Weekly Losses, Jun 15-21: One Stop in a Winning Week
A net positive week with a single contained loss. We publish our losses the way every legitimate fund does, even when the books are green, because a -1.00R stop
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.
We closed the week of June 15 net positive and we still want to talk about the one trade that did not work. The desk gave back 1.00R this week against +22.02R YTD, a give-back small enough to disappear in the noise of a green week and important enough to document anyway. Through Jun 22, 2026, the book stands at +22.02R YTD, which on a $100,000 account risking 2% per trade is $144,042.87 static. The week itself ran +2.64R across six trades at an 83.3% win rate, and exactly one of those six was a loss. We covered the wins in [this week's recap](/blog/weekly-recap-2026-06-15). This report is the other half of the ledger, the part most operators quietly skip.
Act one, the setup
The week opened with the Trend Agent scanning for intraday continuation. On June 15 it surfaced a VWAP pullback long on GBPUSD, a structure where price pulls back into the volume-weighted average and the prior support zone, then resumes higher if the level holds. The setup carried a C+ grade, which in our framework means tradable but not premium. The thesis was simple. If the VWAP and the support beneath it held, the long had room. If they did not, the trade was wrong and we would know quickly.
Act two, the execution
We entered at 14:30 UTC at 1.34315 with a stop at 1.34145, a clean 17.0 pips of risk. That sizing is not a function of how much we liked the trade. Every entry on the desk risks the same fixed amount, so a C+ setup and an A setup cost the same when they fail. For the next several hours the position sat without resolving. No take-profit was reached. The level it leaned on was the entire trade, and the level was still being tested.
Act three, the outcome
At 19:05 UTC, price reclaimed below the VWAP and the support zone. That reclaim was the invalidation. The reason we were long had stopped being true, so the stop did its job and we exited at 1.34145 for -1.00R. One loss, one full unit of risk, no improvisation. Around it, US30 booked two wins, NAS100 booked two wins, and EURUSD booked one. The week ended five wins to one stop, net +2.64R.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Jun 15 | 14:30 UTC | GBPUSD | Long | Claude Opus 4.7 | VWAP Pullback Long | 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.
Pattern of the week
The pattern behind the only loss was a VWAP pullback long that failed when its level did not hold. This is a high-frequency setup for us and usually a productive one. Price trends, pulls back into the volume-weighted average price where intraday participants tend to re-engage, and continues. The edge lives entirely in whether the pullback level holds as support.
On this GBPUSD entry it did not. After several hours of the level being tested rather than respected, price reclaimed below the VWAP and the support zone underneath it. Once that happened, the structure that justified the long was gone. There was no second thesis to fall back on, and we did not invent one. The pattern is sound across a large sample. The cost of trading it is that a fraction of these pullbacks resolve as reclaims instead of bounces, and when they do, the stop is the correct outcome rather than the avoidable one.
Decision highlights
We sized the C+ GBPUSD long to the same fixed risk as any A-grade setup, which is deliberate. Conviction does not change position size on this desk, so a low-grade setup that fails costs exactly one unit and never more, and that rule is precisely why this loss stayed at -1.00R instead of becoming something larger.
We did not add to the position or move the stop while the level was being tested for several hours. The trade was a single bet on whether the VWAP and support held, and when the answer arrived we took it at the predefined level rather than negotiating with a position that had already been invalidated.
We let USDJPY and US500 sit out the week with no trades. Not forcing entries on instruments that produced no qualifying setup is a decision in its own right, and a week where two instruments cleanly abstain is healthier than one where we manufacture activity to look busy.
Six instruments, six stories.
GBPUSD: the week's only loss, a C+ VWAP pullback long stopped for -1.00R when the level reclaimed lower.
All GBPUSD this week →Loss of the week: GBPUSD Long · -1R
Losses worth learning from
GBPUSD Long, the week's only stop
A VWAP pullback long entered June 15 at 14:30 UTC, 1.34315 with a stop at 1.34145, 17.0 pips of risk. Stopped at 19:05 UTC for -1.00R when price reclaimed below the VWAP and support. Setup grade C+.
what was right
The risk was fixed and small. The stop sat where the thesis died, not at an arbitrary distance. When the level failed, we exited cleanly with no averaging down and no widened stop. The loss landed at exactly one unit.
what was wrong
The setup was C+, the lower end of what we trade. The long leaned entirely on a support and VWAP zone that was still being tested rather than clearly holding before we committed.
what we keep
The pattern works across a large sample, so we keep trading it. We keep the fixed risk. We keep exiting at invalidation without improvising a second reason to stay long.
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 | -1R | −$2,000 |
All six agents combined.
From the desk
The arithmetic is worth making concrete. A $100,000 account risking 2% per trade sits at $144,042.87 static and $151,154.08 compounded through Jun 22, 2026. This week's 1.00R give-back is about $2,000 of that on the static basis, a rounding line in the context of a +22.02R year.
That gap between static and compounded, more than $7,000, is what disciplined fixed-fraction sizing produces over time. The same rule that grows the compounded figure is the rule that capped this week's loss at one clean unit. We size every trade the same way whether we love the setup or merely tolerate it, which is exactly why a C+ entry that failed cost about $2,000 and nothing more. Losses are the price of the edge, and the job is keeping each one small and the count of them honest.
What we're tuning
Going forward we are watching the entry timing on VWAP pullback longs more closely than the setup grade alone. The recurring failure mode is committing while the support and VWAP zone is still being tested rather than waiting for a clearer signal that the level is holding. The thesis on this GBPUSD trade was correct in shape and wrong in timing, and the difference between a bounce and a reclaim often comes down to whether we entered into confirmation or into a level that was still up for grabs.
We are not changing the risk, the stop placement logic, or the decision to trade C+ setups at all. Those are working as designed and the loss stayed at exactly one unit because of them. The single variable under review is patience at the entry on this specific pattern, and we will judge any adjustment over a meaningful sample rather than on one stop.
What the numbers actually mean
A single -1.00R stop is statistically unremarkable, and the framework for reading it matters more than the loss itself. Our system has run a 58.82% win rate over 119 trades since inception, with a current R target around 0.73. In a system that wins a little under three times in five and loses the rest, individual losses are not anomalies to explain away. They are the expected denominator of the win rate. A week with one stop in six trades is, if anything, on the favorable side of what the long-run rate predicts.
The useful lens here is the R-multiple view that Van Tharp built his work around. He argued that a trade's outcome is best understood not as a dollar amount but as a multiple of the risk taken, which turns every result onto the same scale. On that scale this loss is a clean -1.00R, the smallest complete loss the system can produce, with no slippage past the stop and no widening. Schwager's interviews with professional traders make the companion point about losing-streak distributions. A losing streak of length one, which is the longest streak we recorded this week, is not just common in a 58.82% system. It is statistically guaranteed to recur often, and streaks several times longer are mathematically expected over a large enough sample. The current system drawdown of roughly 1.95% is shallow precisely because the streaks have stayed short so far.
We will not publish a computed Kelly fraction from these numbers, because doing so would imply a precision we have not earned. With 119 trades the sample is real but still small, and win rate and average R will keep moving as the count grows. The honest statement is that one -1.00R stop in a 58.82% system is exactly what the distribution predicts, and reading it as anything more dramatic would be a misreading of the math.
At a Glance
Drawdown questions
How can a week be reported as a loss week when it was net positive?
This is not a loss week, it is a losses report. We publish a dedicated teardown of every trade that lost, regardless of whether the overall week was green or red. The week of June 15 finished net +2.64R across six trades, and this document exists to account for the single stop inside it rather than to characterize the week as bad.
Why publish a -1.00R stop at all when the books are green?
Because legitimate funds publish their losers in winning periods too, and hiding losses behind a positive headline is how track records become unreliable. A loss is research data about which setups fail and why. Documenting the one stop in a five-win week keeps the record complete and keeps us honest about the cost of the edge.
What does a C+ setup grade mean and why trade it?
A C+ grade marks a setup as tradable but not premium, the lower end of what our framework will take. We trade these because, across a large sample, the pattern still carries a positive expectancy even though a higher fraction of them fail. The grade is a description of conviction, not a prediction, and every setup we trade is sized to the same fixed risk.
When do you exit a trade like this GBPUSD long?
We exit at invalidation, the price level where the reason for the trade stops being true. Here that was the stop at 1.34145, the point below the VWAP and support where the long thesis no longer held. We do not move the stop or invent a second reason to stay in. When price reclaimed below the level at 19:05 UTC, the exit was automatic.
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: +22.02R YTD across 119 trades, see stats strip.
From the SkyAnalyst Journal
A Selective Week, A Clean Close: How the Desk Banked +2.64R
We traded six setups across June 15-21, won five, and sat out two instruments entirely. Pullback longs into support carried a risk-on tape, and the desk closed +2.64R net.
Case Study #93: A NAS100 Long That Rewarded Patient Continuation
A pullback continuation on NAS100, taken only after the read firmed across three evaluations. The market traveled to TP2 (+1.43R full-potential); the ledger logged +0.63R at TP1.
NAS100 Case #92: A Rate-Driven Trend Pullback That Paid +0.89R
A patient 25-hour NAS100 long, entered on falling 10Y yields and a clean pullback into support. We tracked the rate backdrop, bought the dip, and let the move reach TP2.