The gate cleared two setups in five sessions and both stopped inside six minutes of entry. Net minus 2R against +19.60R YTD, and a GBPUSD pullback long entered
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.
Two trades. Two stops. Net minus 2R. That is the entire scoreboard for Jun 8-14, the thinnest fully-counted week on the desk's books this year, and there is no winner-of-the-week section below because there is no winner to put in it. Through Jun 15, 2026, the system has banked +19.60R YTD from Jan 12 inception; a $100,000 simulated account at 2 percent risk per trade sits at $139,204.12 on the static path, or $144,227.48 on the compounded path. The detail worth your time is not the size of the give-back, it is the shape of it. Both losses were longs graded C+, both entered on premises the tape had not yet confirmed, and both were invalidated inside six minutes of entry. A third position, a GBPUSD pullback long filled at 15:46 UTC on Friday, was still open when we went to press. It is deliberately excluded from this week's numbers and will be counted in the window it closes. The calendar arc, the instrument grid, and what we are tuning all follow from those three trades, and this is a short story this week, so we will tell it properly.
The week opened with two sessions of nothing. Candidates were flagged on both days; none cleared the confluence threshold; no entries were sized. At the desk's YTD cadence of roughly five trades a week, a two-session drought is rare enough to notice, and the temptation in a system with an audience is to manufacture activity. The system does not have that setting. The cost of a confluence gate is silence when the tape offers nothing, and the first two sessions of this window were the bill.
The first entry filled at 15:13 UTC on Wednesday, a GBPUSD long at 1.33952 on a post-CPI second-chance read at a VWAP and Fibonacci cluster, stop just over five pips below at 1.3390. The stop was hit five minutes and thirty-six seconds later. A flat 1R, taken exactly as sized, and a stop distance that the [week's losses teardown](/en/blog/weekly-losses-2026-06-08) examines hard: five pips on cable in post-CPI tape sits inside the session's noise band, and the thesis never got tested. Thursday passed without an entry.
Friday at 14:40 UTC the NAS100 trader went long at 29,519 on a post-reversal continuation read, stop at the reversal structure 119 points below. Four minutes and forty seconds later the week's second stop printed, and the simulated curve touched its 4 percent trough at $96,000. Then at 15:46 UTC the GBPUSD trader filled a second long on the pair, a VWAP and EMA pullback read, just over 48 hours after Wednesday's GBPUSD stop and in the same direction. That position was still open at press time. It is the open question the week ends on, and next week's books will answer it.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Jun 10 | 15:13 UTC | GBPUSD | Long | Claude Opus 4.7 | GBPUSD Post-CPI Second-Chance at VWAP/Fib Cluster | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jun 12 | 14:40 UTC | NAS100 | Long | Claude Opus 4.7 | NAS100 Long - Post-Reversal Continuation | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
Dollar figures are simulated on a $100,000 account at 2% risk per trade. Actual subscriber P&L varies with account size. Past performance is not a guarantee of future results.
With two closed trades there is no recurring setup to name, and we will not invent one. What the week does offer is a continuation of the meta-pattern we named in the Jun 1-7 recap: winners enter after a confirmed second leg with volume, and losers enter on premises the tape has not yet proven. Both of this week's losses fall on the wrong side of that line. The Wednesday cable long was a second-chance read priced on a re-offer that never came; the Friday NAS100 long was a continuation entry sized before the continuation printed.
The speed of both invalidations is the tell. Stops hit inside six minutes mean the entries sat inside the market's immediate noise, not at the edge of a developing move. On a two-trade sample this is rhyme, not statistics, and we are treating it accordingly: it earns a line in the tuning queue, not a rewrite of the playbook. But it is the second consecutive window where confirmation, or the lack of it, sorted the winners from the losers, and we said last week that patterns that survive across windows are the ones we act on.
The stand-downs were the week's best decisions. Monday, Tuesday, and Thursday produced no entries because nothing cleared the gate, and nobody lowered the gate to generate content for this page. On a window where both gated entries stopped, the trades not taken are the only line of the ledger that reads well, and that is worth saying without embarrassment.
The Risk Agent sized Friday's NAS100 entry at the same flat 1R as Wednesday's cable trade, with the week already 1R down. No revenge sizing, no fear discount. The give-back stayed at 2R precisely because the second trade was treated as statistically independent of the first, which is the entire point of fixed fractional sizing.
The Friday GBPUSD re-entry at 15:46 UTC is the judgment call we cannot grade yet. Same pair, same direction, 48 hours after Wednesday's stop: exactly the shape the proposed 72-hour cooldown gate exists to scrutinize, and the gate was not live at entry because its back-test is still running. The position was open at press time. Next week's books get the result, and we will write it against the gate's logic either way.
SkyAnalyst runs multiple foundation models in parallel across its four-agent system. When two models trade the same instrument in the same week, the results are directly comparable. This is that comparison.
Same signals, same risk framework, different foundation model.
EURUSD: no trades this week. Nothing on the pair cleared the confluence threshold across five sessions, and the trader stood down.
All EURUSD this week →GBPUSD: one closed trade, net minus 1R. Wednesday's post-CPI second-chance long stopped in under six minutes. A second long, filled Friday at 15:46 UTC, was still open at press time and is not counted in this window.
All GBPUSD this week →US30: no trades this week. The index never fit the setup criteria; no entry was flagged past threshold.
All US30 this week →NAS100: one trade, net minus 1R. Friday's post-reversal continuation long stopped in four minutes and forty seconds at the reversal structure, printing the week's equity trough.
All NAS100 this week →USDJPY: no trades this week. After last week's stopped pullback long, the pair offered no read that cleared the gate.
All USDJPY this week →US500: no trades this week. The trader flagged nothing past threshold; the index sat outside our criteria all window.
All US500 this week →Both of the week's losses are torn down properly in the [Jun 8-14 weekly losses report](/en/blog/weekly-losses-2026-06-08), published alongside this recap. The short version: Wednesday's GBPUSD second-chance long died on a stop placed five pips from entry in post-CPI tape, inside the session's noise band, so the thesis was never actually tested. Friday's NAS100 continuation long placed its stop at real structure 119 points away, and the speed of that stop says the problem was entry timing, not stop placement: the continuation the trade was priced on had not yet confirmed.
The distinction between those two failure modes matters more than the shared headline of fast stops. One is a stop-width problem, and the tuning queue now carries a volatility-keyed stop-width floor for it. The other is a confirmation problem, the same one last week's window surfaced, and the fix there is requiring the confirmed second leg before continuation entries are sized. Two losses, two different lessons, one common thread: both trades were early, and early is just wrong with better grammar.
Each trade risks +$2,000 (1R). The system's actual scale-out behavior may differ, see disclaimer.
| Scenario | R-multiple | Profit on $100k |
|---|---|---|
| Window netActual | -2R | −$4,000 |
A $100,000 simulated account at 2 percent risk per trade entered the week at $143,204.12 on the static path and $150,174.39 on the compounded path, and exits at $139,204.12 and $144,227.48. The static give-back is exactly $4,000; the compounded give-back is approximately $5,947, larger because the compounded path risks 2 percent of a bigger balance. That asymmetry is the honest face of the same property that has compounding $5,023.36 ahead of static through Jun 15, 2026. Fixed-percentage sizing compounds gains and drawdowns alike, and we publish both paths because showing only the flattering one would be marketing, not measurement.
The week itself was a test of process under scarcity, and the process held: three sessions of stand-down without manufactured activity, flat sizing after a loss, stops honored where they were set, and an open position left out of the published numbers because unrealized results are not results. The [losses report](/en/blog/weekly-losses-2026-06-08) has the trade-level autopsy. The GBPUSD position that was open at press time will be in next week's books at whatever it settles to, and next Monday the gate opens on a new tape.
— The SkyAnalyst Team
The cooldown gate proposed in last week's losses report, a half-grade downgrade on any same-instrument, same-direction re-entry within 72 hours of a stop, remains in back-test against the YTD ledger and ships or dies next week against the criteria already published. This week added a live specimen to its evidence file: Friday's GBPUSD pullback long, filled 48 hours after Wednesday's stop on the same pair, graded C+ at entry, exactly the read the gate would have cut to C. Whatever that position resolves to, it becomes part of the gate's record.
New to the queue this week is a stop-width floor keyed to session volatility: when the structural stop distance falls inside the session's realized noise band, the entry waits for a retest at calmer tape or passes entirely. Wednesday's five-pip cable stop in post-CPI conditions is the motivating case. Like the cooldown gate, it is mechanical and testable on the historical ledger, and it will be back-tested before it touches a live evaluation cycle. We do not change the playbook on a two-trade sample; we let the sample nominate hypotheses and make the ledger confirm them.
Because only two setups cleared the confluence threshold in five sessions. The desk's gate requires multiple independent agents to agree before an entry is sized, and on a quiet or unreadable tape that agreement simply does not arrive. The system has no minimum activity quota, and a thin week is the gate working as designed, not a malfunction.
There were no wins. Rather than restructure the page to hide the empty column, we are saying it plainly: zero winners, two losses, net minus 2R. At the YTD win rate of 57.8 percent, a two-trade week with no winners is roughly an 18 percent outcome on a back-of-envelope independence assumption, uncomfortable but well inside normal.
Because it has not resolved. An open position is a risk, not a result, and counting unrealized trades would let any week's published numbers depend on the hour we hit publish. It will be counted, win or lose, in the window it actually closes. We flagged it here because pretending it does not exist would be its own kind of dishonesty.
Not structurally. The YTD ledger stands at +19.60R across 109 trades, and this week moved it by exactly the two stops it recorded. The numbers that would signal a problem are a sustained drift in win rate or average R across hundreds of trades, and nothing in this window's sample, which is two trades, can move those.
TP1 baseline. Winners are credited at the TP1 R distance and losers at minus 1R, so the periods are comparable across the year. The single-trade case studies publish full-TP figures for the same trades, which is why their numbers differ from the snapshot used here.
Subscribers receive the same pre-trade AI analysis three minutes before entry.
We project the recap totals using a TP1 exit on every winning trade. This is the simplest baseline for comparing across periods. Traders running their own scale-out, trail, or TP2/TP3 hold strategies will see different totals. Dollar figures are simulated on a $100,000 account at 2% risk per trade. Actual subscriber P&L varies with account size and execution. Past performance is not a guarantee of future results.
Two closed trades all week, both C+ longs, both stopped inside six minutes of entry. The desk gave back 2R against a +19.60R YTD line, and a third entry was still open when this report went to press.
Seven trades. Three losses inside the dollar bid that broke our Monday open and Wednesday cable. One bounce-rejection short on NAS100 that paid the week. A 57.1 percent win rate that owes most of its margin to one outlier.
Three losses, two of them on the same instrument, same direction, forty-six hours apart. The desk gave back 2.5R against a +21.60R YTD line, then closed the week at a fresh equity peak.