Market evidence — five facts

1. Volatility has collapsed, and both of our measures now agree on it. Average true range is at 54.5% of its 50-day norm, down from 62.4% last week. Twenty-day realised volatility is 2.13% against a 120-day norm of 6.26% — barely a third. Last week the intraday and close-to-close measures disagreed; this week the Bollinger squeeze has activated too. The disagreement we flagged last week has resolved, in the direction of more compression, not less.
2. Brent has moved up to $92–95/bbl from $89.16 a week ago, with the working range widened by a spot-versus-futures discrepancy between sources. India’s import dependence is unchanged at roughly 85%. This is the single driver with proven daily transmission into this pair.
3. The US “Economic D-Day” sanctions package on Iran is detailed today. Measures may extend to buyers of Iranian crude, China chiefly among them, which raises compliance risk for Indian refiners and could narrow the discounted-barrel options India has used to manage its import bill. Hormuz traffic remains well below normal; Gulf production is running roughly 8.3 mb/d below pre-war levels per the IEA.
4. The Fed has three dissents in favour of a hike — the first such trio since September 2016 — with rates held at 3.50–3.75% and the September 15–16 FOMC carrying the dot plot. Market pricing has shifted from expecting cuts to giving real odds to a hike. US 30-year yields sit at roughly two-decade highs.
5. The RBI is out of the picture until 7 October. Repo held at 5.25% for a fourth consecutive meeting, neutral stance. No domestic rate risk inside this window or the next.

Technical evidence — in plain terms

• The structural picture improved. Price is now above the 20-, 50-, 100- and 200-day averages, the exponential stack has turned bullish, and moving-average alignment reads STRONG_BULL against MIXED last week. Price sits 3.12% above the 200-day. This is a genuine improvement for USD/INR.
• Momentum has crossed, finally. Last week we noted the MACD histogram had narrowed for ten sessions and was close to crossing. It has crossed. MACD is now +0.018, above both zero and its signal line. The trend-strength index has flipped positive. That is the confirmation we said was pending.
• But trend strength collapsed while it happened. ADX has fallen from 25.0 to 18.07, with a ten-day slope of −8.28. Directional pressure still reads negative underneath: DI− at 30.22 against DI+ at 23.22. The trend-following stop remains above price at 96.08. Momentum confirmed the break at the precise moment the trend machinery weakened. That combination is the central tension in this week’s evidence.
• Short-term momentum is rolling over. Stochastic K at 86.14 has crossed below D at 90.07 from a high reading; Williams %R at −13.86 is near the top of its band. Three consecutive lower closes. The pair is stretched near-term even as the medium-term structure improved.
• The overhead zone is no longer thin. Last week R rested on two reference points. It now rests on nine and spans 95.75–95.96. Beneath the market the support band carries fifteen. There is still more standing under this pair than over it, but the gap has narrowed.
• Monday’s session was the narrowest in the sample. A 3.5-paise range sits at the 0.1st percentile of 880 observations. See the data-quality note in Appendix D before reading much into a single session.
• Longer-term structure remains constructive for USD/INR: +10.33% over 360 days, +5.25% over 120 days, and 87.0% of the way up the 52-week range.

Full indicator detail and level composition: Appendix A.

Fundamental evidence

Table scrolls sideways on narrow screens. Arrows are shown from the USD/INR side: ↑ means rupee weakness.

Reconciling the three views

Technical: mildly higher, with a genuine improvement in structure — all averages cleared, momentum crossed — undercut by a sharp fall in trend strength and a near-term momentum roll-over.

Fundamental: moderately bullish USD/INR. Oil is up, the Iran sanctions channel is live today, the Fed has turned hawkish and US yields are at multi-decade highs. No driver in the current set argues for a stronger rupee.

Forward market: bullish USD/INR, and more so than last week. The curve repriced upward at every tenor while spot barely moved. See Section 05 — this is the most decisive of the three streams this week and the one that changed most.

Combined: mildly INR-negative. Directional confidence Low. Range confidence High.

All three point the same way, which is the second week of alignment. But the alignment is still soft, and the reason is specific: five signal conflicts are open, up from three, and our own evidence scoring rates indicator agreement Low while data quality and historical sample both score High. Range confidence rises to High this week because compression is now confirmed by both volatility measures. Directional confidence does not rise, because the thing that improved — momentum — improved at the same moment trend strength deteriorated.

Table scrolls sideways on narrow screens.

Immediately beyond the window

Table scrolls sideways on narrow screens.

Adjusting for exposure maturity

Table scrolls sideways. Comparison column measures against the curve published on 19 August value date; only matched tenors are compared.

Where the curve crosses our markers

Last week we wrote that in forward-adjusted terms the range argument had a shelf life of roughly three weeks. That shelf life has roughly halved.

Read that carefully. An importer covering at two months is already paying above the level whose break we would call a confirmed bullish breakout. The forward market has priced past our invalidation point inside eight weeks. Beyond two months, the debate this note is having about spot direction has already been settled by the curve — and settled against the rupee.

Curve shape

The premium rises steeply from the front end — 2.11% at one week — into a broad belly peaking near 2.98% around five months, holds close to that level through nine months, then eases to 2.87% at twelve. A pronounced hump rather than a straight line, and slightly more pronounced than last week.

In practice: near-dated cover remains the cheapest annualised protection, which favours importers layering at the front end. Longer-dated receivables in the 5–9 month belly lock the richest carry, which favours exporters laddering there. The 6-month outright at 97.07 is 142 paise above spot.

Figures from the desk forward calculator, spot value date 27 August 2026, standard month-wise rates. Note the calculator’s spot of 95.66/95.67 differs from our technical close of 95.65 by roughly 1.5 paise — a quoting-convention and timing difference, not an error. Marker crossings above are computed against the ask side, which is the relevant side for an importer.

Table scrolls sideways. Directional impacts are shown from the USD/INR side: ↑ means rupee weakness.

A · Technical evidence

Market structure

• 24 Aug: O 95.6400 · H 95.6700 · L 95.6350 · C 95.6500. Change −4.5 paise (−0.047%). Opening gap −0.057%. Daily range 3.5 paise.
• Position in the 50-day range 59.7%; in the 52-week range 87.0%. 50-day high 96.6725, low 94.1350. 1.35% below the 52-week high.
• Price versus cloud: ABOVE_CLOUD. Cloud 95.4756–95.5125. Conversion 95.5112 below base 95.7963.
• Change by window: 1d −0.047% · 5d +0.050% · 15d +0.328% · 30d +0.031% · 45d +1.410% · 60d +0.684% · 90d +2.439% · 120d +5.246% · 180d +7.207% · 360d +10.329%.
• Completed sessions last week: 14.75 · 6.25 · 5.75 · 16.0 · 9.5 paise. Average 10.45 against a normal 21.0.

Trend indicators

• ADX14 18.07, labelled WEAK_TREND, slope −8.28 over 10 days (was 25.0, slope +5.56). DI+ 23.22 versus DI− 30.22.
• Price above SMA20 (95.4917), SMA50 (95.4285), SMA100 (95.0748), SMA200 (92.7598, +3.116%), EMA8 (95.6199), EMA21 (95.5868), EMA55 (95.3772), EMA100 (94.7123).
• Moving-average alignment STRONG_BULL; exponential stack bullish. Both improved from MIXED / not-bullish last week.
• Supertrend DOWN at 96.0817 X. Parabolic SAR UP at 95.1003.

Momentum

• RSI14 52.34 · RSI7 55.79 · zone NEUTRAL · no divergence.
• MACD +0.01757, above signal −0.00750; histogram +0.02507. The cross flagged as pending last week has completed. But the histogram is no longer rising.
• Stochastic K 86.14 / D 90.07 — K below D from a high reading. Williams %R −13.86 · CCI20 +65.23 · TSI +1.42 (was −4.28).
• ROC12 +0.452% against ROC30 +0.031% — aligned but flattening. Three consecutive down days.

Volatility

• ATR14 0.2103 (21.0 paise) = 54.5% of its 50-day average (was 62.4%). ATR20 0.2511.
• HV20 2.13% · HV60 4.93% · HV120 6.26%. HV20 is 34% of HV120 (was 55%).
• Parkinson 1.74% · Garman-Klass 1.80% · Rogers-Satchell 1.83%. Volatility percentile of history 32.2 (was 55.3).
• Bollinger bandwidth 0.00878, percentile 34.6, squeeze ACTIVE (was not active). Position in band 0.689.
• 1-day VaR 95: −0.548% (≈52 paise). Expected shortfall: −0.869% (≈83 paise).

Support and resistance — full zones

Technical verdict

• Trend: structurally improved — all averages cleared, alignment turned strongly bullish — but trend strength deteriorated sharply and the trend stop remains above price.
• Momentum: crossed positive, then immediately flattened. Short-term readings rolling over from high levels.
• Primary support: 95.4756–95.7032 (round number, SMA20, EMA21, cloud, mid-Bollinger).
• Primary resistance: 95.7458–95.9644 (pivots, 20-day high, Ichimoku base, upper bands).
• Confirmation of upside: two consecutive closes above 95.96.
• Invalidation: a daily close above 96.0817, or two closes below 95.4756.

B · Historical and quantitative evidence

Statistical ranges — unconditional

Based on 870–879 observations of this pair’s own excursions. These carry no directional view. The published base case of 95.46–95.86 is the 5-day typical band, in preference to the narrower conditional band below — the same methodological choice made last week.

Conditional ranges and probabilities

Regime prior WEAK_TREND (base 44%), adjusted by weighted historical evidence (60%) and current market structure (40%), rounded to the nearest 5 points. Historical tilt 0.32, structural tilt 0.45. These sum to 100% and are model estimates of historical frequency, not forecasts. The 5-point shift from base into higher is the only weight change this week.

Weighted historical follow-through — eight streams

Weighted result: 58% higher at five sessions, against 63% last week. Direct outcome sampling on the 20 closest analogues gives 60% higher at 5 days and 65% at 10 days.

Treasury excursion metrics

In comparable historical situations the pair reached the nearby adverse zone 60% of the time within five sessions, and the nearby favourable zone 20% of the time. Read from the exporter’s side, that asymmetry is the strongest single argument against waiting for R with the whole book.

C · Cross-currency context

Correlation of USD/INR to EUR/USD is 0.032 over 30 days and 0.158 over 60, against 0.896 between EUR/INR and GBP/INR. Dollar-major direction is highly relevant to the crosses and close to irrelevant for this pair.

Cross-rate integrity check: EUR/INR quoted at 111.7250 against 111.7259 implied from USD/INR × EUR/USD — a difference of 0.001%, assessed CONSISTENT against a 60-day mean difference of 0.032%. The curves are internally coherent.

Movement decomposition for EUR/INR: over five days, 94.7% of the move came from the EUR/USD leg and only 5.3% from the rupee. Treasury managers with euro exposure should note that a correct rupee call would have explained almost none of last week’s EUR/INR move.

D · Sources, gaps and method

Sources

• Desk quantitative and scenario engines v3.0 / v1.0, run 24 August at 09:15. USD/INR series 880 rows, 2 Jan 2023 to 24 Aug 2026. All integrity checks passed: no duplicate dates, no invalid OHLC, no missing values, no rows dropped.
• Desk forward calculator, standard month-wise rates, spot value date 27 August 2026.
• 30-day geopolitical and macro briefing dated 24 August 2026. Market and commodity prices within it are live fetches of 24 August; Indian cash-market rows reflect the 21 August close.
• Prior week’s published outlook, 17 August, for the scorecard in Section 01 and all week-on-week comparisons.

Data marked unavailable — flagged, not estimated

• FII / capital flows — unavailable for a third consecutive week. Excluded from the view.
• RBI FX reserves and intervention record — no figures. Our reading on policy-suppressed volatility is inference, not measurement.
• India and US government bond yields — excluded from this cycle. This matters more than usual this week, because it is the missing piece that would explain the forward-curve repricing in Section 05. No rate-differential attribution has been performed and none should be inferred.
• India monthly trade balance — the oil-to-currency mechanism is directionally certain and unquantified.
• USD/INR implied volatility and risk reversals — would settle whether the market prices this compression as directionless or skewed. Not available.
• US PCE release date — our source gives 29 August, which is a Saturday. Unresolved; see Section 03.
• Statistical envelope beyond 20 trading days — not produced by the current engine. The longer-horizon view is anchored on observed forward pricing rather than a fabricated band.

Open signal conflicts — five, up from three

Evidence quality is scored HIGH on data quality, historical sample and historical agreement, and LOW on indicator agreement. That combination is why directional confidence is Low while range confidence is High. Overall engine confidence: MEDIUM — “the evidence is usable but not uniform; treat the scenarios as guidance rather than as a firm expectation.”

Method notes

• Levels are drawn from Fibonacci retracements, moving averages, volatility bands, channels, pivots, Ichimoku, swing points and round numbers, then clustered with a 0.1148 tolerance (0.35 × average daily range). Strength is scored 0–10 on reference-point density — see the caveat in Appendix A on why that overstates reliability.
• Historical analogues use a weighted robust-z nearest-neighbour search over backward-looking features only, across 660 candidate days. Best similarity 83.3, average 74.3. Candidates must be at least 20 bars old so outcomes are complete; a ±5-day exclusion prevents one episode filling the list; no outcome value influences the similarity score.
• The Ichimoku lagging line is excluded from all signals to avoid look-ahead bias. The engine is volume-free by design.
• Seasonality is available but carries only 3–4 years per month and is rated VERY_LOW_SAMPLE. It has not been used in this note.
• No rate, level, probability or event in this report has been invented. Every figure traces to a source above.
• This is a market-timing framework, not personalised treasury advice. No hedge percentages are prescribed.