4 of 5 timeframes falling — aligned down
The 15m, 30m and 4H charts agree with the daily downtrend; the 1H chart is sideways.
Participation: 1.57× usual tick volume · rising · OBV divergence: bullish · absorption bar · money flow +0.196 — single-broker tick volume; participation, not direction
Volume–price read: latest bar wide spread on high volume (1.79× usual) — effort matches result (validation) · stopping volume · falling price on rising volume — participation confirms the move — Coulling bar grammar, our thresholds (unmeasured); participation, never direction
A price move needs participation behind it before it deserves trust. A big price move on unusually thin volume is the classic trap signature, while heavy volume with little price progress suggests someone large is quietly absorbing what is offered. FX has no central tape, so volume here is a single broker’s tick volume — the only honest option — used for consistency, not precision. Book-derived hypotheses, unmeasured locally; participation context, never a direction.
Bearish read on USDJPY: down trend with strength 0.42 (ADX 42). Backtested hit rate for this signal grade is about 49% over 5 days.
Path: choppy — direction unreliable (R² 0.003). Stretched: today is 10× its typical daily move. USD/JPY leans down today mostly on JPY strength (+1.61%) — USD -0.63% vs JPY +1.61% across their crosses.
Yellow cone: the ±1σ expected range fanning forward from now, widening with √time over the next 5 sessions (from realized volatility). It answers “where” as a range, not a direction — about 1 in 3 moves ends OUTSIDE it, so it is a range to study, not a boundary or a target.
50 recent stories about this market this week — tone is mixed / neutral.
Unusually heavy news flow this week — more stories than normal.
Negative-toned coverage has persisted across the week.
News describes the mood of coverage, not a buy or sell call.
News measures coverage tone, not price direction. News tone and event context only — not a trade signal. Tone measures how stories read, not what price should do.
USD/JPY: Trending (ADX 42). Likely range over the next 5 sessions is about 153.84–158.25. Bearish lean, low conviction. This is context, not a prediction.
Measured context, not a price prediction or order. You always decide.
Watch: A research setup exists, but its trigger or evidence gates are incomplete. Strategy: Slow Fx Mean Reversion. Net-positive-return probability: Not established.
Bearish read on USDJPY: down trend with strength 0.42 (ADX 42). Backtested hit rate for this signal grade is about 49% over 5 days.
Reasons it could go up: Confirmation improves if USD/JPY holds above near-term support while dependency conflicts fade.
Reasons it could go down: The setup weakens if macro drivers reverse or price closes through the invalidation level.
Chart read changes if: Wrong if price closes above 159.002 (2x ATR) or the trend gate drops.
If USD/JPY reached 159.0019, the idea is wrong and you would be out. This works out what that costs on your account. Nothing here is a recommendation to take the trade — it is the price of being wrong.
Run down the checklist before you act on USD/JPY. This is discipline, not a signal — a green gate means you aren’t forcing the trade, not that it will win.
0 of 2 checks — this is a B-setup. The edge is in passing on these. Trades are like buses; another comes along.
Continuation
Invalidation
Bearish read on USDJPY: down trend with strength 0.42 (ADX 42). Backtested hit rate for this signal grade is about 49% over 5 days.
A research setup exists, but its trigger or evidence gates are incomplete.
4 of 5 timeframes falling — aligned down
The 15m, 30m and 4H charts agree with the daily downtrend; the 1H chart is sideways.
Price chart1D · Trending (ADX 42)
Why the agents think this
Confirmation improves if USD/JPY holds above near-term support while dependency conflicts fade.
The setup weakens if macro drivers reverse or price closes through the invalidation level.
Wrong if price closes above 159.002 (2x ATR) or the trend gate drops.
Bias and directional model score now come from the deterministic advisor read. Evidence status separately controls whether any setup is actionable.
What might happen next
Price respects the current regime and dependency confirmation remains intact.
Wrong if price closes above 159.002 (2x ATR) or the trend gate drops.
Want every agent’s score, key support/resistance levels, and the cross-market signals behind this read? Switch to Pro in the top bar for the full desk.
AI analysis to study, not orders to follow. You always decide — and never risk money you can’t afford to lose.
If USD/JPY reached 159.0019, the idea is wrong and you would be out. This works out what that costs on your account. Nothing here is a recommendation to take the trade — it is the price of being wrong.
Run down the checklist before you act on USD/JPY. This is discipline, not a signal — a green gate means you aren’t forcing the trade, not that it will win.
0 of 2 checks — this is a B-setup. The edge is in passing on these. Trades are like buses; another comes along.
The current rise has lasted longer than 95% of this market’s past moves — old by its own history, though that alone predicts nothing.
Why this matters: moves made in the market’s busiest hours carry more information than moves in the quiet stretches, and an old move usually has less of its typical life left — knowing the rhythm helps you avoid mistaking noise for meaning.
Descriptions of the market’s rhythm from trading books — not advice to buy or sell.
a regular session is open, outside the London hours
Older than 95% of past moves.
Price edged just below a prior low and slipped back (a 2B probe).
Coverage incomplete: Nikkei, spx futures unavailable.
Session liquidity: WHEN a move happens matters as much as the move. The hours when London and New York are both open carry roughly half of daily FX volume, so moves there carry real information — moves in the quiet late-New-York lull are usually noise.
Move maturity:compares the current run with this market’s own past moves. Old, stretched moves have historically had less room left, and chasing them is the most common beginner mistake.
False breaks: when price breaks a level everyone watched and then snaps back, the traders who chased the break are trapped on the wrong side — classic trading books rate failed breakouts as more informative than successful ones.
Driver tension: some related markets tend to move first (oil for the Canadian dollar, gold for the Australian dollar). A related market moving hard without this pair following is worth attention — and it can resolve in either direction.
WHEN a move happens matters as much as the move itself. The London–New York overlap carries roughly half of all daily FX volume, so moves made there involve real participation — while the late-New-York lull is so thin that moves there are usually noise. Prices drifting into the 4pm London fix, especially at month-end, tend to reflect scheduled rebalancing flows rather than fresh opinion. One caution from our own measurement: the much-cited “reversal after the fix” was real in 2015–2018 data but has faded to statistical zero since 2019 — treat the fix as a flow window, not a direction signal.
When price breaks a level everyone was watching and then snaps back, the traders who chased the break are trapped on the wrong side. That is why several classic trading books rate failed breakouts as more informative than successful ones. Each event starts unconfirmed and is later graded confirmed or failed — forward only, never rewritten with hindsight.
This compares the current run with this pair’s own completed moves — like asking how old a trend is relative to its life expectancy, a veteran trader’s insurance-actuary analogy. Old, stretched moves have historically had less room left, and chasing them is the most common beginner mistake. An old move is a caution note about what may be left, not a prediction that it ends here.
Some markets tend to move first: gold and copper for the Australian dollar, oil for the Canadian dollar, US yields for the yen pairs. When a driver moves hard and this pair has not followed, that gap is worth attention — though it can close in either direction. Legs whose live correlation has broken are muted rather than pretended, because these relationships come and go.
Book-derived structure context; descriptive only — no bias, no confidence, no gate; thresholds are hypotheses pending local measurement. Context to study, never a signal.
Asian hours (00:00–07:00 UTC) typically cover 51% of the day's high–low range (95% range 50%–52%, 2999 days); a flat 24-hour clock would give 51%. Quietest weekday: Monday (median range 0.83× ATR vs 0.94× on Wednesday); Tuesday–Thursday run 1.05× Monday/Friday (95% range 1.02–1.09). None of this says which way it will go.
| UTC hour | Under house cost | 95% range | Under measured spread |
|---|---|---|---|
| 00:00 | 37% | 35%–39% | 8% |
| 01:00 | 46% | 44%–48% | 10% |
| 02:00 | 57% | 55%–59% | 14% |
| 03:00 | 64% | 63%–66% | 17% |
| 04:00 | 62% | 60%–64% | 16% |
| 05:00 | 59% | 57%–61% | 14% |
| 06:00 | 50% | 49%–52% | 12% |
| 07:00 | 43% | 41%–45% | 10% |
| 08:00 | 46% | 44%–48% | 10% |
| 09:00 | 52% | 50%–53% | 12% |
| 10:00 | 55% | 53%–57% | 14% |
| 11:00 | 54% | 53%–56% | 13% |
| 12:00 | 41% | 39%–43% | 10% |
| 13:00 | 37% | 36%–39% | 8% |
| 14:00 | 35% | 33%–36% | 7% |
| 15:00 | 39% | 38%–41% | 9% |
| 16:00 (now) | 51% | 49%–53% | 12% |
| 17:00 | 56% | 55%–58% | 14% |
| 18:00 | 57% | 55%–59% | 14% |
| 19:00 | 62% | 61%–64% | 17% |
| 20:00 | 70% | 68%–71% | 20% |
| 21:00 | 78% | 77%–80% | 25% |
| 22:00 | 68% | 66%–70% | 20% |
| 23:00 | 57% | 56%–59% | 14% |
| NY hour | Median spread (pips) | bp of mid |
|---|---|---|
| 00:00 | 1.6 | 1.00 |
| 01:00 | 1.6 | 1.00 |
| 02:00 | 1.6 | 1.00 |
| 03:00 | 1.6 | 1.00 |
| 04:00 | 1.6 | 1.00 |
| 05:00 | 1.6 | 1.00 |
| 06:00 | 1.6 | 1.00 |
| 07:00 | 1.6 | 1.00 |
| 08:00 | 1.6 | 1.01 |
| 09:00 | 1.6 | 1.01 |
| 10:00 | 1.6 | 0.99 |
| 11:00 | 1.6 | 0.99 |
| 12:00 | 1.6 | 0.99 |
| 13:00 | 1.6 | 1.00 |
| 14:00 | 1.6 | 1.00 |
| 15:00 | 1.6 | 1.00 |
| 16:00 | 1.8 | 1.12 |
| 17:00 | 6.1 | 3.82 |
| 18:00 | 1.9 | 1.19 |
| 19:00 | 1.7 | 1.06 |
| 20:00 | 1.6 | 1.01 |
| 21:00 | 1.6 | 1.01 |
| 22:00 | 1.6 | 1.00 |
| 23:00 | 1.6 | 1.01 |
| Horizon | Days | 95% band covered | 95% range | Calm-σ̂ third | Wild-σ̂ third |
|---|---|---|---|---|---|
| 1d | 2993 | 93.5% | 92.5%–94.3% | 91.8% | 94.4% |
| 5d | 2989 | 92.6% | 91.6%–93.5% | 88.5% | 95.1% |
| 20d | 2974 | 92.7% | 91.7%–93.6% | 89.2% | 95.5% |
Measured session, cost and coverage facts for this pair. They say when and how much it typically moves — never which way. Nothing here is a call, a bias or a confidence; past behaviour does not guarantee future behaviour. Corpus: OANDA M15 mid bars, complete, 2015-01-01 → 2026-08-02 (docs/experiments/.cache-m15-trend-onset). Notes: pair-vol-meters, oanda-ba-spread-probe, asian-range-width-day-range, london-ny-sign-relation, interval-coverage-crps, yen-cross-session-comovement, yield-change-attribution, levels-vs-returns-correlation-audit.
A research setup exists, but its trigger or evidence gates are incomplete.
Slow Fx Mean Reversion · exploratory
Not established
Not established
carry, macro, bidAsk, events, independent holdout, forward paper, net expectancy after costs
Confirmation improves if USD/JPY holds above near-term support while dependency conflicts fade.
The setup weakens if macro drivers reverse or price closes through the invalidation level.
Wrong if price closes above 159.002 (2x ATR) or the trend gate drops.
Bias and directional model score now come from the deterministic advisor read. Evidence status separately controls whether any setup is actionable.
If USD/JPY reached 159.0019, the idea is wrong and you would be out. This works out what that costs on your account. Nothing here is a recommendation to take the trade — it is the price of being wrong.
Run down the checklist before you act on USD/JPY. This is discipline, not a signal — a green gate means you aren’t forcing the trade, not that it will win.
0 of 4 checks — this is a B-setup. The edge is in passing on these. Trades are like buses; another comes along.
Institutional analytics
Market detail views surface data traceability and scenario context for the selected symbol.
Exposure, leverage, concentration, and drawdown analytics are built on versioned market data. Values appear only when they trace to a trusted dataset — nothing is estimated for display.
Factor exposures and stress scenarios are computed only where a measured data source exists, and every shock definition is stated explicitly.
Spread, slippage, and other cost analytics stay educational until measured cost evidence passes deterministic checks.
Correlation regime: Calm.
Cross-market inputs are mixed for USD/JPY; confidence remains near the base agent view.
Cross-market adjustment to the model score: -3 pts
Directional model score across recent runs. This is not a probability of profit.
| R3 | 163.73 | H4 | 160.90 |
| R2 | 161.33 | H3 | 159.91 |
| R1 | 160.13 | P 157.74 | |
| S1 | 156.53 | ||
| S2 | 154.14 | L3 | 157.93 |
| S3 | 152.93 | L4 | 156.94 |
Source: USDJPY=X via Yahoo chart API (daily, 5y). Descriptive measurements, not signals or advice.
Average percent move of each currency against its crosses. Descriptive, not a signal. Strongest-vs-weakest is where a pair's direction usually comes from.
| Agent | Bias | Model score | Assessment | Key Levels |
|---|---|---|---|---|
| Technical agent | bearish | 49% | USD/JPY technical structure is bearish inside a Trending (ADX 42) regime.Invalidation: 1H close above 156.98ATR 0.031⚠ event or volatility risk | S 155.1 / 154.64R 156.98 / 157.6 |
| News / sentiment agent | bearish | 41% | USD/JPY macro tone is primarily driven by dollar strength, event risk, and rate expectations.conf adj -0.070 | — |
| Regime / correlation agent | bearish | 47% | Measured moderate positive correlation with DXY (0.586); regime classified as Trending (ADX 42).regime Trending (ADX 42)stability 0.715corr regime calmrolling ρ 0.586driver DXY | — |
| Risk agent | neutral | 62% | Position sizing should be reduced until event and volatility risk normalizes.disagreement 0.506⚠ reduced size⚠ monitor event window | — |