Five longs stopped, two shorts paid: the week in review
Seven trades, five of them losses, a net give-back of 2.24R. Yet the setups graded A- for the week and both winners were shorts. Here is what the tape actually
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.
Some weeks the system is right about the market and wrong about direction, and this was one of them. Seven trades cleared our setup filter between July 13 and July 19. Two won, five lost, and the ledger closed at minus 2.24R, roughly minus 4,480 dollars on a 100,000 dollar account risking two percent a trade. The setups themselves graded A- for the week. The market simply refused to reward the long side. That distinction matters, so we will be precise about it. Every one of the five losses was a long. Both of the two winners were shorts. This was not a week where we misread structure or forced marginal entries. It was a week where the index tape faded every bounce we bought and rewarded both continuation shorts we sold. The grade was high and the result was negative, and those two facts sitting side by side are the whole story. Context keeps a week like this in proportion. Through Jul 20, 2026, the system has banked plus 26.24R since the January 12 inception. A 100,000 dollar account at two percent risk sits at 152,498 dollars on a static basis. This week gave back a slice of that. It did not dent the thesis.
Act I — Monday's clean short
The week opened the way we would have scripted it. On Monday the trend agent flagged a bearish continuation on NAS100 off a failed retest, graded it B+, and the short filled at 14:10 UTC. It ran straight to TP1 for plus 1.46R and lifted the simulated account to its high-water mark for the week, about 102,916 dollars. One trade in, we were green and the read looked easy.
Act II — the long-side grind
Then the tape turned on the buyers. Tuesday brought three long entries, US500, US30, and GBPUSD, and all three stopped at exactly minus 1R. Wednesday brought two more longs, US500 and US30 again, and both stopped as well. Five consecutive losses, none of them larger than a single R, walked the account from its Monday peak down to roughly 92,916 dollars by Wednesday afternoon. Nothing broke. The risk policy sized every one of those trades identically, so the drawdown was linear and shallow rather than jagged.
Act III — Friday's recovery short
Thursday we took nothing. No setup cleared the filter, and the system sat out a day in the middle of a losing streak rather than reach for a trade to feel better. Friday the pattern that had worked on Monday returned. A GBPUSD short filled at 15:32 UTC and paid plus 1.3R, lifting the account back to about 95,519 dollars and trimming the full week to minus 2.24R. The two trades that bracketed the week were both shorts, and both paid.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Jul 13 | 14:10 UTC | NAS100 | Short | Claude Opus 4.7 | NAS100 Short - Bearish Continuation off Failed Retest | B+ | +1.46R(TP1) | +$2,916(TP1) | TP1 hit · ★ Trade of the week | Read case → |
| Jul 14 | 14:37 UTC | US500 | Long | Claude Opus 4.7 | US500 LONG — Opening Range Breakout | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 14 | 14:40 UTC | US30 | Long | GPT-5.5 | US30 NY AM Pullback/Second-Chance Long | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 14 | 15:06 UTC | GBPUSD | Long | GPT-5.5 | GBPUSD London pullback buy | B | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 15 | 14:37 UTC | US500 | Long | Claude Opus 4.7 | US500 Bullish VWAP/Structure Pullback Buy | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 15 | 14:52 UTC | US30 | Long | GPT-5.5 | US30 pullback-long continuation | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 17 | 15:32 UTC | GBPUSD | Short | GPT-5.5 | GBPUSD SHORT | C+ | +1.30R(TP1) | +$2,603(TP1) | TP1 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.
Pattern of the week
The pattern of the week was direction, not setup quality. We bought five pullbacks and breakouts and the market stopped all five. We sold two continuation shorts and the market paid both. When the winners and losers sort that cleanly by direction, the lesson is not that the reads were wrong. It is that the index and currency tape was in a posture that punished dip-buying and rewarded selling strength, and our long setups kept arriving into it.
There is a second, quieter pattern worth naming. On both Tuesday and Wednesday, the US500 long and the US30 long fired within minutes of each other. Those two instruments are close cousins, both broad US equity exposure, and taking them long in the same session is close to doubling a single directional bet. Twice they stopped together. That is correlation showing up as a cost, and we return to it below.
Decision highlights
The risk policy held flat through a five-trade losing streak. Every one of the five losses was sized to exactly minus 1R, with no revenge sizing after the third or fourth stop. That discipline is why the streak cost 5R gross rather than the open-ended damage a discretionary trader chasing the drawdown would have booked.
The system did not develop a directional bias mid-week. Even as five straight longs stopped, it still took the Friday GBPUSD short at face value and it still holds no standing preference for one side. It read each setup on its own structure, which is precisely why both winners were shorts while the losers were longs.
Thursday produced no trade at all. In the middle of a losing streak, with the account near its weekly low, the temptation to force a recovery entry is real, and the system felt none of it because it does not feel. No setup cleared the filter, so nothing was taken, and the streak ended on its own terms the next day.
Six instruments, six stories.
EURUSD: No trades this week. No setup cleared the filter, and the pair spent the week outside our entry criteria.
All EURUSD this week →GBPUSD: Two trades, net plus 0.3R. A Tuesday long stopped for minus 1R, but Friday's short paid plus 1.3R and turned the pair green for the week.
All GBPUSD this week →US30: Two trades, both long, both stopped for a combined minus 2.0R. Half of the week's index drag came from here, and both entries fired alongside a correlated US500 long.
All US30 this week →NAS100: One trade, the short of the week. It ran to TP1 for plus 1.46R, the best result of the window and the Monday high-water mark.
All NAS100 this week →USDJPY: No trades this week. The pair stayed outside our setup criteria the entire window.
All USDJPY this week →US500: Two trades, both long, both stopped for a combined minus 2.0R. The other half of the index drag, twice paired with a US30 long in the same session.
All US500 this week →Win of the week: NAS100 Short · +1.46R
Loss worth learning from
The instructive cluster this week was the four index longs, two on US500 and two on US30, that stopped across Tuesday and Wednesday. Each was a pullback or breakout continuation long, and each graded C+, marginal but acceptable by our filter. The read in every case was the same: buy strength into an intraday uptrend and lean on prior structure for the stop. The market's answer was also the same every time. It faded the bounce, tagged the stop at minus 1R, and moved on. Four identical setups, four identical outcomes, no single one of them a mistake in isolation. What made the cluster expensive was not any one entry but the fact that the correlated pair kept arriving together.
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 |
|---|---|---|
| Window netActual | -2.24R | −$4,480 |
All six agents combined.
From the desk
A note, before we move on.
We publish the losing weeks with the same detail as the winning ones because the losing weeks are where a track record earns its credibility. Anyone can show you the green. The question that matters to someone deciding whether to trust a system is what the red looks like, and this is what ours looks like: five stops of exactly 1R each, a day of sitting out rather than forcing, and two shorts that trimmed the damage without a hint of heroics.
Hold the week against the year and the shape of the thing becomes clear. Through Jul 20, 2026 the system has banked plus 26.24R since inception. On a 100,000 dollar account at two percent risk, that is 152,498 dollars on a static basis, where every trade risks the same fixed amount, and 163,402 dollars compounded, where each trade risks two percent of the growing balance. The gap between those two figures, almost 11,000 dollars, is what disciplined sizing produces over 151 trades. This week subtracted about 4,480 dollars from that path. The path did not change direction.
We would rather you saw this week than a curated one. A system that only ever shows you its Mondays is selling you a story about trading. We are trying to show you a system, and a system includes the Tuesdays and Wednesdays where the long side simply does not work.
— The SkyAnalyst Team
What we're tuning
There is nothing structural to fix in the losses themselves. Each was a graded setup that stopped at its planned risk, which is the system working as designed rather than failing. The flat 1R policy did its job and kept a five-loss streak to a linear 5R rather than a spiral.
The one honest observation is correlation. Twice this week a US500 long and a US30 long ran in the same session, and twice they stopped together, effectively converting two positions into one doubled directional bet. We do not currently enforce a portfolio-level correlation check across the two index books, and this week is a clean argument for one. It would not have turned the week green, but it would have kept us from paying the same directional wrong twice at once.
At a Glance
Week at a glance
Why publish a losing week at all?
Because the losing weeks are the honest test of a track record. A system that only shows its winning weeks is curating, not reporting. We log every trade the same way, so a minus 2.24R week appears here in the same detail a strong week would, with the full trade index and the reasoning behind each entry.
How did the setups grade A- while the week still lost money?
Setup grade measures the quality of the read at entry, not the outcome. This week the reads were sound but the market punished the long side, and five of our entries were longs. High-grade setups still lose when direction goes against them. Over a large sample those grades correlate with results, but any single week can break the link.
What kept five straight losses from becoming a real drawdown?
A fixed risk policy. Every trade risked the same 1R regardless of how the week was going, so five losses cost 5R in a straight line rather than an accelerating spiral. There was no revenge sizing after the third or fourth stop. That flat discipline is the difference between a rough week and a damaging one.
Why were both winners shorts and all five losses longs?
The tape was in a posture that faded strength and punished dip-buying, and our long setups kept arriving into it while our two shorts sold into that same weakness. The system holds no directional bias, so it took each setup on its own structure. The clean split by direction is a fact about the market that week, not a preference in the system.
Does a losing week change how the system trades next week?
No. The risk policy, the setup filter, and the agent coordination are unchanged by a single week's result. The one note we took was about correlation, taking US500 and US30 long in the same session doubled a directional bet twice, and that is a portfolio observation rather than a change to how any single trade is judged.
Get next week’s trades before they print.
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.
From the SkyAnalyst Journal
The short the system refused to chase, and why it paid
London sold Cable hard, and the tempting play was to hit the lows. The system did the opposite. It waited for price to retrace to VWAP, shorted the failure to reclaim, and never went a single pip underwater.
Five stops in a row: what a losing week actually costs
Five trades stopped in a row for a combined 5R, and every one of them was a long. None was a mistake in isolation. This is what a normal losing streak looks like inside a system that still carries a real edge.
The first trade Claude Fable 5 ever took as a beta model.
Claude Fable 5 is a new beta trader running in the SkyAnalyst sandbox, outside the tracked Pro and Lite record. Its very first trade was an eight-of-eight-confluence EURUSD pullback long that ran clean to TP3.