Prep audit — Review date: 2026-07-20 | Prep file: public/data/reports/2026-07-20-xauusd-session-preparation.md | Prep frontmatter date: 2026-07-20 ✓ | Prep lead scenario: "Range holds — two-sided chop into the FOMC" (45%) — no sustained H4 close outside ~$3,985–$4,030 | Prep directional lean: "Neutral / Wait"
Session Summary
Gold spent Monday July 20 inside the range the preparation described, opening near $4,008, probing the twice-rejected $4,020–$4,029 cap once more without a sustained break, and closing near the session midpoint around $4,007–$4,008 — essentially flat on the day. The preparation's top-weighted scenario (range holds, 45%) was the operative outcome, and the session's real catalyst was a building hawkish Fed chorus rather than the data-light calendar the prep correctly flagged.
Session: GOLD A-Cluster
Symbol: XAUUSD
Window: 22:00–22:00 UTC (active: London 07:00–16:00 / NY 13:00–21:00 UTC)
Regime: Range/chop confirmed; cap rejected a third time; floor untested
Preparation: Accurate
Surprises: Low
Confirmed session open: ~$4,008 (broker H4 close, 05:00 UTC) | Confirmed session close: ~$4,007–$4,008 (spot ~$4,006.74; COMEX August settle $4,015.90) | Net direction: essentially flat, marginally softer on the day
Note: direct MT5/candle access was unavailable in this session. The open and close above are corroborated by three independent sources: the published July 20 preparation's own confirmed reference price, the July 21 preparation's "yesterday's call" recap (which explicitly states Monday closed near $4,007–$4,008 within the $3,985–$4,029 band), and web-sourced spot/futures prints (Reuters/CNBC: spot $4,006.74, down 0.2%; COMEX August futures settle $4,015.90). All three converge on a flat, range-bound session, so the directional grade below is treated as confirmed rather than estimated.
Pre-Session Expectation
The preparation entered Monday with a Neutral/Wait directional lean and a scenario map weighted toward continued balance:
- Lead scenario (45%): range holds — two-sided chop into the FOMC, with no sustained H4 close outside roughly $3,985–$4,030.
- Secondary scenario (33%): an NY/COMEX reclaim above ~$4,030, opening $4,050–$4,065.
- Tertiary scenario (22%): a COMEX breakdown below $3,985, re-exposing the $3,959–$3,969 floor.
The regime was described as a ranging consolidation nested inside a broader multi-month downtrend, with the $3,959–$3,969 floor twice-defended (July 16 and 17) and Friday's strong reversal close ($4,015.75) leaving the near-term picture genuinely two-sided. The prep called Monday data-light — no tier-1 US catalyst — and identified the NY/COMEX window (13:00–15:00 UTC) as the session's resolution engine, with the July 28–29 FOMC as the real macro resolver. The sentiment view feeding the prep was explicitly flagged as stale background context, not a live signal.
What the Market Actually Did
Open / overnight (already largely complete by the prep's writing): Price had gapped down to roughly $3,996 over the weekend, poked the $4,029 cap, and settled near $4,008 by 05:00 UTC — essentially the session's opening reference.
London (07:00–09:00 UTC) and early US: No confirmed break of either range edge. Gold held inside the $3,985–$4,029 band through the London session, consistent with the Judas-roundtrip prior for this window on a data-light day.
NY/COMEX (13:00–15:00 UTC) and the Fed-speak overlay: No tier-1 data print forced a decision, exactly as the prep anticipated — but a growing hawkish Fed chorus became the session's substantive driver. Cleveland Fed President Beth Hammack made a public case for further rate hikes to curb inflation, joining a swelling group of policymakers ahead of the July 28–29 FOMC blackout. Traders responded by pushing the CME-implied probability of a December hike to roughly 80%, up from 73% the prior week. Brent crude traded above $90 as US strikes against Iran entered a ninth consecutive day, with tanker damage reported near the Strait of Hormuz — normally a gold-supportive geopolitical backdrop, but the hawkish real-yield repricing offset it, leaving gold's correlation to oil inverted for the session.
Close: Gold settled essentially flat — spot down about 0.2% to $4,006.74, COMEX August futures down roughly 0.1% to $4,015.90 — closing near the session's midpoint. Neither range edge gave way: the $4,020–$4,029 cap held (its third rejection, counting Friday's high and the pre-session Monday probe), and the $3,985–$4,000 pivot held as well, with the deeper $3,959–$3,969 floor never tested.
Preparation vs Reality
| Pre-session view | What actually happened | Assessment |
|---|
| Lead scenario: range holds — two-sided chop into the FOMC, no sustained close outside ~$3,985–$4,030 (45%) | Session traded within $3,985–$4,029 all day, closing near the midpoint (~$4,007–$4,008); no decisive break either way | Correct — the top-weighted branch fired |
| Secondary scenario: NY/COMEX reclaim above $4,030, opening $4,050–$4,065 (33%) | Cap held at $4,029; no sustained close above $4,030 | Did not occur |
| Tertiary scenario: COMEX breakdown below $3,985, re-exposing the $3,959–$3,969 floor (22%) | Pivot held; low of session ~$3,996, well above $3,985 | Did not occur |
| Directional lean: Neutral/Wait | Session closed essentially flat versus the open | Correct |
| $4,020–$4,029 cap: twice-rejected immediate resistance | Rejected again intraday — a third test, still held | Correct |
| $3,985–$4,000 pivot: session fulcrum | Held throughout; low ~$3,996 stayed inside the zone | Correct |
| $3,959–$3,969 floor: twice-defended demand | Not tested — price never traded close enough to challenge it | Not tested (consistent with range holding) |
| Sentiment view flagged as stale, background only | The session's real catalyst (Hammack-led hawkish Fed chorus) was fresh, same-day information the cached sentiment view could not have captured | Correctly caveated — the "background only" framing was validated |
Overall: Accurate. The scenario map's highest-weighted branch fired cleanly, both key levels held on a closing basis, and the Neutral/Wait lean matched a session that closed within a few dollars of its open. The one gap in the original framing — calling Monday "data-light" — was technically correct for the scheduled calendar but missed that Fed-speak volume, not scheduled data, would be the session's actual information event. It did not change the outcome, since the hawkish chorus reinforced rather than broke the range, but it is worth naming precisely.
What Caught Us Off Guard
A hawkish Fed chorus, not the calendar, was the session's real catalyst. The preparation correctly dismissed Monday's scheduled prints (CB Leading Index, Canada CPI) as non-events for gold, but did not anticipate that a cluster of Fed officials — led by Cleveland's Beth Hammack — would make an unscheduled, coordinated case for further rate hikes, pushing December hike odds from 73% to roughly 80% intraday. This was foreseeable in principle (Fed-speak calendars are public), but the prep's "data-light" framing implicitly treated the day as catalyst-free rather than flagging the speaker slate.
Gold's correlation to oil inverted. The Iran conflict escalated further (ninth consecutive day of strikes, tanker damage near Hormuz, Brent above $90) — the kind of headline that typically lifts both oil and gold together on a risk-premium basis. Instead, gold stayed capped because the same escalation fed inflation fears that hardened rate-hike bets, and the real-yield channel dominated the safe-haven channel. The prep's instrument-characteristics section named the dollar and real yields as the "structural" driver and geopolitics as a "fast-moving overlay," which is exactly what played out — this is a case where the prep's own stated hierarchy was vindicated by the day's events, not a miss.
Beyond those two dynamics, the session unfolded within the expected parameters. No material surprises to the range itself.
Implications for Next Preparation
-
Track Fed-speak density explicitly, even on nominally data-light days. Hammack's remarks were the session's substantive catalyst despite no scheduled tier-1 print. Add a same-day Fed speaker check to the "data-light" designation so a cluster of hawkish officials isn't treated as a non-event.
-
The $4,020–$4,029 cap has now rejected three consecutive tests (Friday's high, the Monday pre-session probe, and Monday's intraday retest). Raise conviction that a genuine break needs stronger COMEX displacement than the prep's default $3–15 threshold — a maturing supply zone this well-defended likely needs to see a full ATR of follow-through to be trusted.
-
Make the oil/gold correlation-inversion condition explicit rather than implicit. The prep's instrument-characteristics note (real yields dominant, geopolitics an overlay) predicted Monday's disconnect correctly but only in general terms. Next preparation should state directly: "if hawkish Fed-speak coincides with a geopolitical oil spike, expect gold to track real yields, not oil" — this is now a repeatable, testable rule rather than background color.
-
December rate-hike odds (CME FedWatch) jumped from 73% to 80% intraday — track this metric explicitly in the regime section as the fastest-moving real-yield proxy with the FOMC just over a week away; a further move toward 85%+ ahead of July 28–29 would tilt the lean short even without a level break.
-
The $3,959–$3,969 floor has now gone untested for a third straight session (last challenged July 17). An untested level isn't confirmation of continued defense — it simply wasn't challenged. Next preparation should avoid treating the floor's "twice-defended" status as strengthening further until it is actually retested.