Same 62% Score, Two Answers: A GBPUSD Short Back to London's Low
SkyAnalyst AI journal entry: GBPUSD Short on Sep 22, 2026 closed +2.34R on TP3. Full workspace view, decision log, and AI reasoning, unedited.
Before the trade, meet the system.
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 specific setup the trend agent almost passed on.
At 14:02 UTC on Tuesday, SkyAnalyst looked at a GBPUSD short, scored it at 62 percent, and said wait. Seventy-nine seconds later it looked again, scored it at 62 percent again, and said enter. Nothing in the thesis changed between the two reads. What changed was the price: it had slipped below the 1.33600-1.33610 entry zone on the first look and come back into it by the second. The fill was 1.33607 at 14:03 UTC, with the stop 14.8 pips higher at 1.33755. TP1 at 1.33500 filled 13 minutes later. TP2 followed at 15:33, and at 15:47 UTC TP3 printed at 1.33260, for +2.34R (TP3) in 1 hour 43 minutes. The broker closed the position at TP1, so the realized result logged to our track record is +0.72R (TP1). The target came from the morning's map. TP3 sat 0.4 pips above the low London had printed earlier in the session, at 1.33256. The morning analysis had mapped both ends of London's range, sold the failed rally in the middle, and aimed for the far end. Across the whole trade, price went 1.5 pips against the entry. Compare that with the previous day's Cable short, which reached TP3 only after coming 1.1 pips from its stop overnight. This one never tested its stop.
London drew the map before New York sat down
By the time the NY AM analysis ran, London had already done two things. It had washed GBPUSD out to a session low of 1.33256, and it had then rallied the pair back into 1.3370-1.3374, a band where the five-day EMA, the prior day's high and the Trend Agent's resistance and invalidation zone overlapped. That rally failed. Price slid back below the daily open at 1.33615 and below the 60-minute VWAP near 1.3365-1.3367. The analysis read London as a mild bearish template for the New York overlap and wrote the instruction in plain words: sell the rally, do not chase longs. The same London-first reading sat behind a Cable short on September 2.
The dollar side agreed. DXY was at 100.44, above its five-day EMA of 100.33 and above the prior day's high. VIX was calm and below its own five-day EMA, which made this a steady dollar bid rather than a flight from risk. Cable's inverse relationship with the dollar was intact, so the playbook's divergence gate had nothing to object to.
The agents lined up with it without shouting. The Macro Agent leaned bear on GBPUSD at 72 percent in the analysis, citing Fed-BoE policy divergence and a quiet UK data calendar. The Trend Agent was bearish at 63 percent but called the regime TRANSITIONING and the strength WEAK, and its recommendation was to reduce size. Macro above 70 with the dollar aligned against Cable triggers a hard rule in the playbook: no longs. The analysis applied it and wrote that no long setup qualified.
One thing was not clean. A speech by President Trump was scheduled for 9:55 ET, right at the start of the window this setup would trade in, and the analysis listed headline whipsaw around 10:00 ET as its primary risk. That risk became one of the two reasons behind the first evaluation. If you want to see a morning map like this drawn on your own markets, try the same analysis engine on your charts with a 21-day free trial.
Retrace Rejection Short. The pattern sells a market that has already shown its direction, on the bounce back into a level it recently lost. The retrace is where short covering and late buyers push price up; the rejection is the candle that shows they ran out. Professional traders use it because the stop can sit just above the retrace, close to where the idea is proved wrong, while the targets sit at levels the session has already printed.
The zone was a stack, not a line
The entry zone was 1.33600 to 1.33610, and it was built from levels that sat together: the daily open at 1.33615, the flip zone where the prior day's low had turned from support into resistance, and the 38.2 to 61.8 percent retracement of the intraday drop. The Trend Agent's VWAP at 1.33656 and key resistance at 1.3376 sat above it, so a buyer had to get through several layers before the short was wrong.
The trigger had two parts
The analysis wrote the rule before price arrived. First, a 5-minute retrace into that zone. Second, a 5-minute close back below 1.3360 with RSI failing under 50 or the MACD histogram rolling down. It added a timing clause for the 9:55 ET speech: if the headline flow spiked price, wait for a 50 to 61.8 percent retracement of the spike before selling, and do not chase the first move.
The trigger does not ask for a new low. It asks for proof that the bounce failed at the level, which is information a touch alone cannot give.
The targets came from London
The stop went at 1.33755, just under the Trend Agent's 1.3376 invalidation, which put 14.8 pips of risk on the 1.33607 fill. The targets stepped down through levels the tape had already respected: 1.33500, then 1.33365, then 1.33260 just above London's 1.33256 low. Against the actual stop that is 0.72R, 1.64R and 2.34R.
Where the plan was weaker than it read
The analysis described TP1 as roughly 1.0R and said the 15-pip minimum stop was respected. Measured from the real fill, the stop was 14.8 pips and TP1 was 0.72R, below the 1R to 1.25R the playbook prefers for a first target. The playbook allows a close TP1 only when a strong TP2 at 2R or more exists. TP2 here was 1.64R, so that exception did not cover it either, and the gap between the stated and the measured numbers belongs in the record.
SkyAnalyst doesn't favor any single strategy. This month alone the same system has taken a retracement short into resistance on Cable and bought a Nasdaq pullback on the index side. It reads the tape first and trades whatever pattern the session offers.
London set a mild bearish NY-overlap template for GBPUSD. On the 60m, Cable first washed out to the London low 1.33256, then rallied into 1.3370-1.3374, tagging resistance near the 5-day EMA / yesterday’s high / Trend Agent resistance-invalidation zone, and failed back below the daily open 1.33615 and 60m VWAP ~1.3365-1.3367. That argues for sell-the-rally, not chasing longs.
Macro also leans that way: the Macro Agent is lean_bear at 72% on Fed-BoE divergence and a UK data vacuum; the Trend Agent is bearish 63% but weak/transitioning. DXY is above its 5-day EMA (100.44 > 100.33) and above yesterday’s high, which supports Cable downside. VIX is below its 5-day EMA, so this is not a panic risk-off move; it is more of a steady USD-bid backdrop. The inverse GBPUSD/DXY relationship is still intact today, so there is no divergence-gate problem. Per the hard rule, with Macro confidence >70 and DXY aligned bearishly for Cable, longs are offside.
Directional Bias: Bearish
Volatility: Normal
Setup #1: GBPUSD SHORT
- Entry: 1.33600-1.33610
- Entry Trigger: 5m retrace into the daily open / prior-day low flip zone / 38.2-61.8% intraday retrace, then a 5m rejection close back below 1.3360 with RSI failing back under 50 or MACD histogram rolling down.
- If the 9:55-10:00 ET headline flow creates a spike, wait for the 50-61.8% retracement first; do not chase the initial move.
- Stop Loss: 1.33755-1.33760
- Minimum 15 pips is respected here.
- If execution buffer would force the stop materially above 1.33760, skip the trade.
- Targets: TP1=1.33500, TP2=1.33365, TP3=1.33260
- R-Multiples: TP1≈1.0R, TP2≈1.6R, TP3≈2.3R
- Quality Score: 7.1/10
- Confidence: Medium-High - 6/7 confluences: London bearish mean-reversion bias; DXY supportive; Macro bear 72%; Trend Agent bearish 63%; 60m below EMA/VWAP with MACD below zero; defined 5m entry level. Main risks: trend strength is weak/transitioning, 15m still has bounce potential, and headline noise around 10:00 ET can create whipsaws.
- Invalidation: Stand down on any 15m hold above 1.3376, or if price reclaims and accepts above VWAP / 1.3366+ while DXY loses its bid.
No long setup qualifies.
Reason: hard-rule conflict (Macro confidence >70 and DXY trend align bearishly for Cable), plus London already rejected resistance near 1.3370-1.3376.
Decision log
14:02 UTC, 62 percent, WAIT. The bearish case was intact: the last completed 5-minute candle had closed back below 1.3360, RSI had dropped to 44, and price was under VWAP and both EMAs. The timing was the problem. Price was already below the 1.33600-1.33610 entry zone, so selling there meant selling lower than the plan allowed. It was also 10:02 ET, with the 10:00-10:05 candle after the headline window still unfinished. The system called it chasing and passed.
14:03 UTC, 62 percent, ENTER. Seventy-nine seconds later the score had not moved, but the price had: it had retraced back into the 1.3360-1.3361 zone without touching the stop. The system pointed to the 13:55 5-minute candle, which had probed up into the retrace area and closed back below 1.3360 as RSI fell to 44 and the MACD histogram softened, and judged the trigger met. It named its own caveat, a candle still forming just after the headline window, and called the entry valid but not top-tier. The fill was 1.33607.
On a $100k account at 2.0% risk per trade.
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 | +0.72R | +$1,440 |
| TP2 hit | +1.64R | +$3,280 |
| TP3 hit (max potential) | +2.34R | +$4,680 |
All six agents combined.
What this trade teaches
Same confidence, opposite answers. Many readers expect an AI system to enter when a score crosses a line. Here the score never crossed anything. It was 62 percent on the wait and 62 percent on the entry. The entry rule was about location, and location was the thing that changed.
That distinction matters on a pair like Cable. The stop is set at structure, so selling below the zone does not move the stop. It only moves the fill further from the stop and closer to the targets, which shrinks every R-multiple on the plan. With TP1 only 10.7 pips from the fill, a few pips of chasing would have taken a real share of the first target.
The second lesson is about the target. TP3 was London's low with 0.4 pips of room. When a session has already printed both extremes, the far one is a level the market has already shown it will trade to, and the analysis used it. TP3 filled at 15:47 UTC, and the trade monitor also recorded a print at 1.33245, slightly through the London low.
Valid but not a top-tier clean entry. SkyAnalyst entry evaluation, 14:03 UTC
One honest caveat. The backdrop was supportive but not strong: the Trend Agent graded the trend WEAK and suggested reduced size, and the setup carried a C+ grade. A trade can be well built and still be an average-quality setup. The realized result on the ledger is +0.72R (TP1), because the broker closes the whole position at TP1. The +2.34R (TP3) headline is how far the market went.
From the desk
We picked this trade for the 79 seconds in the middle of it. At the same 62 percent, the system declined to sell from below the zone and accepted the same idea once price was back inside it. A quiet trade shows those mechanics plainly: sold at 10:03 ET, finished before noon in New York, never more than 1.5 pips against the entry.
This is one trade, and one trade proves very little. What it does show is an entry refused until price came back to where the plan said to sell, and a final target that London had drawn before the setup was written.
At a Glance
What this teaches about AI-driven trading
What is a retrace rejection short?
A retrace rejection short sells a market that is already falling, on a bounce back into a level it recently lost. The trader waits for price to climb into that level and then for a candle to close back below it, which shows the bounce failed. The stop sits just above the level, so the risk is small compared with targets placed at lower levels the session has already printed.
Why would a trading system pass on a setup and take it a minute later at the same confidence?
Confidence measures how strongly the evidence supports the direction. Entry rules also cover location and timing. If price has already moved past the planned entry zone, selling there means a worse fill against the same stop and targets. When price returns to the zone, the same thesis becomes a valid entry. The score can stay flat while the decision changes, because the market's position changed, not the view.
How do session highs and lows work as take-profit targets in forex?
Session extremes mark prices where one side ran out of orders earlier in the day. When a later move heads back toward them, those levels often attract price and then slow it down, because stops and resting orders cluster there. Placing a target just inside the extreme, rather than exactly on it, raises the chance of a fill if price stalls a few pips short of the level.
When should a trader wait after a scheduled speech or data release?
Scheduled headlines can spike price in either direction for a few minutes before the market settles. A common approach is to let the first 5-minute candle after the event complete, or to wait for price to retrace 50 to 61.8 percent of the spike before entering in the planned direction. That avoids entering on the initial move, which often reverses once the news is absorbed.
What is the difference between full-potential R and realized R on this trade?
Full-potential R measures how far the market travelled before the setup was exhausted: here TP3, for +2.34R (TP3). Realized R is what the account booked. This desk closes the entire position at TP1, so the logged result is +0.72R (TP1). Reporting both shows the whole move and the conservative result in the track record.
Run your markets with SkyAnalyst
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.
From the SkyAnalyst Journal
The Cable Short That Came 1.1 Pips From Its Stop, Then Ran to TP3
A GBPUSD short sold a failed pivot retest at 1.33735, got a hold verdict at 20:02 UTC, then watched price squeeze to 1.33874 overnight. London took it to TP3 by 08:31 UTC.
Every One of This Week's Six Losses Cost Exactly 1R.
Six stop-outs, a four-loss streak on Monday, and a 9.9% drawdown on the simulated account. Not one of them cost more than the 1R it was sized for, and that is the only part of this week we are happy about.
The Regime Flipped on Monday. We Kept Selling Until Friday.
Seven trades, one winner, minus 4.95R. Every loss was a short, the average short died sixty-seven minutes after it was filled, and the only trade that went the other way was the only trade that paid.