is not investment advice, and neither side is a recommendation.
Market Regime
What kind of session this is.
- TRENDweak ↔ strong
- BREADTH50% advancing
- VOLATILITYcompressed
Neither Breadth nor the spread between stocks is confirming a direction. Moves on tapes like this one have reversed about as often as they have extended in this market's history, so the session carries less information than its size suggests.
REGIME BREAKS IF ▸ turnover pace passes 120% of the average session — drift with turnover behind it is a different regime.
Audio Brief
The note, read aloud
Simulated player — it produces no audio. Production narrates in Bangla or English.
Your Names Today
Your own names, and why each is on today's tape.
Machine vs Crowd
Where this briefing's reading differs most from the crowd's.
The five-factor composite ranks it 29/100, below where the crowd sits — the gap is carried by the earnings trend and the yield.
The five-factor composite ranks it 30/100, below where the crowd sits — the gap is carried by the yield and its Shield score.
The five-factor composite ranks it 37/100, below where the crowd sits — the gap is carried by the yield and the earnings trend.
Crowd figure withheld — 46 contributors, below the 50-contributor floor.
Crowd positioning is simulated: this build has one local user, so the figures are generated rather than observed. No crowd figure is shown below 50 contributors, and 7 of 26 listings are withheld on that rule right now.
Risk Radar
What could go wrong, and what would tell you first.
- Election drift35%
Index −2% over a month if turnover keeps fading ahead of the poll
WHAT SETTLES IT ▸ in all three prior cycles the drift began only after turnover fell below its 20-session average — that crossing is the observable, and it has not happened yet
- Excise surprise45%
BATBC −6% on a larger tax step than the tape has priced, Sep 15
WHAT SETTLES IT ▸ in prior excise cycles roughly 70% of the total move landed in the five sessions before the session itself, so the approach is where the information is
- Steel margin squeeze55%
BSRM / GPH −4% as scrap costs reach first-quarter margins
WHAT SETTLES IT ▸ first-quarter gross margin is the figure that settles it; in the two prior squeezes gas relief did not offset input costs
Simulated probabilities. Each row names the observable that would resolve the risk rather than a position to take — nothing here is a recommendation.
Call Tracker
every next-session call this briefing published, scored against what happened
- Session minus 20, called UP at 82% confidence, correct
- Session minus 19, called UP at 74% confidence, correct
- Session minus 18, called DOWN at 61% confidence, wrong
- Session minus 17, called UP at 88% confidence, correct
- Session minus 16, called DOWN at 57% confidence, correct
- Session minus 15, called UP at 79% confidence, correct
- Session minus 14, called UP at 66% confidence, wrong
- Session minus 13, called DOWN at 71% confidence, correct
- Session minus 12, called UP at 58% confidence, wrong
- Session minus 11, called UP at 84% confidence, correct
- Session minus 10, called DOWN at 62% confidence, correct
- Session minus 9, called UP at 76% confidence, correct
- Session minus 8, called DOWN at 69% confidence, wrong
- Session minus 7, called UP at 91% confidence, correct
- Session minus 6, called DOWN at 73% confidence, wrong
- Session minus 5, called UP at 59% confidence, wrong
- Session minus 4, called UP at 81% confidence, correct
- Session minus 3, called DOWN at 64% confidence, correct
- Session minus 2, called UP at 77% confidence, wrong
- Session minus 1, called DOWN at 68% confidence, correct
UP at 55% confidence, scored at the close.
Calls stated at 70% confidence or above were right 82% of the time; below that line, 44%. The confidence figure is what separates them, not the wording.
A simulated record: this build has no published history, so the twenty calls are fixed rather than accumulated. The two confidence figures are counted from the cells above them.
TOMORROW — THREE SCENARIOS
BULL CASE
34%DSEX pushes through 5960 and holds it; leaders extend and laggards get dragged up late.
CONFIRMS ▸ breadth clears 2:1 · Pharma keeps its flow lead · foreign desk stays a net buyer
BASE CASE
40%A range day: the index digests while rotation does the work underneath. Stock-picking beats index-watching.
CONFIRMS ▸ breadth stays inside 1.5:1 either way · turnover holds near its average pace
BEAR CASE
26%DSEX loses 5889; money hides in defensive yield names while junk flow spikes as traders chase whatever still moves.
CONFIRMS ▸ decliners pass 2:1 · yield names take the flow lead · foreign desk flips to sell
Probabilities are computed from current breadth and the index move on simulated data, and sum to 100. They describe how this tape has resolved before — they are not a forecast, and not investment advice.