Day 3 · Gold, Asia, London and New York (from Thailand)

Market context

After Monday's $170 fall, gold opened Asia at 4,121 between yesterday's New York double bottom (4,110–4,120) and the broken London low (4,140–4,148). No major US news; US–Iran messages through mediators kept headline risk alive, and retail traders were 75% long.

Key levels

The plan

• Buy a failed breakdown at 4,110–4,120: 5m close back above 4,121.5, confirmed 1m higher low; skip if the dip goes below 4,115.5 (stop over $8).
• Sell a rejection at 4,140–4,148: 5m close back below 4,138.5, confirmed 1m lower high; target 4,122.
• London: sell a rejection at 4,140–4,148 (5m close back below 4,138.5, confirmed 1m lower high), stop above the nearest 5m swing high; buy a failed breakdown at 4,110–4,116.

What happened

The Asia markup froze on a permission prompt, so I did it by hand just before Tokyo opened. At 07:15 Thai a $7–8 candle dropped into the support zone and wicked to 4,114.49, past my skip line. I bought a pullback at 4,123.56 anyway, with the stop under a small 1-minute swing inside the zone. The 1-minute was still making lower highs, so the higher low wasn't confirmed; it slid into my stop (−£4.44), dipped to 4,113.46 and then rallied $20 without me. Later price crept up to tap 4,140 twice, closed back below 4,138.5 and made a confirmed 1-minute lower high. I missed the first signal, waited for a bounce and sold 4,136.55. It fell steadily with lower highs; I trailed the stop and was taken out at 4,129.91 (+£4.98) as the 1-minute turned up off a 4,124–4,125 shelf from earlier in the morning. London: gold spent 40 minutes inside the 4,140–4,148 zone, made a double top at 4,148.9 and lower highs, but kept finding buyers at 4,137.8–4,140. The 15:30 Thai candle closed at 4,138.08, below my 4,138.5 trigger but only 70 cents under that shelf. I sold 4,138.69 with the stop above the nearest swing (4,143.94). There was no bounce first; buyers defended the shelf and it hit the stop (−£4.01). Afterwards the hourly chart showed rising lows (4,113 → 4,122 → 4,136 → 4,138) under a flat 4,149 ceiling: a squeeze, not a sell. New York: no trade. After the 21:00 Thai JOLTS news gold pushed up through 4,160–4,171 (5m close 4,174.55), but the retest failed and the 21:30 candle closed back inside the zone at 4,168.24: a failed breakout. The only clean sell was the first lower high back into the zone (about 4,169.6, stop above the 4,175.17 fake-out high, about $5.90), which I first misjudged as too wide by measuring from the current price. It dipped to 4,162.74 and bounced; it never reached the 4,160 target.

Trades I skipped (and why)

• 08:03 Thai: Anything in the middle of 4,113–4,129. Two hours of sideways chop inside and above the support zone: no level, no trade.
• 10:30 Thai: First tap of 4,140. Only a tap of the zone's edge, no close back below yet.
• 15:08 Thai: Sell after the first test of 4,140–4,148. No clean 5m close below 4,138.5, and price went back above the last lower high (4,145.3): the rejection failed.
• 16:10 Thai: Anything inside 4,137–4,149. The hourly was squeezing (rising lows under a flat ceiling): middle of the range, and two losses already today.
• 21:20 Thai: Buy the breakout above 4,160–4,171. Needed a retest that held and a confirmed 1m higher low; the retest failed and the 21:30 5m closed back inside the zone.
• 21:40 Thai: Sell the failed breakout's first lower high (~4,169.6). Wrongly judged too wide (measured from the current price, lesson 19). It dipped to 4,162.74 but didn't reach the 4,160 target.

Lessons

• A 1-minute higher low is only confirmed once a 1-minute candle closes above the previous candle's high. Before that it's a guess.
• Never tighten a stop to fit the limit. The stop goes beyond the real invalidation (the dip low); if that's too wide, skip. My rule stop at about 4,113 would have survived and the trade would have hit 4,140.
• Missing a signal is fine: waiting for a bounce and a second confirmed lower high gave a clean entry.
• Exits need the session's own 5-minute levels too. The 4,124–4,125 shelf stopped the move; the target should have been about 4,126, just before it.
• The stop goes $1.50 beyond the NEAREST 5m swing that formed the rejection, not the top of the whole zone test (lesson 17). Otherwise almost every trade is too wide.
• Don't sell into the session's own support: a trigger must clear any 5m shelf built earlier in the session by about $1.50 (lesson 18). The close at 4,138.08 was sitting on it.
• Work out the risk from the planned entry (the pullback lower high or higher low), not from where price is when you look. Measured from the bounce, the failed-breakout sell was a $5.90 stop, not $9 (lesson 19).