Orion Sold Gold Where Price Wasn't, Then Waited for It

Everything sold at once

Gold spent the night bleeding from 3964 to 3895 through Tokyo and London, without threatening the previous day's low at 3881.87. Between 7:00 and 8:00 AM ET it squeezed to 3971.49, tagged the prior high of 3970.48 by about a point, and failed. By the time the plan was written, it had given back 65 percent of that leg, and the day had covered 76 points of range against the previous session's 88.

JOLTS broke the shelf

Job openings at 10:00 AM ET printed 7.52 million against 7.15 million expected. The two 15-minute bars that followed fell from 3945.9 to 3922.86, through a shelf that had held from 9:00 to 9:45 AM ET with its base at 3940.24. The plan read the tape as risk-off across the board rather than a haven bid: the 10-year yield at 4.95 percent and up, the dollar index at 101.93 against a day high of 101.947, the Dow below its prior-day low, and $ADD at -368. Gold was falling with risk, not against it.

The agents had only just turned

The Macro Agent's 10:02 AM ET row had gold at lean bear, 59 percent, with a bias score of -28. It cut size to 0.85 times normal and called the setup fragile, with "upside asymmetry present". The Trend Agent had been bullish at 61 percent as late as 9:51 AM ET. It went neutral after JOLTS and turned bearish at 62 percent at 10:26 AM ET, with an invalidation level of 3946.25. On the 15-minute chart price had lost VWAP at 3930.2 and its EMA20 for two closes, and the 60-minute EMA stack was bearish.

Breakdown retest. When a level breaks on news, the first move often overshoots, and the price where the break began becomes the place sellers defend if price returns. The pattern sells that return instead of the overshoot. The entry sits at the breakdown's origin, the stop goes above the structure that would show the break has failed, and the trade simply does not happen if the market never comes back.

Why not short at 3924.5

The plan was explicit about the price on the screen. At 3924.5 gold sat on the session low of 3922.86, with 3920 and 3919.48 just under it. The 5-minute chart was 2.3 ATR below its EMA20 with RSI at 32. A shallow retest to VWAP or the pivot at 3930-3933 did not fix it either: with an honest stop, the session low would sit less than 1R away.

The zone and the stop

The zone was 3936-3939, at the 3936.25 high of the 10:15 AM ET bar where the breakdown began, just under the broken 3940.24 shelf. The stop went to 3947.4, above the 3945.90 post-release high and the Trend Agent's 3946.25 invalidation, and clear of the one-point overshoot the morning had shown at 3970-3971. From the near edge that was 11.4 points, 1.22 times the 15-minute ATR of 9.33.

The targets

TP1 at 3923.2 sat just in front of the 3922.86 session low, 12.8 points from the near edge, or 1.12R. TP2 at 3914.4 was the 15-minute 78.6 percent fib, alongside the 60-minute 23.6 percent fib at 3913.1, for 1.89R. TP3 at 3910.6 was a broken swing high, for 2.23R.

The doubts it wrote down

The plan put the chance of TP1 before the stop at 5.0 out of 10. Its first doubt was reachability: the zone needed roughly a 12-point bounce inside the order's 30 minutes. It also noted that yields had come off their 5.037 high to 4.954, that the tape had changed direction twice in four hours, and that an 11.4-point stop was smaller than the 13.39-point range of the 10:15 AM ET bar. It asked for size at no more than 0.5 to 1 percent of equity.

Our system doesn't favor shorts over longs, or one pattern over another. The same Orion line bought the Nasdaq that morning because its range was holding, and sold gold because its shelf had broken, and in both cases the plan described where the trade stopped making sense before it described where to enter.

What this trade teaches

The plan's main doubt was the one that resolved first. It wrote that the zone might simply not fill, because reaching it needed a bounce of about 12 points in 30 minutes. The bounce took 11 minutes. A plan that waits for price to come to it accepts that some days it will not, and we would rather miss those trades than move the order down to the screen.

Price is sitting on the obstacle. SkyAnalyst setup analysis, 15:31 UTC

The fill came before the confirmation

The gate described 5-minute bars stalling inside 3936-3939 with upper wicks, then a 5-minute close back below 3936. The order filled 2 seconds after price touched the zone. The 5-minute bar it filled in closed at 3935.62, below 3936, after the fill. The next three bars closed between 3936.5 and 3937.32, inside the zone, and no 15-minute bar closed above 3940. The worst price before TP1 was 3938.67 at 10:53 AM ET, 2.67 points against the entry and 8.73 under the stop.

A slow winner

TP1 took 3 hours 39 minutes. For most of the first half hour gold drifted inside a few points of the fill, and the Trend Agent's 10:43 AM ET read, a minute after the entry, had already gone back to neutral in a choppy regime. Patience was not a decision anyone made in that window, because the position was simply open with its stop and target in place. After TP1, price reached 3917.46 at 3:03 PM ET, 3.06 points short of TP2, then rallied. The tracked record reached the stop at 3947.4 at 8:15 PM ET. Because the broker closes at TP1, the ledger entry was fixed at 2:22 PM ET, the same rule that protected Monday's Dow long.

From the desk

Our ledger records +1.12R (TP1), booked at 2:22 PM ET, and the hero panel matches. The result panel's 3 hours 39 minutes run from the fill to TP1, and its Max drawdown of 2.67 points is the 10:53 AM ET high, the worst price before TP1. The session label reads NY AM, since the fill came at 10:42 AM ET.

The Macro Agent stayed lean bear on gold all day, between 55 and 59 percent across its updates. At 3:01 PM ET, two minutes before the post-TP1 low of 3917.46, its flag read "Gold can sell off on yields but rebound sharply on headlines". That flag fits the evening, though nothing in our record ties the rally to a specific headline, and we leave the cause open.

One profitable short does not tell us the line is right about gold. What it shows is a plan that was specific about the price it would not accept, and a market that came to the price it would. The last losses report is where the trades that went the other way are written up in the same detail.