SkyAnalyst AI journal entry: Orion XAUUSD Short on Nov 19, 2026 reached +2.42R (TP3); the ledger booked +1.08R (TP1). Full workspace view, decision log, and AI reasoning, unedited.

SkyAnalyst is not one AI trader. It is four specialist agents, each with its own data pipeline, each maintaining state between evaluations, and each required to agree before a position is sized. They don’t chat in prose. They write structured messages to a shared state object that each reads on every evaluation cycle. That’s what makes the system auditable, and it’s what this case study will show, step by step, on a gold short built on a morning when weak US data brought no haven bid, approved automatically at the zone, and carried to TP3.
Gold came into Thursday off a down day with a 138-point range: it opened near 3854 on Wednesday and closed at 3724.79. Overnight, Tokyo and early London held a 3725-3745 range, and that range gave way into the New York morning. Then the Philly Fed printed 6.3 against a 20.6 forecast. A weak factory reading tends to pull yields lower and soften the dollar, and both of those usually help gold. Neither happened.
The Orion plan logged the dollar index at 103.90, close to its 103.99 high, and the 10-year yield holding at 5.08%. Equities sold at the same time: the plan noted the Dow below the previous day's low and NYSE breadth, $ADD, at -1367. The Atlas US500 short from this morning came out of the same tape. When stocks and gold fall together while yields hold up and the dollar firms, the selling usually reflects tighter money rather than fear, and fear is what a haven bid needs.
At 16:02 UTC the Macro Agent scored XAUUSD bearish at 72 percent confidence with a bias score of -70. Its reasons were hawkish Fed repricing, 10-year real yields that had climbed from 2.66% on November 9 to 2.88% on November 18, a dollar index of 103.849 above its five-day average, and about $1.37 billion of outflows from the largest gold ETF over 30 days. It also tagged the regime high volatility, with VIX at 25.23, and cut its position size multiplier to 0.65. The Trend Agent's last read before the order, at 15:53 UTC, was bearish at 82 percent with a strong trend, and it recommended reduced size because volatility was elevated, noting that high VIX "has not translated into safe-haven support." It also warned that momentum was stretched.
By 11:31 AM ET the 15:00 UTC hourly bar had broken the previous day's low at 3719.75 and the 3693 hourly swing on 1.55 times normal volume. Gold was trading near 3680, about 44 points, or roughly 4 average true ranges on the 15-minute chart, below session VWAP. The 5-minute chart showed bullish divergence, and the last four 5-minute bars were holding 3674-3682 above the 3669.59 session low.
Pullback short into a broken swing. When a trend breaks a level and runs far from its average, traders often skip the extended price and wait for the first pullback into the level that broke. Former support becomes the place to sell, and the stop goes above the point where the break would be undone.
The plan put it plainly: "Chasing a short here at -4 ATR is wrong. The trade, if any, is selling a bounce into broken structure." With price 44 points below VWAP and divergence on the 5-minute chart, a short at the low would have put the stop far away and the nearest target close.
The zone, 3687.0-3692.0, sat on the 5-minute 38.2-50% Fibonacci retracement, under the broken 60-minute swing low at 3693.36 and the 15-minute highs of 3693.75 and 3692.45. The plan defined invalidation as acceptance back above about 3694, then the 5-minute 61.8% level at 3699.4 and the round 3700. It put the stop at 3702.4, beyond all three, for 15.4 points of risk from the zone edge against a 15-minute ATR of 10.5.
The only structure below price was the session low at 3669.59, so TP1 went just in front of it at 3670.5, about 1.07R from the zone edge. TP2 and TP3 used the round numbers 3660 and 3650, at 1.75R and 2.40R. The plan scored the setup 5.3 out of 10 and gave 5.3 out of 10 as its probability of reaching TP1 before the stop.
The plan described a gate: a 5-minute upper-wick rejection inside 3687-3693, no 15-minute close above 3693, and the dollar index holding above about 103.85. Orion does not run that check at the moment of entry. It approves automatically when price enters the zone, once the news block and the hard gates pass, so the order filled on the first touch.
SkyAnalyst doesn't favor any single strategy; the system reads the tape first. On a tape that had already run 4 ATR from its average with the dollar firm, the plan picked a pullback short over a breakdown chase and placed the risk above the structure that had just failed.

The tendency I'm relying on: after a 15-minute-close acceptance below a level with the dollar confirming, gold's pullbacks are shallow. This morning's evidence:
Probability of TP1 before the stop: 5.3/10. Trend, macro and location agree, but there are two real doubts: fading momentum with divergence, and the late-session clock.
Directional Bias: Bearish Volatility: High
Setup #1: XAUUSD SHORT Pullback
Risk: standard size is no more than 1% of equity. Consider 0.5% given the high VIX and the stretched tape.
16:35 UTC, approved automatically on zone entry. There was no AI read at this point: price touched the zone at 16:35:24 UTC, the news block and the hard gates passed, and the order filled at 3687.1 at the near edge of the zone, three minutes and 39 seconds after monitoring began. The confidence bar shows zero because no model confidence was recorded, not because the setup was scored at zero.
Each trade risks +$2,000 (1R). The system's actual scale-out behavior may differ, see disclaimer.
| Scenario | R-multiple | Profit on $100k |
|---|---|---|
| Stop hit (invalidated) | -1R | −$2,000 |
| TP1 hitActual | +1.08R | +$2,160 |
| TP2 hit | +1.77R | +$3,540 |
| TP3 hit (max potential) | +2.42R | +$4,840 |
Weak data helps gold only through a second step: lower yields and a softer dollar. On Thursday that second step never came. The plan did not short gold because the data was bad. It shorted because the dollar index sat near its high, the 10-year held 5.08% and stocks were selling alongside gold.
The 15-minute bar from 15:30 UTC topped at 3693.75, before the plan existed. After the fill, gold climbed to 3692.41 at 11:55 AM ET and then fell to TP1. After TP1 it climbed again, to 3692.51 at 12:32 PM ET. From 15:47 UTC to the exit, no one-minute bar traded above 3693. The worst print against the fill was 5.41 points, about 0.35R, and the stop at 3702.4 was never closer than 9.89 points.
The plan asked for a 5-minute rejection and a dollar check before entry. Orion filled on the first touch of the zone, so neither condition was confirmed at entry. On this trade the zone held. On another morning, an automatic fill can land before any rejection forms. We report that difference because a subscriber reading the plan could otherwise assume the gate was enforced.
The plan worried that the London close and the midday lull would stall the move. TP1 came fast, 31 minutes after the fill, when gold dipped to 3667.39, below the morning low. After that the move did stall. From 12:12 PM to 2:46 PM ET gold held between 3672.27 and 3692.51, and TP2 came 3 hours 53 minutes after TP1, at 3:59 PM ET. TP3 at 3650 followed at 4:24 PM ET.
Our ledger records +1.08R (TP1) on this trade, or +$2,160 (TP1) on the simulated $100,000 account at 2% risk. The move reached +2.42R (TP3), which would have paid +$4,840 (TP3) on the same account. Closing the whole position at the first target left 1.34R of the move unbooked, and on a trade that needed almost four more hours to get there, we accept that trade-off. The plan itself suggested considering 0.5% risk because of the VIX and the stretched tape; the simulated panel keeps the same flat reference we use in every case study.
We last covered gold with an Orion gold short on November 3, a different morning with a different driver. The dollar side of this story showed up in the Atlas EURUSD short on November 18. What ties them together is the part we care about most: the macro read is on record with a timestamp, and anyone can check it against what happened.
Gold usually benefits from weak data because it tends to lower bond yields and weaken the dollar, which reduces the cost of holding a metal that pays no interest. If yields stay high and the dollar stays firm despite the weak number, that support never arrives. When stocks and gold fall together, markets are often pricing tighter policy rather than fear, and a haven bid depends on fear.
A real yield is the return on a bond after inflation, commonly read from inflation-protected Treasuries. Gold pays no interest, so when real yields rise, the opportunity cost of holding it rises too. Over time, higher real yields have tended to weigh on gold and lower real yields have tended to support it. The relationship is not fixed day to day, but it is one of the main drivers analysts watch.
It is a short taken after price breaks below a support level and then bounces back toward it. The broken level often acts as resistance, so the trader sells near it rather than chasing the breakdown. The stop goes above the level, or above the point where reclaiming it would cancel the idea. This keeps risk defined and usually gives a better ratio of reward to risk than selling an extended low.
Round numbers such as 3700 attract orders, so price often reaches them or overshoots them briefly. A stop placed exactly at a round number, or just in front of it, can be hit by that ordinary probe even when the trade idea is still valid. Placing the stop a few points beyond the round number, and beyond nearby technical levels, gives the trade room to survive normal noise.
An automatic zone entry fills as soon as price reaches a predefined range, as long as basic safety checks pass. A confirmed entry waits for an additional signal inside that range, such as a rejection candle or a momentum stall. Automatic entries never miss a quick touch, but they can fill before the setup proves itself. Confirmed entries filter some failures but can miss fast moves entirely.
21-day free trial. No credit card. Full access to the Trend Agent, Macro Agent, and six-factor confluence scoring.
Trading involves substantial risk of loss. Past performance is not indicative of future results. The analysis shown was produced by an AI model operating on SkyAnalyst’s live trading infrastructure; it is shared for educational and research purposes only and is not financial advice. About reported results. Every AI Trader publishes three take-profit targets (TP1, TP2, TP3) per trade. The broker closes 100% of the position at TP1, so two distinct R-multiples appear in this article. The hero R-multiple is the full-potential R: where the market actually traveled (the highest take-profit hit, or the stop loss) before the setup was invalidated or exhausted. The realized R, shown on the TP1 row of the simulated returns panel, is TP1’s R (or -1R on a stop out). The realized R is what we log to our running track record. Both numbers are honest. Showing both is what lets readers see the full arc of the move and the conservative ledger entry it produced. Simulated returns in this article are calculated against a hypothetical $100,000 account at 2% risk per trade (1R = $2,000). These are educational reference figures and do not reflect any specific account or broker execution. Your actual result depends on your position size, your risk parameters, and live market conditions.
The trades in this article are taken by the SkyAnalyst agents in live demo accounts at a regulated broker, in a live market environment, so the demo account reflects every transaction SkyAnalyst makes. The app logs each trade and confirms it against the live price feed of the broker; the capital is simulated and no real capital is at risk. The demo account is linked to MyFxbook, which publishes its results publicly. We publish these results to study how the agents trade and reason, for education and trade analysis, not as a recommendation. Trade at your own risk.
A +5.88R week still opened with three straight stops. Four losses at exactly 1R each, three of them Nasdaq shorts sold into a bounce that had not finished, and one Atlas euro short approved before its trigger fired.
Thirteen trades, all of them shorts, nine wins and +5.88R. Tuesday opened with three stops in a row, then seven straight winners from Thursday morning through Friday turned a 3R hole into the week's result.

The Atlas GBPUSD trader sold Cable at 1.3112 only after the London high was rejected. Eleven minutes later the market tested that rejection again, and it held. TP2 printed at 2:26 PM ET.