Weekly Losses, Jun 22-28: Three Honest Stops in a Winning Week
The week closed net positive at +3.64R across ten trades and a 70 percent hit rate. We still booked three full stops. Here is the unvarnished account of where e
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.
The desk gave back 3.00R across three stops this week against +25.66R YTD. Through Jun 29, 2026 the book stands at +25.66R for the year, with a static balance of $151,332.35 as of Jun 29, 2026. A losing column inside a winning week is not a contradiction, and we would rather show you the three trades that did not work than dress up the seven that did. For the full ledger, the wins included, see [this week's recap](/blog/weekly-recap-2026-06-22). What follows is only the loss side: three trades, each stopped for exactly one R, totaling a 3.00R give-back against a year that remains comfortably in the green.
Act one, the setup that looked clean
The week opened with the kind of structure we like. Levels were respected, pullbacks were orderly, and our first entries reached target without drama. By midweek the book was carrying a healthy positive balance and the win column was doing most of the talking. That is exactly the environment where a desk gets loose, and we want to be honest that the temptation to size up was present even as we held line.
Act two, the chop arrives
Around Jun 24 the tape on GBPUSD stopped trending and started ranging. A continuation entry that should have followed through instead reversed into our stop. A day later a pullback buy on US30 walked into resistance and did not hold. The signal was not that any single trade was wrong, but that the conditions had shifted underneath us while the surface still looked tradable.
Act three, the books close green anyway
When the week settled we were net +3.64R across ten trades with a 70 percent hit rate. Three stops, three full R lost, and a year-to-date figure that did not flinch. The arc here is ordinary on purpose. A system that wins seven of ten and still takes three clean losses is a system behaving exactly the way the math promises it will.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Jun 22 | 14:50 UTC | GBPUSD | Long | Claude Opus 4.7 | GBPUSD Bullish Pullback Continuation Long | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jun 24 | 15:10 UTC | GBPUSD | Short | Claude Opus 4.7 | GBPUSD Short — Bearish Continuation on Pullback | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jun 25 | 14:57 UTC | US30 | Long | GPT-5.5 | US30 NY AM pullback buy into 52660 support band | 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
All three losses shared a shape. Each was an entry that depended on a level holding or a move continuing, and in each case the thing we were leaning on gave way. The GBPUSD long on Jun 22 was a bullish pullback continuation that needed the prior support to act as a floor, and it did not. The GBPUSD short on Jun 24 was a bearish continuation that needed sellers to press, and they refused. The US30 long on Jun 25 was a pullback buy into a zone that needed buyers to absorb resistance, and the absorption never came.
The common thread is follow-through risk. Pullback and continuation entries are bets that an existing move resumes after a pause. When the pause turns into a reversal, the stop does its job and we are out for one R. None of these were structural errors in reading the chart. They were the ordinary cost of trading continuation in a week where continuation stopped paying.
Decision highlights
We honored every stop to the tick. Across all three losers there was no widening, no averaging down, and no hope-trade held past its level, which is the single most important number on a loss report even though it never shows up as a number.
We did not let a winning week loosen our sizing. With the book green by midweek the pull to add risk was real, and we held flat at the planned allocation, so the three stops cost exactly what they were designed to cost and nothing more.
We read the GBPUSD cluster as information rather than noise. Two stops on one instrument inside three days told us the pair had gone range-bound, and we tightened our willingness to take continuation entries there for the remainder of the window.
Six instruments, six stories.
EURUSD: no losses this week, entries behaved and reached target without incident.
All EURUSD this week →GBPUSD: the week's busiest and choppiest name, two of the three stops landed here, a Jun 22 long and a Jun 24 short, both continuation bets that the range refused to pay.
All GBPUSD this week →US30: one stop, a Jun 25 NY AM pullback buy into resistance that did not hold for a clean minus one R.
All US30 this week →Loss of the week: US30 Long · -1R
Losses worth learning from
US30 Long, Jun 25, a grade-B pullback that did not hold
what was right
The location was defensible. We bought a New York morning pullback at a level that had given a clean bounce earlier in the session, and the entry trigger fired on a recognizable reversal candle. Risk was defined before we clicked, and the stop sat below structure where a break would genuinely invalidate the idea.
what was wrong
We were buying into resistance overhead rather than into open space. The pullback was real but the path upward was congested, and that congestion meant any push needed more buying pressure than a normal continuation. We gave the setup a grade B, and in hindsight the overhead supply should have pulled it lower.
what we keep
The discipline holds. We took a defined-risk entry at a sensible level, honored the stop, and lost exactly one R. The lesson is narrow: a pullback buy into a resistance shelf needs cleaner confirmation than one we would take into clear air.
GBPUSD Short, Jun 24, a grade C+ continuation that failed
what was right
The bias matched the lower-timeframe structure at the moment of entry. Price had been pressing lower, our continuation trigger fired on a retest, and risk was defined and small. We took the trade for the same reason we take most continuations, the path of least resistance looked downward.
what was wrong
The pair had already started ranging and we read the pause as a continuation rather than a top. Sellers did not press. The grade C+ reflects the marginal quality, and a choppier filter would have kept us out, especially with another GBPUSD stop already on the books that week.
what we keep
Defined risk, honored stop, one R. The keeper is the instrument signal: a second GBPUSD loss in three days is the tape telling us to stand down on that name until it trends again.
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
The dollar figures keep the give-back honest. Starting from $100,000 at 2 percent risk, the year-to-date result is $151,332.35 static versus $162,280.25 compounded. This week's 3.00R give-back is roughly $6,000 of that figure, which is real money and also a small fraction of a book that remains well in the green for the year.
The gap between static and compounded is the entire argument for disciplined sizing. Static holds the risk-per-trade flat against the original base, which is conservative and easy to reason about. Compounded lets the unit grow with the account, which is why the same trade history produces a larger number. We report both so the give-back is never flattered by the more aggressive math. Three honest stops cost about $6,000 against a year that has earned far more, and that proportion is the point.
What we're tuning
The one adjustment worth making is on continuation entries when a single instrument starts repeating losses. The GBPUSD cluster is the tell. When a pair hands us two stops inside three days, the prior is that it has gone range-bound, and we want our filter to demote continuation setups on that name until trend structure returns.
The second tune is on pullback buys into overhead resistance. The US30 stop was a defensible idea in a poor location, and the fix is not to stop taking pullbacks but to require cleaner confirmation when the path forward is congested. Neither change alters the system. Both tighten the conditions under which we are willing to press a continuation.
What the numbers actually mean
Three full stops in a ten-trade week feels heavy when you read them in a column, but it is statistically unremarkable for a system that wins a little under sixty percent of the time. Our hit rate sits at 59.69 percent across 129 trades since inception. At that rate, a cluster of three losses inside a ten-trade window is not a warning sign. It is a routine draw from the distribution, and any week is roughly as likely to deliver it as not.
Van Tharp's framing on R-multiples is useful here because it strips the dollars out of the question. Each of these three trades lost exactly one R, the unit of risk we defined before entry, which means the loss side of the week is fully described by a single number: minus three R. There is no hidden tail, no oversized stop, no trade that quietly cost more than planned. When every loss is capped at one R, the only thing that varies is how many of them cluster together, and clustering is governed by probability rather than skill.
Schwager's writing on losing streaks makes the same point from the other direction. Strings of losses are not anomalies to be explained away, they are guaranteed features of any edge that wins less than one hundred percent of the time. A system with our hit rate will produce runs of consecutive losers as a matter of course, and our own longest losing streak to date is just one, which if anything suggests the variance has been kind so far. The honest caveat is sample size. At 129 trades the numbers are directional, not settled, and we hold them loosely. What we will not do is read three ordinary stops as a signal that the system is broken, because the distribution we signed up for guarantees weeks exactly like this one.
At a Glance
Drawdown questions
How can a week be a winning week if it booked three losses?
Because the seven trades that worked outweighed the three that did not. The week closed net +3.64R across ten trades at a 70 percent hit rate. A losing column inside a winning week is normal for any system that does not win every trade, and the net figure is what compounds the book over time.
Why were two of the three stops on the same instrument?
GBPUSD was the busiest and choppiest name on the board that week. It went range-bound while still looking tradable on the surface, so two continuation entries that needed follow-through instead reversed into their stops. A cluster of losses on one instrument usually means the instrument has changed character.
What did the desk actually lose in dollars?
The 3.00R give-back is roughly $6,000 measured against the static figure. Each of the three trades was capped at one R, the risk unit defined before entry, so there was no oversized loss hiding in the column. The year-to-date book remains at $151,332.35 static and $162,280.25 compounded.
When does a string of losses mean the system is broken?
Not at three stops in a ten-trade week. With a 59.69 percent win rate over 129 trades, short runs of losers are a guaranteed feature of the edge, not a defect. The longest losing streak on record is one, and the system drawdown sits near 2.18 percent, which is shallow by any reasonable measure.
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: +25.66R YTD across 129 trades, see stats strip.
From the SkyAnalyst Journal
The Week the Tape Turned: How We Read a Bearish Open and Bought the Rotation
A two-regime week on the desk. We pressed shorts into an early-week slide, then flipped long as the rotation took hold, banking +3.64R across ten trades at a 70% win rate.
How Falling Yields Set Up a Clean NAS100 Pullback Long
Case study #100. With the 10Y compressing to fresh five-day lows, the desk passed on the chase and bought NAS100 into a VWAP and Fibonacci confluence. A clean TP1 win.
GBPUSD Long: A Post-Data Pullback at the 1.3200 Confluence
Case study #99: cable flushed to yesterday's low on a heavy London session, then reclaimed VWAP into the 1.3200 confluence. We waited for the reclaim, not the low, and banked +0.83R at TP1.