Three stops, minus 3R, and the prior-high lesson we paid for twice.
Three losses at -1R each, a maximum drawdown of 2.59% from Wednesday's peak, and two US500 longs that asked untested prior highs to hold. The books, reviewed in
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.
Three trades stopped out this week, and this article is about those three. The desk finished the Jul 6-12 window net positive — +1.19R across eight trades — but the wins have their own articles. Here we review the other column: one EURUSD short on Monday and two US500 longs on Thursday and Friday, each closed at exactly -1R, for a combined give-back of 3R. The give-back was orderly. Monday's loss took the simulated account to its low of $98,000 before Tuesday and Wednesday rebuilt it to the week's peak of $103,032.57; the two US500 stops then pulled it back at most 2.59% off that peak before Friday's final winner recovered most of the ground. No loss exceeded its budgeted 1R. No losing streak ran past one. By the standards we publish every week, this was a losing column doing what a losing column is supposed to do — costing exactly what was underwritten, and nothing more. The context is the ledger. Through Jul 13, 2026, the system stands at +28.48R YTD, which on a $100,000 simulated account at 2% static risk is $156,978.99. This week's 3R of losses is real money against that figure — and it is also the recurring, budgeted cost of a 60.42% win rate doing business. We publish the losses at full size because that is what any fund worth evaluating does with its books.
Act I — Monday's fee at the open
The week's first trade was its first loss. At 14:34 UTC Monday, the EURUSD desk shorted a New York AM VWAP rejection graded C+. The rejection it was selling never materialized — the touch had paused, not reversed — and the stop closed the position at -1R. The simulated account marked its weekly low of $98,000 within the first hour of the first session. What the loss bought was a clean read on the week's dividing line, visible only in hindsight at that point: the level EURUSD was leaning on had never been defended.
Act II — Two clean days rebuild the peak
Tuesday and Wednesday produced no losses at all — the GBPUSD and US30 desks banked the week's two strongest trades, and by Wednesday 14:10 UTC the account stood at its high-water mark of $103,032.57. We include the clean days in a losses review because they define the denominator: drawdown is measured from a peak, and Wednesday set the peak that Thursday and Friday would be measured against.
Act III — The same stop, two sessions running
Thursday at 14:06 UTC the US500 desk bought a pullback at yesterday's high, framed as opening-range-breakout continuation, graded C+. It stopped at -1R. Friday at 14:15 the same desk bought a cleaner version of the same idea — a pullback to the prior day high at the 7,550 zone, graded B — and stopped again. That second stop marked the week's maximum drawdown, 2.59% below Wednesday's peak, at $100,365.91. Ninety minutes later the US30 desk's final winner pulled the account back to $102,393.68, but the pair of US500 receipts is the reason this article exists.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Jul 6 | 14:34 UTC | EURUSD | Short | Claude Opus 4.7 | EURUSD NY AM VWAP rejection short | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 9 | 14:06 UTC | US500 | Long | Claude Opus 4.7 | US500 LONG — Pullback Buy at Yesterday's High / ORB Breakout | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 10 | 14:15 UTC | US500 | Long | Claude Opus 4.7 | US500 LONG — Pullback to Prior Day High / 7,550 Zone | B | -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.
Pattern of the week
Viewed from the losing column, the week's pattern is unusually legible. All eight of the week's entries were retracements into prior structure. The five that won entered at levels that had already been broken and defended at least once. The three that lost entered at levels whose authority was still hypothetical — a VWAP touch that had not yet printed a rejection, and two prior session highs that had never been tested from above.
That is not a coincidence of three trades; it is one entry criterion showing up three times in one week. The losses were not mispriced risk, late exits, or regime misreads. They were tuition paid to the same instructor on three visits, and the receipts all say the same thing: defense history is the property that separates a level from a line.
Decision highlights
Monday's EURUSD stop was honored without a revenge re-entry. The short thesis was never loudly invalidated — price chopped rather than reversed — and a discretionary desk might have re-shorted the same level that afternoon. The system did not, and when it returned to EURUSD on Thursday it was on the long side, at a proven breakout zone, for +0.67R.
Friday's US500 entry was taken with full knowledge of Thursday's stop, because the evaluation loop scores each setup on its own evidence rather than on the memory of yesterday's loss. The B grade was honest — the structure genuinely was cleaner — and the outcome was identical. We highlight the decision because re-engaging a valid read at fixed size is correct process even when it produces a second receipt.
Nothing escalated after the Wednesday peak. No position sized up to defend the high-water mark, no trade added to recover Thursday's loss faster. The Risk Agent held 2% flat through the give-back, which is the entire reason the week's maximum drawdown reads 2.59% instead of something with a whole number in front of it.
Six instruments, six stories.
One loss. Monday's NY AM VWAP-rejection short stopped at -1R; the desk partially recovered with Thursday's +0.67R pullback long, but the stop stands on this page.
All EURUSD this week →No losses this window. The desk's single trade, Tuesday's retracement short, banked +1.07R and ran through TP3.
All GBPUSD this week →Clean sheet. Two trades in opposite directions, two winners, nothing given back.
All US30 this week →No losses this window. One trade, one TP1 winner minutes after Monday's opening stop.
All NAS100 this week →Desk offline — retired from the rotation. No trades counted this window.
All USDJPY this week →Both index-long stops live here: pullback buys at untested prior highs on Thursday and Friday, -1R each, -2R total, and both of this week's teardowns.
All US500 this week →Loss of the week: US500 Long · -1R
Losses worth learning from
US500 long — pullback to the prior day high, 7,550 zone (Friday, grade B)
What was right: the trend context was genuinely constructive, the pullback was orderly, and the 7,550 confluence was real — this was the week's second-highest-graded entry, and the grade was defensible. What was wrong: the level itself had no history. The prior day high had never been tested from above, so the "support" the entry leaned on was an assumption about where buyers should be, not evidence of where they had been. The market probed straight through it, and the stop closed the trade at -1R with no drama. What we'd do the same: honor the stop at the structural level, size at fixed 2%, and take a B-graded setup when it prints — the criterion that needs work is upstream of all three, in what earns a level the right to be called support.
US500 long — pullback buy at yesterday's high / ORB breakout (Thursday, grade C+)
What was right: the opening-range-breakout framing was coherent and the risk was correctly bounded below the tested zone. What was wrong: everything Friday's trade got wrong, a day earlier and with a weaker grade. Yesterday's high had not been defended even once; volume on the approach was unremarkable; and the C+ grade was the system telling itself the evidence was thin. The trade stopped at -1R in the same session. What we'd do the same: the exit. What the pair of losses changes: a C+ read at an unproven level is now a named pattern in our review queue, not an isolated data point — twice in two days converts an anecdote into feedback.
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 drawdownActual | -3R | −$6,000 |
All six agents combined.
From the desk
Publishing a losses review on a week the desk finished up +1.19R might look like theater. It is the opposite. The habit of reviewing the losing column only on losing weeks is how trading operations end up narrating their own performance instead of measuring it. The three receipts above cost $6,000 of simulated equity at full size — $2,000 each — and they get the same scrutiny they would have gotten if the winners hadn't covered them.
The running account is the frame that keeps this honest in both directions. Through Jul 13, 2026, the ledger shows +28.48R since the Jan 12 inception: $156,978.99 on a $100,000 simulated account at 2% static risk, or $171,192.60 if the same sequence of trades is compounded with each position sized to the account as it grew. This week's give-back is about $6,000 of motion inside a figure that has added nearly $57,000 static since January. The compounded number being higher — by more than $14,000 — is worth one sentence of dwelling: it means the equity path has spent enough time above water that sizing to the growing account outran sizing to the starting one. Losing weeks that stay inside their budget are precisely what keeps that true.
We would rather show you three named, bounded, understood losses than a highlight reel. The highlight reel is in the recap; this is the part of the books most operations don't publish.
— The SkyAnalyst Team
What we're tuning
One candidate adjustment comes out of this window: weighting defense history in the confluence math for index pullback longs. Both US500 stops entered at prior highs that had never been retested from above; all five of the week's winners entered at levels with at least one prior defense. Had the evaluation loop required that single property, the losing column would have been one trade instead of three, and no winner would have been excluded.
Two occurrences is a sample, not a verdict, so nothing changes this week. The pattern goes on the review queue with a named trigger: if unproven-level entries keep underperforming defended-level entries at anything like this spread over the next month of published trades, the confirmation threshold for the former rises. Tuning from published evidence, at the speed of evidence, is the whole discipline.
What the numbers actually mean
The numbers above are the week's losses placed against the system's running record: a 60.42% win rate across 144 published trades, an average winner near +0.84R at the TP1 baseline, a maximum drawdown this window of 2.59%, and a longest losing streak of one. The useful question is not whether those numbers are good — it is what they make <em>expected</em>.
Start with the win rate. A 60% system loses four trades in ten as a matter of arithmetic, and those losses do not arrive politely spaced. Over a 144-trade sample, runs of three, four, and five consecutive losses are not warning signs; they are what the binomial math produces even when nothing is broken. Jack Schwager's interviews in the <em>Market Wizards</em> series return to this point repeatedly — the traders with durable records are distinguished less by avoiding losing streaks than by having position sizing that renders streaks survivable and psychologically boring. A week whose longest streak was one is, by that standard, a quiet week.
Van Tharp's framing in <em>Trade Your Way to Financial Freedom</em> is the second lens: think in R-multiples, not dollars or win rate alone. Every loss this week was exactly -1R — the unit of risk the system defined before entry. Expectancy comes from the full distribution of R-multiples over hundreds of trades, which is why we neither celebrate a +1.19R week nor apologize for a -3R losing column inside it. Both are single draws from the same distribution.
The last number worth framing is the 2.59% drawdown. At 2% risk per trade, three full-size losses cost roughly 6% of account if nothing wins in between; this week's winners meant the realized dip was less than half that. Drawdowns of this scale are the operating temperature of the strategy, not an alarm. What would concern us is a change in the <em>shape</em> of the losses — stops exceeding -1R, streaks clustering beyond what the win rate implies, or losses concentrating in a pattern we cannot name. This week's losses we can name precisely, which is the subject of the teardowns above.
At a Glance
Drawdown questions
Why publish a losses article on a week that finished positive?
Because the losing column deserves review on its own terms, not only when it outweighs the winners. Every legitimate fund reports drawdowns and loss attribution as a matter of routine; doing it only on bad weeks turns reporting into damage control. The wins from this same window are covered in the weekly recap — this page is the other half of the books.
Is a 2.59% drawdown something to worry about?
No — it is the operating temperature of a system risking 2% per trade. Three full-size losses cost about 6% if nothing wins between them; this week's winners kept the realized dip to less than half that. What would worry us is losses exceeding their -1R budget or streaks running beyond what a 60.42% win rate statistically implies. Neither happened.
Why are both teardowns the same instrument?
Because the two US500 stops are the week's actual lesson. They entered the same species of level — a prior high never tested from above — on consecutive sessions, with grades of C+ and B. One occurrence is noise; the same failure with a better grade a day later is feedback about the entry criterion, and that earns the full teardown treatment.
The system took the Friday US500 trade after Thursday's identical stop. Isn't that a flaw?
We consider it correct process with a losing outcome, which is a category retail traders systematically misjudge. Friday's setup was independently graded on its own evidence and the grade was honest — the structure was cleaner. Skipping valid setups because the last one lost is how win rates get flattered and expectancy gets destroyed. The fix under review is upstream: requiring defense history before a level qualifies at all.
What would actually make you change the system?
Published evidence at sample sizes that mean something. The defense-history pattern flagged this week goes on a named watch: if unproven-level entries keep losing at this spread over the next month of trades, the confirmation threshold rises. What we will not do is rewire the entry logic off a two-trade sample in either direction.
Trade with the system that publishes its drawdowns.
Subscribers receive every signal — winners and losers — three minutes before entry, with full reasoning.
Dollar figures are simulated on a $100,000 account at 2% risk per trade. Drawdown trajectories shown reflect a small window sample size and are not projections of forward performance. Past performance — including losses — is not a guarantee of future results. Actual subscriber P&L varies with account size and execution. YTD context: +28.48R YTD across 144 trades, see stats strip.
From the SkyAnalyst Journal
The entry we refused at 80% confidence and took at 78%.
Confidence is not a trigger. The system sat through evaluations at 80% and 82% because the pullback trigger had not printed, then entered a Dow long with fourteen minutes left before its own midday cutoff.
Two agents disagreed. The dollar index cast the deciding vote.
Our Macro Agent leaned bearish. Our Trend Agent read a breakout. The tiebreaker was the dollar index, falling below its 5-day average, and the result was a 23-hour EURUSD long that never traded a pip underwater.
The week two shorts carried what the index longs gave back.
Eight trades, five wins, +1.19R. A stopped-out Monday open, two retracement shorts that ran to TP3, and a US500 long thesis that cost 2R in two sessions. The week rewarded proven levels and punished aspirational ones.