Open Market JournalOpen book
XAUUSDReviewCautious

GOLD Session Review — August 28, 2026

Reclaim Fake-Out Gives Way to the Sharpest Breakdown of the Corrective Leg

Gold nearly completed Thursday-Friday's reclaim attempt -- tagging $4,631.25, just above the $4,629.53 confirmation level -- before reversing inside the same hour into a $150, 6.5x-H4-ATR collapse that closed the day at $4,454.44, deep in the mid-$4,400s the prep's breakdown-continuation path had named as its extended target. The prep's primary 13:00-15:00 UTC NY decision window resolved as neither branch's trigger described it -- directionless chop across the reclaim level -- and the real move fired five hours later, inside the window the prep had explicitly labeled management-only, with no confirmed tier-1 catalyst in the calendar feed. Carry-forward: treat 17:00-21:00 UTC as a live decision zone on days following a failed reclaim, and pair any NY-window trigger with an explicit late-fire fallback.

Prep outcomepartial
Lead scenario45% · hit
Leanneutral · incorrect
Day typetrend → trend
Surprisehigh
Grade card5 of 11 correct

Every prep call is graded like this. See the full track record →

Session chart
XAUUSD — session vs prep levelsH1
Shaded band: the reviewed session. Price lines: the preparation's key levels — see which held and which broke.
Powered by Cortiq

Graded against live session data from Cortiq

Every call in this review is checked against the session data captured in Cortiq — the AI trading workspace that powers our daily preparation, review, and calibration loop.

See how Cortiq works
Session
GOLD Reclaim Fake-Out to Breakdown Session
Symbol
XAUUSD
Window
21:00 UTC (Aug 27) – 22:00 UTC (Aug 28)
Day type called
Trend (bearish continuation), tentative
Day type actual
Trend -- decisive same-session breakdown, largest range day of the corrective leg
Lean outcome
Incorrect
Regime
Reclaim attempt reversed into a $150 breakdown
Preparation
Partially accurate
Surprises
High

Grade card

5 of 11 correct
  1. Day-type callCorrect
    Called
    Trend (bearish continuation), tentative
    Actual
    Trend -- decisive same-session breakdown, the leg's largest range day
  2. LeanIncorrect
    Called
    Neutral/Wait (conditional short below $4,565.13 / long above $4,594.52, both NY-window-held)
    Actual
    Session resolved sharply directional -- D1 close $148.65 below the open
  3. Conditional lean -- short triggerCorrect · no trigger within the stated window
    Called
    Held, displaced H1 close below $4,565.13 surviving the 13:00-15:00 UTC NY window
    Actual
    Never touched $4,565.13 in that window (session low there was $4,588.95); the level didn't break until 18:00 UTC, five hours later
  4. Conditional lean -- long triggerPartial
    Called
    Held, displaced H1 close above $4,594.52 surviving the 13:00-15:00 UTC NY window
    Actual
    Price traded above $4,594.52 for most of 09:00-17:00 UTC but chopped either side of it inside 13:00-15:00 itself ($4,600.02 / $4,593.27 / $4,601.61) -- never one clean held, surviving close
  5. Lead scenario -- breakdown continuation (45%)Correct · (path/target); wrong trigger mechanism and window
    Called
    Trigger: NY-window close below $4,565.13; path: break → $4,540.74 → mid-$4,400s if decisive
    Actual
    Trigger fired late (18:00 UTC close $4,524.39), outside the stated window, via an unmapped late-session reversal; final path matched almost exactly -- close $4,454.44, low $4,445.24, squarely in the mid-$4,400s target zone
  6. Reclaim -- overnight break rejected (30%)Note · Contradicted (was live, then failed)
    Called
    Held NY-window close above $4,594.52 → $4,605.14 → $4,629.53
    Actual
    Nearly completed: price cleared $4,605.14 for hours and tagged $4,631.25 -- through the $4,629.53 confirmation level -- at 17:00 UTC, then reversed within the same hour
  7. Range / two-sided chop (25%)Incorrect
    Called
    No held, displaced close outside $4,565.13-$4,594.52 through 16:00 UTC
    Actual
    Price left the band to the upside for roughly eight hours (09:00-17:00 UTC)
  8. Key level $4,629.53 (2nd bullish confirmation)Correct · held as resistance
    Called
    Expected to confirm reclaim on a break-and-hold
    Actual
    Tagged intrasession (high $4,631.25) but rejected within the same hour
  9. Key level $4,594.52 (primary reclaim trigger)Partial · a real bullish tell for hours, then failed
    Called
    Earliest bullish tell on a held close above
    Actual
    Held as resistance-turned-support for roughly eight hours, then broken decisively in the reversal
  10. Key level $4,565.13 (primary continuation trigger)Correct
    Called
    Held close below confirms the breakdown
    Actual
    Broken at 18:00 UTC; the breakdown accelerated immediately after
  11. Key level $4,540.74 (first target)Incorrect · magnitude underestimated
    Called
    "Not expected to break on a single session"
    Actual
    Broken by roughly $95 -- session low $4,445.24

The tape

  1. Overnight / Asian (21:00-07:00 UTC, already unfolded by the time the prep was written)

    Opened $4,603.09, tested up to $4,611.22, then broke down through $4,594.52 and $4,582.92, printing a fresh multi-week low of $4,574.93 by 05:00 UTC before drifting to a session low of $4,571.23 by 08:00 UTC. This matched the prep's own account of the overnight break precisely.

  2. London open (07:00-09:00 UTC)

    The 08:00 UTC hour closed at $4,573.23, still below every mapped resistance. Then, inside the 09:00 UTC candle, price reversed hard -- low $4,572.21, high $4,613.04, close $4,612.53 -- a single-hour, displaced reclaim of both $4,594.52 and $4,605.14 that the prep had flagged as Judas-prone but not this violent.

  3. Pre-NY drift (09:00-13:00 UTC)

    Consolidated above the reclaim trigger, $4,600-$4,614, holding the reclaim gains through Canada's GDP print (12:30 UTC, not a gold driver).

  4. 13:00-15:00 UTC NY/COMEX -- the window the prep named as the decision point

    Chopped $4,588.95-$4,609.71 across the $4,594.52 level, closing 4,600.02 / 4,593.27 / 4,601.61 in successive hours -- neither the breakdown nor the reclaim trigger held a clean, surviving close. The 14:00 UTC Michigan Sentiment release (in line to slightly firmer, softer inflation expectations) produced no visible directional break.

  5. 15:00-16:00 UTC NY overlap

    Extended the reclaim -- high $4,627.10, close $4,605.18 -- approaching the $4,629.53 second confirmation level.

  6. 17:00 UTC -- the session's real pivot

    Spiked to $4,631.25, tagging through the $4,629.53 reclaim confirmation, then reversed inside the same hour to a low of $4,529.87, closing at $4,576.61 -- a single-candle round trip from confirmed-reclaim territory to just above the breakdown trigger.

  7. 18:00-21:00 UTC -- the window the prep labeled management, not fresh-entry

    The breakdown trigger broke on the 18:00 UTC close ($4,524.39), and the decline continued without pause -- $4,479.24 (19:00), $4,469.24 (20:00), $4,456.80 (21:00) -- printing the session low of $4,445.24 inside the 21:00 UTC hour.

  8. Late / close (22:00-23:00 UTC)

    Stabilized $4,452-$4,467, closing the D1 session at $4,454.44 -- $148.65 below the open and roughly $95 through the $4,540.74 level the prep had called unlikely to break intraday.

Full notes

What We Learned

Surprises: The dominant surprise was magnitude and timing, not direction. A near-completed reclaim (tagging through the second confirmation level) reversed into a $150, ~6.5x-H4-ATR collapse inside a single hour, with the bulk of the move firing in the 18:00-21:00 UTC window the prep had explicitly reserved for position management rather than fresh entries. No tier-1 event in the calendar feed lines up with the 17:00 UTC pivot or the acceleration that followed; the only ambient risk the prep had named for that stretch was the dateless, ongoing Jackson Hole Economic Symposium, which it correctly flagged as unscheduled headline risk (No-Trade Condition 5) but could not pin to a trigger time. This was not foreseeable with better level work -- the map's structure was sound -- but it was foreseeable that a failed reclaim this close to a confirmation level, sitting inside an active symposium window, carried live continuation risk beyond the stated "management" hours.

  1. Right scenario, untradable trigger → tighten the tradability standard. The breakdown-continuation branch's ultimate path (break → $4,540.74 → mid-$4,400s) is exactly what printed, but its trigger was written narrowly to a single window (13:00-15:00 UTC NY-held close). When that window resolved as contradictory chop instead, the branch had no fallback trigger for a delayed confirmation -- by the letter of the prep, the real break, when it came five hours later, was untradable. Fix: pair any NY-window trigger with an explicit late-fire fallback ("if the window contradicts the lead, re-arm the same trigger on the next H1 close through the level, without requiring window-survival, once price has already changed sides once").

  2. Right scenario, untradable trigger → widen the late-session window on post-reclaim-failure days. The 19:00-21:00 UTC bucket is labeled management-only per the standing priors; today it hosted the single largest continuation leg of the session. Propose treating any GOLD session that (a) sits inside an active, dateless Jackson Hole/FOMC-adjacent symposium window and (b) has just failed a near-complete reclaim attempt, as carrying live decision risk through the late-session bucket rather than management-only by default.

  3. Tighten the tradability standard on target sizing. $4,540.74 was sized as "not expected to break on a single session" without an explicit ATR-multiple caveat; the session traveled roughly 6.5x live H4 ATR intraday. Any target beyond ~2x live H4 ATR from the anchor level should carry an explicit "extends past this level on a greater-than-3x-ATR day" caveat instead of an unqualified containment claim.

  4. Driver-stack read was sound discipline, not the failure point. Three of four drivers were silent (no fresh confirmed read) ahead of the primary window, and that correctly capped the lead at 45% rather than an outright lean -- the map's caution was procedurally right given the available inputs. The lesson is not to change the driver stack; it is that a fully silent driver stack ahead of an active, dateless symposium window is itself a signal of pending uncaptured event risk, and should tighten the no-trade window around those specific hours rather than being folded into generic "ambient noise" framing.

Reviewed prep: 2026-08-28-xauusd-session-preparation