Nine Refusals in Thirteen Minutes, Then a Trade at 67
SkyAnalyst AI journal entry: GBPUSD Short on Sep 10, 2026 closed +2.04R 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.
Between 14:02 and 14:13 UTC the system scored the Cable short nine times, at 85, 85, 82, 85, 82, 84, 80, 82 and 86 percent, and returned WAIT on every one of them. At 14:15 it scored 67 and sold at 1.35184 with a stop at 1.3544, a risk band of 17.1 pips. Price ran to the third target at 1.34835 for +2.04R. Nine refusals at scores in the eighties, then an entry at a number lower than all of them. Written as a list it looks like a malfunction. It is the opposite, and the reason is the single most useful thing this desk has published about how it works.
London had already done the work
By the time the New York overlap arrived, the move most traders would have wanted was finished.
London set a clearly bearish tone in Cable and carried it the whole way. The 60-minute chart sold from the 1.3557 area down to 1.3491, broke the daily open at 1.35375, traded below the prior day's low at 1.35344, and stayed under 60-minute VWAP throughout. That is a trending session rather than a ranging one, and the direction was not in question.
The problem is where it left price. London had stretched into an intraday extreme near 1.3500 and the 1.34913 support, which makes the New York tactic straightforward to state and difficult to sit through: do not chase the lows, wait for a rebound into resistance, sell the retracement.
The macro backdrop supported the same direction without ambiguity. DXY at 99.042 was above its 5-day EMA of 98.88 and above the prior day's high. VIX at 18.01 was above its 5-day EMA of 16.39 and pressing fresh highs, a risk-off configuration that is dollar-supportive. Cable was falling while the dollar rose, so the normal inverse relationship was intact and the divergence gate was not tripped.
The Trend Agent read bearish at 65% in a trending regime, naming resistance at 1.35284, support at 1.34913, VWAP at 1.35402 and invalidation at 1.3544. The 15-minute was below both EMAs and below VWAP with momentum still net bearish, even as the 5-minute bounced off oversold.
Sell the Retracement into Resistance. Nine refusals is the most this desk has logged on a single setup, and every one of them was declining to chase a low that London had already produced.
What the nine scores were measuring
They were measuring a thesis, and the thesis was excellent. London trending lower, daily open broken, prior day's low broken, price under VWAP, DXY firm and above the prior day's high, VIX pressing new highs, Trend Agent bearish. Scores between 80 and 86 are an accurate description of that picture.
None of it constitutes a trade. Price at 14:02 was near an intraday extreme with the nearest structural level for a stop far above it. Selling there means either a stop wide enough to ruin the arithmetic or a stop close enough to be removed by the ordinary bounce that follows a stretched move.
What the tenth score was measuring
By 14:15 price had rebounded into the 1.3528 resistance area, with the Trend Agent's invalidation at 1.3544 sitting just above. That geometry is the whole trade. It puts a defensible stop 17.1 pips from an entry at 1.35184, and it does so at a level the market has to reclaim for the idea to be wrong.
The score of 67 is lower than the nine refusals because the entry carries what the thesis does not. The 5-minute had bounced off oversold, which is what created the entry and also what makes it uncomfortable, and a VIX at 18 makes counter-moves sharper. Both are real, and the number says so rather than inheriting the confidence of the setup behind it.
The pattern, stated plainly
Professional traders: this is the fourth time in three weeks that this desk has refused a setup in the eighties and taken it at a materially lower number, and it has won all four. The list is Cable on September 2 at 62 after four refusals, the Dow on September 8 at 76 after two at 83, the Dow on September 9 at 71 after one at 88, and this one.
SkyAnalyst does not favor any single strategy and it does not treat its own conviction as a trigger. It reads the tape first, scores the entry actually in front of it, and declines a setup it agrees with until the geometry exists to risk money on it.
London set a clear bearish tone into the NY overlap: 60m price sold from the 1.3557 area down to 1.3491, broke the daily open (1.35375), traded below yesterday’s low (1.35344), and stayed under 60m VWAP. That confirms London was trending, not ranging. However, London already stretched into an intraday extreme near 1.3500 / 1.34913 support, so the correct NY AM tactic is not to chase lows but to wait for a rebound into resistance and sell the retracement.
Macro/risk regime also favors GBPUSD downside. DXY is above its 5-day EMA (99.042 vs 98.88) and above yesterday’s high, while VIX is above its 5-day EMA (18.01 vs 16.39) and pressing fresh highs: that is a risk-off USD-supportive backdrop. Cable is falling while DXY is rising, so the normal inverse relationship is intact; the DXY divergence gate is not tripped. Trend Agent is bearish, 65% confidence, trending regime, with R=1.35284, S=1.34913, VWAP=1.35402, invalidation=1.3544. 15m remains below both EMAs and below VWAP; momentum is still net-bearish even though the 5m has bounced off oversold. No long setup qualifies here; the risk-off compound means only shorts qualify.
Directional Bias: Bearish
Volatility: High
Setup #1: GBPUSD SHORT
- Directional Bias: Sell the NY retracement into resistance, not the low
- Entry: 1.3518-1.3522
- Entry Trigger: 5m rejection at/just above 1.3520 with a close back below 1.3514, or a 5m bearish turn in MACD histogram after testing the zone
- Stop Loss: 1.35355
- Targets: TP1=1.35045, TP2=1.34915, TP3=1.34835
- R-Multiples: TP1=1.0R, TP2=1.8R, TP3=2.3R
- Quality Score: 8.0/10
- Confidence: High, 6/7 confluences
- London bearish bias aligned
- DXY supports shorts; inverse GBPUSD/DXY correlation intact
- Trend Agent bearish at moderate confidence
- 60m structure bearish: below VWAP, bearish EMA condition, MACD below zero
- 15m confirms: below EMAs/VWAP, RSI below 50, MACD below zero
- 5m entry zone is defined: retracement into 15m 61.8% / 5m 50-61.8% / round 1.3520
- No new high-impact USD/UK event within 30 minutes
- Miss/limitation: Macro Analysis Agent was not provided, so that confluence is unconfirmed
- Invalidation: Any 5m acceptance above 1.3528 weakens the setup; a 15m hold above 1.3532 materially damages it; hard invalidation remains 1.3544
- Risk Note: London already pushed for 3+ hours and tagged a major intraday extreme, so TP1 should be treated as a forced take-profit. Do not hold full size looking for runners if price hesitates near 1.3505/1.3500.
No long setup qualifies under current rules because DXY is firm, VIX is elevated, and the risk-off compound restricts valid trades to shorts only.
Decision log
14:02 UTC, 85 percent, WAIT. London's bearish work was complete and the score reflects it: daily open broken, prior day's low broken, price under 60-minute VWAP, DXY above the prior day's high, VIX pressing new highs. Price was also sitting near the 1.3491 intraday extreme, which is the one place this setup cannot be entered.
14:03 UTC, 85 percent, WAIT. Unchanged. A high score on a completed thesis does not decay just because it is not acted on, and it does not become an entry either.
14:04 UTC, 82 percent, WAIT. A marginal dip as the 5-minute began bouncing off oversold. That bounce is the thing the setup needed, but a bounce in progress is the worst moment to sell it.
14:05 UTC, 85 percent, WAIT. Back to 85. The macro frame had not moved and neither had the structural problem: no rebound into resistance yet, so no defensible stop location.
14:06 UTC, 82 percent, WAIT. Holding through the bounce. Every point of rebound at this stage improves the eventual entry and worsens the immediate one, which is why the score oscillates in a narrow band rather than trending.
14:08 UTC, 84 percent, WAIT. Price working higher into the zone the setup wanted. Still short of the 1.3528 resistance area, and approaching a level is not failing at it.
14:10 UTC, 80 percent, WAIT. The session's lowest pre-entry score, arriving as the bounce gathered pace. The number behaving correctly: as the counter-move strengthened, the case for selling into it weakened.
14:11 UTC, 82 percent, WAIT. The bounce continued toward resistance. Nothing to do but let it arrive.
14:13 UTC, 86 percent, WAIT. The session's highest score, and the ninth refusal. Price was now at the resistance the setup had been waiting thirteen minutes for, which is precisely when the thesis looks strongest and the entry is still one confirmation away.
14:15 UTC, 67 percent, ENTER. The rejection printed at the 1.3528 resistance area with the Trend Agent's invalidation at 1.3544 just above. The Risk Agent placed the stop at 1.3544 and sized against a 17.1 pip band from a 1.35184 entry. The score of 67 prices in the two real reservations, a 5-minute that had bounced off oversold and a VIX at 18 that makes counter-moves sharper. Short.
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 | +0.81R | +$1,620 |
| TP2 hit | +1.57R | +$3,140 |
| TP3 hit (max potential) | +2.04R | +$4,080 |
All six agents combined.
What this trade teaches
The nine refusals are worth more to a reader than the entry is.
Every one of them was a correct decision that produced nothing. There is no way to evaluate them from the outcome, because the outcome of a refusal is that nothing happens. The only way to judge them is to ask what selling at 14:02 would have required, and the answer is a stop either far enough away to make 34.9 pips worth less than 1R or close enough to be taken out by the bounce that arrived two minutes later.
The second lesson is about what a high confidence number is for. It is not a signal to act. It is a description of how much the inputs agree about direction, and direction is the easy half. On this trade the desk was at 86 percent certainty one evaluation before it entered at 67, and the 67 is the number that had money behind it.
Four instances in three weeks now follow this shape, all winners. It is not a quirk of one setup. It is what happens when a system scores entries rather than opinions.
From the desk
Cable was the instrument that broke August. The long book took ten trades, won two, and gave back 6.18R, more than double the entire month's net loss, while the short book lost a negligible 0.11R across two trades. On August 27 every trader was split into independent long and short books, the GBPUSD long book came off, and the short book, which had done nothing wrong, kept trading.
This is that short book, and it has now won both of its September trades: the September 2 short for +1.02R and this one for +2.04R.
Six winners this week, all of them shorts, into a week where yields added ground every session and finished above 4.92, VIX ran to eighteen, and Brent broke past 104. When a tape looks like that, the honest thing for a desk to report is that it took the side the market was paying rather than that it found anything clever.
The year stands at +28.75R across 220 trades through the August close, and September sits on top of that as an open month.
At a Glance
What this teaches about AI-driven trading
Why refuse a setup nine times and then take it at a lower score?
The nine scored the bearish thesis, which London had completed before New York opened. The tenth scored an entry, which needed a rebound into resistance to exist at all. The lower number also prices in reservations the thesis does not carry: a 5-minute bouncing off oversold and a VIX at 18 making counter-moves sharper. It is a narrower measurement of a specific position.
What does +2.04R mean in dollars?
R is the trade's risk unit, the distance from entry to stop, here 17.1 pips. On a $100,000 account risking 2% per trade, 1R is $2,000, so +2.04R is roughly $4,080. Reporting in R rather than pips keeps results comparable across account sizes and across instruments quoted in completely different units.
What would selling at the first 85 have cost?
Price was near the 1.3491 intraday extreme with no structural level close enough to anchor a stop. The choice would have been a stop wide enough that 34.9 pips returned well under 1R, or one close enough to be removed by the bounce that began two minutes later. Both are worse than waiting thirteen minutes.
Has this pattern happened before?
Four times in three weeks, all winners. Cable on September 2 entered at 62 after four refusals at 79 to 84. The Dow on September 8 entered at 76 after two at 83. The Dow on September 9 entered at 71 after one at 88. This trade entered at 67 after nine in the eighties, the most extreme instance so far.
How do you count R on a trade that reaches a third target?
This case study reports full potential, the R distance to the furthest target price reached, here the third target at 1.34835. Our weekly, monthly and year-to-date recaps use a stricter TP1 baseline crediting only the first target, so this trade enters those totals at +0.81R. The two numbers are deliberately different and we never mix them.
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
Four Shorts in Forty-Eight Minutes, on One Reading
Between 14:20 and 15:08 the desk sold Cable, the Nasdaq, the S&P and the Dow. One macro read, four instruments, four stops, four winners. This is the Dow leg, and the question is what that correlation costs.
The Winner That Paid Less Than a Loss Would Have Cost
This trade was right and returned +0.67R. The target sat 14 points away and the stop sat 21, so being correct paid two thirds of what being wrong would have cost. It is the least flattering trade of the week.
Half the Stop, One Day Later: The Same Book at +2.32R
The Nasdaq short on September 9 risked 153.8 points to make 271. This one risked 76.8 to make 178. Same instrument, same direction, one day apart, and the smaller move returned a third more.