Open Market JournalOpen book
XAUUSDReviewCautious

GOLD Session Review — August 17, 2026

A Support Trap Reverses Into a Near-Miss on $4,435

Gold's session opened as forecast at the twice-rejected $4,435 shelf but did not resolve as the prep's Range/Neutral call expected: a real, displaced break of the $4,396.75 pivot at the 15:00 UTC NY overlap fully reversed within the hour, and gold ran to $4,428.69 -- within $6.31 of the ceiling -- before closing $40.02 (+0.91%) above the open, a trend-shaped day the prep's map never weighted.

Prep outcomepartial
Lead scenario40% · missed
Leanneutral · incorrect
Day typerange → trend
Surprisehigh
Grade card4 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 Third Test of Resistance Session
Symbol
XAUUSD
Window
00:00 - 23:00 UTC (Asian open through late-NY close)
Day type called
Range
Day type actual
Trend
Lean outcome
Incorrect (Neutral/Wait; session resolved directionally, +0.91% net, closed near the day's high)
Regime
Support trap that reversed into a trend-shaped rally, stalling just under the key ceiling
Preparation
Partially accurate
Surprises
High

Grade card

4 of 11 correct
  1. Day-type callIncorrect
    Called
    Range
    Actual
    Trend (open near low, close near high, $40.02 net move, 63% of D1 ATR)
  2. LeanIncorrect
    Called
    Neutral/Wait (combined same-direction weight capped at 40%)
    Actual
    Session resolved directionally bullish, closing near the day's high
  3. Conditional lean -- short trigger (held H1 close < $4,396.75)Incorrect · a live trap -- would have stopped out a short
    Called
    Confirms reject-and-fade, targets $4,376.04 then $4,351-4,343
    Actual
    Fired at 15:00 UTC (close $4,384.81, low $4,377.19 -- real displacement past the level), then fully reversed the very next hour; price never approached $4,376
  4. Conditional lean -- long trigger (held H1 close > $4,435.00)Correct · no trigger
    Called
    Confirms break-and-hold, targets $4,449.47 then $4,470-4,500
    Actual
    Never fired; session high $4,428.69 was $6.31 short of a held close at $4,435
  5. Lead scenario -- Reject-and-fade (40%)Incorrect
    Called
    Lose $4,396.75 -> retest $4,376.04 -> possible extension to $4,351-4,343
    Actual
    Trigger touched and closed, but reversed within the hour with no follow-through toward $4,376
  6. Break-and-hold (35%)Incorrect · directionally closest, not confirmed
    Called
    Clear $4,435 -> test $4,449.47 -> extension to $4,470-4,500
    Actual
    Approached to $4,428.69, no held close at or above $4,435
  7. Chop -- range holds (25%)Incorrect
    Called
    No held close outside $4,396.75-$4,435.00 through the 15:00 UTC window
    Actual
    Closed outside the band (below $4,396.75) exactly at the 15:00 UTC deadline
  8. Key level $4,435.00 (decision ceiling)Correct
    Called
    Key session decision level; held close above confirms the break
    Actual
    Held for a third straight test -- day high $4,428.69, no close at or above it
  9. Key level $4,416.37 (Asian-high pivot)Correct
    Called
    Earliest tell for the break branch
    Actual
    Reclaimed at 17:00 UTC (close $4,417.16), and the session closed almost exactly on top of it ($4,416.42)
  10. Key level $4,396.75 (fade pivot)Partial
    Called
    Fastest tell for the reject-and-fade branch
    Actual
    Broke with genuine displacement at 15:00 UTC as designed, but the break was a trap rather than a durable fade signal
  11. Key level $4,376.04 (fade continuation shelf)Correct
    Called
    Two-point pivot confirming a fade extending, not just pulling back
    Actual
    Never seriously tested after the prep's publish -- the reversal began before price reached it

The tape

  1. Asian session (00:00-07:00 UTC)

    Mostly complete by the time the prep published. Opened $4,376.40, swept down to $4,367.14 by 02:00, rebounded through $4,400 by 03:00, spiked to a session high of $4,416.37 at 04:00, then eased back to a $4,393-4,397 close by 06:00-07:00 -- exactly the "compresses, extremes act as liquidity" pattern the instrument's session clock describes.

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

    Choppy and inconclusive, ranging $4,386-$4,404 with no held break either way -- consistent with the Judas-prone, no-conviction read the prep assigned this window.

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

    A grind higher to $4,411.04 (10:00 high) faded back to $4,392-$4,400 by noon. The Empire State Manufacturing print landed at 12:30 UTC as a large beat (20.6 actual vs. 8.4 forecast, 15.6 previous) -- a print that would normally read dollar-positive and gold-negative -- but produced no immediate directional move; the 13:00 candle chopped $4,391.84-$4,405.64 and closed nearly flat at $4,398.87.

  4. NY/COMEX primary window (13:00-15:00 UTC)

    Rather than a genuine held break in either direction, price drifted down through $4,396.75 across the 14:00 and 15:00 candles, closing 15:00 at $4,384.81 with a low of $4,377.19 -- a real, displaced close below the reject-and-fade trigger, confirming the lead scenario exactly at the window's deadline.

  5. NY overlap (15:00-16:00 UTC) -- the reversal-prone window the prep flagged by name

    The fade confirmation lasted one hour. The 16:00 candle erupted from $4,384.85 to a $4,415.92 high, closing $4,408.54 -- erasing the entire reject-and-fade signal and validating the instrument's own "reversal-prone" tag on this exact window.

  6. Extension (17:00-19:00 UTC)

    The rally continued. The 17:00 close of $4,417.16 reclaimed the $4,416.37 Asian-high pivot, invalidating reject-and-fade per the prep's own rule and flipping the tape toward the break branch. The 18:00 close reached $4,427.46, and the 19:00 candle printed the session high of $4,428.69 -- $6.31 short of the $4,435 decision level, with no held close at or above it.

  7. Late session (20:00-21:00 UTC)

    Gave back nearly $30 off the highs -- a slide to $4,411.10 (20:00 low, coinciding with the 20:00 UTC TIC data) and a deeper dip to $4,398.07 (21:00 low), closing 21:00 at $4,406.56.

  8. 22:00 UTC "Asian-resume" window

    Bounced back to a $4,420.82 high and $4,419.64 close, pushing directly into the trap-prone window the prep named -- worth flagging for the next prep rather than trusting the bounce.

  9. Close (23:00 UTC)

    Settled at $4,416.42 -- almost exactly on the $4,416.37 Asian-high pivot, and $40.02 above the open.

Full notes

What We Learned

The session unfolded with a genuine surprise, not a within-parameters day. The reject-and-fade trigger fired exactly as the prep defined it -- a held H1 close with real displacement below $4,396.75 -- and still failed completely within the next candle, something none of the three mapped branches described. A trader who acted on the prep's own conditional-short rule would have been trapped and reversed against within an hour.

  1. Day-type call was wrong, and the precondition check that produced it was incomplete. The prep's Day Type reasoning weighed only Trend-vs-Range preconditions ("Trend requires prior-session compression or a fresh unpriced catalyst... none of that is present"). It never evaluated the framework's own Breakout-and-fail/trap preconditions against $4,396.75 and $4,435.00 -- both obvious, multiply-tested levels (the textbook trap magnet per the day-type taxonomy) -- and never weighted a whipsaw branch despite genuinely contradictory signals (a well-tested support with a still-intact multi-week uptrend beneath it). Routing: chronically-wrong-day-type-call category -- re-examine precondition checks, specifically add an explicit trap/whipsaw evaluation whenever a level has already rejected twice in the prior week, as $4,435 had.

  2. The reject-and-fade trigger fired but was untradable -- the same displacement discipline applied to the break branch was missing on the fade side. The break-and-hold branch had an explicit $3-15 displacement sweet spot to separate a real break from a fakeout; the reject-and-fade trigger had no equivalent filter, just a single H1 close beyond the level surviving one window. That gap is exactly what let a real, displaced close turn into a one-candle trap. Routing: right scenario, untradable trigger -- tighten the tradability standard by adding a displacement and/or two-consecutive-close requirement to fade-side triggers, mirroring the break-side rule.

  3. A large data beat produced no visible gold reaction, worth a driver-stack note rather than a rule change. The Empire State beat (20.6 vs. 8.4 forecast) is the kind of surprise that would normally support a dollar-positive, gold-negative read, yet gold rallied hard through the afternoon regardless. With no confirmed DXY or real-yield print available tonight, this can't be attributed with confidence -- flag it as a single data point on Moderate-tier Empire State prints and gold, not yet a pattern.

  4. The $4,435 ceiling and the $4,416.37 pivot both worked precisely as designed. This is worth stating explicitly alongside the misses: the level-selection layer of the prep is not the part that needs rework. The next preparation should carry that same level discipline forward while fixing the day-type and trigger gaps above.

Reviewed prep: 2026-08-17-xauusd-session-preparation