Brent surged more than 8% on the week to about $104, briefly touching $109.68 before easing below $105 on profit-taking. Attacks on tankers in the Strait of Hormuz intensified and Iran-aligned Houthis threatened Saudi exports in the Red Sea. This was the week's central variable
Dominant
Global bond yields
The US 10-year moved above 4.81%, its highest since 2023, as oil-driven inflation fears sent global yields to multi-year highs. Higher yields cut the appeal of risk assets and sustained FII outflows
Compounding
US labour data
Non-farm payrolls rose 162,000 against expectations near 53,000, with unemployment steady at 4.1%. Stronger data revived expectations of a hawkish Fed
Hawkish repricing
Rupee
Weakened to roughly ₹95.46–95.50, a fourth consecutive session of decline and its steepest weekly loss since mid-May, down about 1% on the week despite RBI intervention
Under pressure
Breadth of the selloff
Fourteen of sixteen major sectors logged weekly losses. Nifty IT fell 5.8% — its biggest weekly drop since April and the worst of any major sector. Financials lost 1.9%; Reliance, the third-heaviest Nifty weight, fell 4.9%
Broad-based
Friday's own session
Nifty Realty −2.70% and Nifty Metal −2.30% were the drags; Nifty Private Bank +0.48% outperformed, with HDFC Bank up 2.02%. The index recovered from an intraday low of 23,231.40
Late recovery
Next session
NSE and BSE are closed on Monday 14 September for Ganesh Chaturthi. The next trading day is Tuesday 15 September
Market holiday
Key insight: The character of this decline changed decisively this week. For a month the
index drifted lower on narrow weakness while domestic flows absorbed everything and small caps outperformed.
This week the selloff went broad — fourteen of sixteen sectors down, IT off 5.8%, Reliance off 4.9%, financials
off 1.9% — and the Nifty closed at its lowest level since 11 June. Five consecutive losing
weeks now total 4.77%. The cause is singular and external: Brent up more than 8% to about $104 with a spike to
$109.68, feeding an inflation narrative that pushed global yields to multi-year highs. Notably the market still
recovered 622 Sensex points off its intraday low, and private banks closed green — the bid has not vanished,
it has narrowed.
What breaks the pattern from here. The prior longest streak in this
market was six weeks, ending 3 April, when the indices fell about 12% as the US–Iran conflict began. This one is
five weeks and 4.77%. Every sector reading on this dashboard now traces back to one variable — whether Brent
holds above $100. If it eases substantially the inflation-and-yields chain unwinds quickly; if it does not, the
correction has further to run regardless of what domestic flows do.
Relative Rotation Graph
Sector performance & momentum relative to Nifty 50 · benchmark at 100/100 · NSE close 2 Sep 2026 — two sessions behind the Overview tab, see note below · rotation runs clockwise Improving → Leading → Weakening → Lagging
LeadingImprovingWeakeningLagging
Benchmark Nifty 50 as at 2 Sep · 3M +1.56% · 1M −2.85% · 1W −0.73% · hover to read, click to pin
Engine inputs · NSE closing data, 2 September 2026 two sessions behind
Sortable. RS-Ratio = 100 + (sector 3-month return − Nifty 3-month return). RS-Momentum =
100 + (sector 1-month return − Nifty 1-month return). Tail step = sector 1-week − Nifty 1-week,
which is the displacement drawn behind each node.
Sector
Symbol
Index level
RS-Ratio
RS-Momentum
Tail step
Bdry dist
Quadrant
Coverage · Ten sectors
One card per sector in the required universe. An RRG is only valid when every series comes from the same
provider priced to the same close, so a sector is either plotted on that common basis or it is not — there is
no middle state. Each card shows the figures behind its status, with the date and provider for anything partial.
The engine basis is now seven sessions stale — read this first. The multi-horizon sector series behind this chart is priced to the 2 September close and has not refreshed. Since then the Nifty has fallen from 23,914.45 to 23,398.10 — a drift of 516.35 points, or 2.16%, with the week of 7–11 September alone accounting for 2.09%. That is no longer a rounding difference: a move of this size over a week in which fourteen of sixteen sectors fell, IT dropped 5.8% and Reliance lost 4.9% will have moved these quadrants. Treat every position below as a 2 September reading, not a current one. Re-dating the axes without re-sourcing all eight series would be fabrication, and splicing an 11 September benchmark onto 2 September sectors would corrupt the comparison outright — so the basis is held and stamped instead. Verified 11 September sector moves are reported on the Overview tab and in the ledger, deliberately not merged into these axes.
Methodology. This is the canonical Julius de Kempenaer construction: two normalised
series centred on 100, with RS-Ratio on the horizontal axis and RS-Momentum on the
vertical, and the conventional colour scheme — green Leading, blue Improving, amber Weakening,
red Lagging. Rotation runs clockwise: Improving → Leading → Weakening → Lagging.
On the tails. A true RRG tail is drawn from a daily RS series. Only summary period
returns are published for these indices, so each tail here is a two-point derivation —
the current node plus its position approximately one week earlier, backed out from the
one-week excess return. It shows direction of travel over the last week and nothing finer.
It is not a full daily tail and should not be read as one.
A better source has been identified for the next rebuild. The Government of India's DIPAM Market Monitor (dipam.gov.in), published daily and sourced from Bloomberg, BSE, NSE, RBI, FBIL, CDSL, NSDL and PIB, carries a full sector table that does include Nifty Healthcare, Nifty Metal and Nifty Oil & Gas — the three sectors absent here — alongside FII/DII flows, the repo rate, INR/USD and the Brent futures curve. It publishes point-in-time closes rather than period returns, so the engine can be rebuilt on it by pulling three editions and computing the returns directly. The most recent edition indexed at compile time was 11 August, which is why it could not be used for this build.
Sectors not on the chart, and what was tried. Nifty Energy, Nifty Metal and Nifty Healthcare are absent. The multi-horizon source carries ten NSE indices and does not publish the 1W/1M/3M triple for those three. TradingView was fetched directly this run for CNXMETAL and CNXENERGY — its symbol pages render quote values client-side, so the response returns "Market closed / No trades" with every numeric field empty. It is usable for constituent lists, never for figures. NSDL blocks automated access, the CDSL mirror resolves only historical fortnights, and there is no RSS feed for NSE sectoral index returns. The partial data that does exist — Metal 13,193.90 on 1 Sep, Healthcare 16,552.20 on 3 Sep with 1W −0.93% and 1M −1.38% from a different provider — carries no 3-month figure on the common basis, so neither can be placed on the x-axis at all.
Migration. Week-on-week quadrant migration is still not shown. The measurement basis
has changed again this edition — from percentage-point excess to the 100-centred JdK scale —
so a comparison against the prior edition would be measuring two different things. It returns
once two consecutive editions share this basis.
Horizon
Horizon matrix 2 Sep basis
Every tracked index across all three horizons, with the benchmark row pinned. Colour is scaled to the absolute
return so the eye reads magnitude and sign together; the selected horizon column is emphasised. Sortable.
Index
Symbol
Level
Stocks
1 Week
1 Month
3 Month
Excess return over the benchmark
Diverging bar — the correct chart when values carry a sign and rank matters. Shared zero axis, sorted by
magnitude. Each bar is sector return minus Nifty return over the selected horizon, so zero means
"performed exactly like the index".
Session strip · breadth, flows and volatility
The three readings that describe market structure rather than direction. Absorption is
DII buying ÷ absolute FII selling — above 100% means domestic institutions more than covered the
foreign sale.
Every widget on this tab is driven by the same eight verified series plus the benchmark — no separate or
decorative datasets. Percentages are as published by the exchange data source for 2 September 2026; excess
returns and absorption ratios are computed from those figures and the formulas are stated above each chart.
Nifty 50 · the five-week drawdown
Area chart — the correct form for a level moving through time. The shaded region is the distance below the
7 August peak, so the depth of the drawdown is read directly rather than inferred from a table. Nine verified
closes; the gap between 4 and 10 September is drawn as a break, not interpolated, because 8 and 9 September
closes were never individually published.
Weekly change · waterfall
Waterfall — built for showing how signed period changes accumulate to a total. Each bar starts where the last
ended, so the compounding is visible. The five weekly moves compound to exactly −4.77%, which matches the direct
peak-to-close calculation.
Institutional flows · FII against DII
Back-to-back bars on a shared zero axis, because the two flows are opposed and their relative size is the point.
The line is the absorption ratio — DII buying divided by absolute FII selling — with the 100% threshold marked.
Sessions where FIIs bought have no ratio and the line breaks.
India VIX
Line chart with the cycle low marked. The final point is drawn hollow because sources state only that VIX rose
above 12 on 11 September — no exact close was published, so 12.00 is a floor, not a reading.
Sector moves · verified, by horizon
Diverging bars from a shared zero, sorted. Only five sector figures are verified for this period and they span
two different horizons, so each bar is labelled with its own — mixing a weekly and a daily move on one unlabelled
axis would be misleading.
Underlying tables
Charts show the shape; these let you check every number behind it.
Date
Nifty 50
Change
%
Confirmed in
Week ending
Close
Change
%
Direction
Date
FII cash
DII cash
Absorption
Note
Date
VIX
Change
Note
Two things this edition resolves.
(1) The 2 September flow figure was right, and I was wrong to exclude it. Three earlier editions withheld
a reported FII inflow of ₹6,688.37 crore for 2 September because only one outlet carried it and that outlet had
a documented error elsewhere. A weekly review has since published the full day-by-day series for 31 August to
4 September — −₹7,985.88, +₹1,143.38, +₹6,688.37, −₹2,345.87, −₹3,111.94 — which sums to −₹5,611.94 crore
and matches the independently reported weekly total exactly. The figure is corroborated and is now in the table
above. Caution was right at the time; the exclusion is now wrong, so it has been reversed.
(2) The losing streak is now genuinely five weeks. A prior edition corrected an outlet that called the
week to 4 September the "fifth consecutive" decline — it was the fourth, because the week to 7 August was a
0.77% gain. That correction still stands for that edition. With the week to 11 September down 2.09%,
the count has legitimately advanced to five, totalling 1,172.55 points or 4.77% since 7 August. The
weekly table above shows every step, so the distinction is checkable rather than asserted.
Still unverified: a precise India VIX close for 11 September. Sources state only that it rose more than
4% and moved above 12; the table carries 12.00 as a floor with that caveat rather than a false precision.
Sessions 7, 8 and 9 September are not individually in the ledger because only the weekly aggregate and the
Monday close (23,779.15) were published.
Options positioning · 11 September
A strike ladder showing where writers are committed. The highest Call open interest marks working resistance,
the highest Put OI working support, and the band between them is the range the writing community is defending.
Friday's intraday low is drawn in because of what it did to that band.
Highest Put OI · 138.02 lakh contracts · only 98.10 pts below spot
Put wall vs Call wall
1.01×
Put OI is 1.0% heavier — writers are defending the floor slightly harder than they are capping the ceiling
PCR
~1.05–1.06
Two sources agree; a third reading of 1.30 is from a different expiry and is excluded
52-week position
29.0%
−11.28% from the 26,373.20 high; +5.48% above the 22,182.55 low
The most important thing the option chain says. Friday's intraday low of 23,231.40 fell
68.60 points below the 23,300 put wall — the level writers are most committed to — and the index
then closed back above it at 23,398.10, a recovery of 166.70 points off the low. That is the defended floor being
tested and holding within a single session. It also explains the shape of the recovery: the bid that appeared was
structural, coming from writers protecting a position, rather than a change of view. The band being defended runs
23,300 to 24,000, just 3.00% wide, and spot sits at the bottom of it.
Institutional flow · confirmed sessions
NSDL fortnightly sector flows · 1–15 August 2026
Net FPI investment by sector, the only dataset that shows foreign positioning at sector level rather than in
aggregate. Diverging bar with a shared zero axis. The residual is the gap between the thirteen named sectors and
the published total, shown explicitly rather than absorbed.
Where the two datasets disagree — and why that matters. NSDL's fortnightly data shows foreign
money moving into Financial Services (+₹6,535 Cr) and Automobiles (+₹4,405 Cr) during the first half of
August, and out of Telecom, Capital Goods, Power and Realty. But the rotation engine, built on price,
puts Realty at the top of three-month relative strength and Auto in the Weakening quadrant. Flows and prices are
measuring different things over different windows: foreign investors were selling Realty into a rally, and buying
Auto into what has since become underperformance. Neither dataset is wrong; the disagreement is the signal.
On the 2 September gap: confirmed FII/DII cash figures for the session were not available from a source that
also reported the session correctly, so they are excluded — see the integrity note on the Overview tab. The last
fully confirmed session remains 28 August. The NSDL fortnight covers 1–15 August; its thirteen named sectors sum
to +₹15,000 Cr against a published total of +₹16,618 Cr, leaving ₹1,618 Cr across sectors not individually
reported, shown above as its own bar.
◆Candidate listing only. Names surfaced by the rotation engine and flow data.
Contains no entry levels, targets, stop-losses, position sizing or buy/sell calls — names to research,
not trade instructions.
On coverage: cards are anchored to the sector's position in the rotation engine, which is computed from
three months of index data rather than a single session. Stock-level moves cited are from the most recent session
in which that name was verifiably reported, and the date is shown. Brokerage views are reported as facts about
what analysts said, never carried across as recommendations. Names without a verified figure carry an amber tag
and show no number.
Quadrant census
What changed this edition
Change
Detail
Status
Rotation engine rebuilt
Momentum axis moved from a single session to 1-month excess return; relative strength axis now 3-month. Velocity vectors, boundary flags, click-to-isolate and NSE ticker symbols added
New
Analytics tab added
Horizon matrix, diverging excess-return bar, session strip and VIX trajectory — all driven by one shared verified dataset with a single horizon toggle
New
Sector coverage
Eight sectors now carry full multi-horizon data where prior editions had one verified magnitude. Seven further sectors are absent from the source and omitted rather than estimated
Improved
Session updated
Overview and Smart Money moved to the 3 September close: Nifty 23,873.45 (−0.17%), Sensex 76,152.86 (−0.55%, lowest since 24 July), FII −₹2,346 Cr against DII +₹4,978 Cr
Current
Two-date disclosure
The rotation engine and analytics remain on the verified 2 September basis because the multi-horizon source had not refreshed. Both tabs are explicitly labelled rather than silently mixing dates
Flagged
Questions worth asking
Pro tip: The most useful reading on this dashboard is the disagreement between the two axes.
Realty leads three-month relative strength by 12.57 pp, yet its one-week velocity is −1.08 pp. Auto sits 4.68 pp
ahead over three months but −2.39 pp behind over the past week — the widest negative velocity on the board, and
it has already crossed into Weakening. IT is the mirror image: +5.40 pp over three months and +3.65 pp over one
week, the only sector where both horizons agree strongly. When long-horizon strength and short-horizon velocity
point opposite ways, the velocity is the newer information — but it is also the noisier. That tension is exactly
what the boundary rings are for: Auto and Infra are both within 1.00 pp of changing quadrant.