GOLD Session Review: September 29, 2026, The Floor Held but the Short Lean Failed
Gold held the 4,110.50 floor and closed 56.49 higher at 4,174.70, turning a range call and short lean into a partial preparation grade. The main gap was a bullish trend day that no scenario mapped, even though the preparation correctly identified the floor, the delayed New York decision window, and the need to avoid the first reaction to the 14:00 UTC data.
Every prep call is graded like this. See the full track record →
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.
- Session
- GOLD Floor Hold and Bullish Range Traverse
- Symbol
- XAUUSD
- Window
- 00:00 to 23:59 UTC
- Day type called
- Range
- Day type actual
- Trend
- Lean outcome
- Incorrect
- Regime
- Bullish trend from support, close near the session high
- Preparation
- Partially accurate
- Surprises
- Moderate
Grade card
8 of 12 correct- Day-type callIncorrect
- Called
- Range after the prior session's 1.78 ATR decline
- Actual
- Gold opened 4.91 above the low, rose 56.49, and closed 2.60 below the high, which is a trend-day profile
- LeanIncorrect
- Called
- Short-leaning, conditional on a held loss of 4,110.50 or a failed repair below 4,244.28
- Actual
- Neither short trigger fired, and price closed 1.37% above the open
- Lead scenario, post-breakdown digestion (36%)Partial · not the scenario that resolved
- Called
- Hold 4,110.50, stay below 4,200.00, rotate around 4,155.25, then test a range edge
- Actual
- The containment and midpoint rotation occurred, but price did not test 4,200.00 and closed near the high with directional intent
- Driver stackIncorrect · as a directional input
- Called
- Bearish price structure supported a short lean, while real yields and the dollar were unverified
- Actual
- The bearish structure did not produce continuation. The two primary macro drivers remained unavailable as confirmation, and gold rallied after both 14:00 UTC releases missed forecasts
- 4,510.61 resistanceCorrect · remained out of play
- Called
- Remote range ceiling, not an intraday target
- Actual
- Session high was 4,177.30, leaving the level remote
- 4,370.94 resistanceCorrect · remained out of play
- Called
- Major repair threshold
- Actual
- Never tested
- 4,303.11 to 4,315.55 resistance bandCorrect · remained out of play
- Called
- Supply after a deep rebound
- Actual
- Never tested
- 4,280.20 resistanceCorrect · remained out of play
- Called
- Full-session retracement marker
- Actual
- Never tested
- 4,244.28 decision edgeCorrect · trigger stayed inactive
- Called
- First credible repair level, requiring displaced H1 acceptance and a held retest
- Actual
- Never tested, so the bullish repair branch did not activate
- 4,200.00 liquidity pivotCorrect · remained overhead
- Called
- Sweep target whose reaction would matter more than the touch
- Actual
- Session high stopped at 4,177.30
- 4,110.50 to 4,114.70 support and close zoneCorrect
- Called
- Immediate decision zone. A hold favors digestion; acceptance below opens price discovery
- Actual
- Price reached 4,113.30, held 2.80 above the floor, and rallied 64.00 to the high
- 4,100.00 beyond-range liquidityCorrect · remained out of play
- Called
- Sweep target only, not confirmed support
- Actual
- Never reached
The tape
- Open, 00:00 to 07:00 UTC
Gold opened at 4,118.21 and tested the support zone twice. The low printed at 4,113.30 during the 04:00 UTC hour, only 2.80 above the 4,110.50 floor. Sellers never produced an H1 close below the floor, much less the 3 to 15 dollar displacement and failed retest required by the bearish continuation branch. Price recovered to 4,140.11 before London, then pulled back to a 4,126.57 close at 07:00 UTC.
- Mid-session, 08:00 to 16:00 UTC
London lifted gold through the 4,155.25 midpoint area, but the 14:00 UTC data did not create a clean immediate break. Consumer Confidence printed 81.9 against 89.4 expected, while JOLTS printed 7.079 million against 7.106 million expected. The 14:00 UTC candle opened at 4,153.24 and closed at 4,153.36 after trading both sides, followed by a pullback to 4,148.45 in the next hour. That left the first reaction indecisive, in line with the preparation's warning not to act inside the release window.
- Late and close, 16:00 to 23:59 UTC
The durable advance started later. Gold reached 4,174.95 during the 17:00 UTC hour, gave most of that push back by 20:00 UTC, then surged again from 4,143.79 to a 4,170.76 close during the 21:00 UTC hour. The session high of 4,177.30 followed at 22:00 UTC. Gold closed at 4,174.70, up 56.49 from the open and only 2.60 below the high. The full-day range was 64.00, about 0.67 of the preparation's 95.47 D1 ATR.
What We Learned
The moderate surprise was not that 4,110.50 held. The lead scenario allowed that. The surprise was that gold converted the hold into a trend-day close without touching 4,200.00 or activating the mapped bullish repair above 4,244.28. The weak US releases did not produce an immediate clean rally, and the lasting move formed later, first at 17:00 UTC and then through the 21:00 UTC push. That timing was foreseeable because the preparation named a second decision window, but the bullish outcome inside the lower range was not mapped.
- Day-type precondition checks: A post-trend digestion setup can still produce a directional traverse when the session opens directly above a major floor and never accepts below it. The next range call should state what would keep the close near the midpoint. If price holds the lower edge and advances through the midpoint without returning, the day-type override should switch to Trend before the close.
- Driver stack mis-ordering: The short lean rested on bearish price structure while real yields and the dollar, gold's two primary drivers, were unverified. Add a priors rule that an unconfirmed primary driver stack cannot support a directional lean merely because bearish scenario weights sum above 50%. Keep the lean neutral until a mapped price trigger fires or the primary drivers confirm it.
- Right scenario, untradable trigger: The digestion trigger was too permissive. Two H1 closes inside a 4,110.50 to 4,200.00 band confirmed containment but did not distinguish mean reversion from a directional move through the band. Tighten the tradability standard by requiring a rejection from the midpoint or both-way rotation before calling digestion active. Otherwise classify the setup as unresolved containment.
Reviewed prep: 2026-09-29-xauusd-session-preparation
