Weekly Losses, One US30 Long Stops Monday Inside a 3-Win Week
Four trades on the week, three wins and one stop-out. The single loss came on Monday's opening US30 long. YTD ledger holds at +13.04R, compound balance $1,527 a
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 system entered four trades between April 27 and May 3, 2026: three winners and one stop-out. The stop came first, on Monday morning, when a US30 long at 14:34 UTC hit its 1R invalidation within 90 minutes. The other three entries (one each on Wednesday, Thursday, Friday) ran cleanly to their first targets. The week closed with a positive R total even after the Monday give-back. Through May 4, the running ledger holds at +13.04R YTD across 75 trades (44 wins, 31 losses, 58.67 percent win rate). A $100,000 simulated account at 2 percent risk per trade carries $126,071 static or $127,598 compounded. The week's give-back from the single Monday loss is approximately $2,000 of that figure. The compounded balance now sits $1,527 ahead of the static path, the largest compound gap of the year so far and the first time the divergence is meaningful at the per-trade-decision level. The Monday stop is the more interesting trade of the week. A loss landing first in the week tests the system in a way that a loss landing fifth does not. The Trend Agent did not adjust its threshold after the Monday stop, did not enter a revenge trade on Tuesday, and did not deviate from the gate on the three subsequent winners. The discipline of NOT trading on Tuesday is the kind of decision Weekly Losses articles exist to make visible.
Act 1: Monday opens the week with the only stop-out
The week opened with a US30 long at 14:34 UTC on Apr 27. The setup cleared the gate at C+ on a structural support cluster with the Macro Agent's regime read leaning bullish for indices. The trade hit the 1R stop within 90 minutes. The Trend Agent's read had been honest at the entry, but the chart did not hold the structural level on the first test. Realized R was exactly -1.00.
Act 2: Tuesday produced no entries
Tuesday is the more important sentence of this article. After a Monday stop-out, the temptation is to chase the loss with an aggressive Tuesday entry. The Trend Agent ran its scheduled evaluations across all six instruments on Tuesday and cleared none. The gate did not relax. No chase, no revenge trade, no compensatory positioning. The position book sat flat for the full session.
Act 3: Wednesday, Thursday, and Friday delivered three clean winners
The remaining three trading days produced one trade each, all of which ran to TP1 cleanly. An index long Wednesday, a forex carry trade Thursday, and a structural-failure short Friday. The combined R from the three winners more than offset the Monday stop, leaving the week net positive even after the early give-back. The cadence of one quality entry per session is roughly the system's long-run pace.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Apr 27 | 14:34 UTC | US30 | Long | Claude Opus 4.7 | US30 Pullback Long to VWAP/Fib Confluence | 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: the gate does not adjust after a stop
A loss landing early in the week is psychologically more taxing than a loss landing late. The temptation when the book opens red is to find a setup that will close the deficit. Discretionary traders fall into this routinely; the academic literature on behavioral trading (Kahneman, Thaler) documents the loss-aversion mechanism that drives it.
The system has no loss-aversion mechanism. The Trend Agent's gate is calibrated on long-run expectancy across the full setup catalogue. A stop-out in the morning does not change the gate's threshold in the afternoon. The Macro Agent's regime read does not get more permissive after a loss. The Risk Agent's position scalar does not increase to compensate.
What happens after a stop-out is exactly what was going to happen if the stop had not been hit: scheduled evaluations on the next session, gate clearance only when confluence supports it, position size determined by the regime tag at the time of the entry. The Monday stop did not cause Tuesday's no-trade outcome. Tuesday's no-trade outcome was caused by the chart on Tuesday not producing a setup. The two days are independent.
Decision highlights
The Trend Agent's decision to take the Monday US30 long cleared the gate at C+ on a structural support cluster. The macro alignment was supportive, the higher-timeframe structure was bullish, and the 5m entry trigger fired at a value reference. The grade was honest at C+; the outcome was the grade's expectancy resolving on a single sample.
The Trend Agent's decision NOT to enter a Tuesday trade is the more important second decision. After a Monday stop, the gate did not adjust. Scheduled evaluations on Tuesday returned no above-gate clearances and the position book sat flat. The discipline of refusing setups when the chart does not produce them, regardless of recent P&L, is the system's structural anti-revenge-trading mechanism.
The Risk Agent's stop placement on the US30 long was tight to the structural support cluster. When the stop hit, the realized R was exactly -1.00 with no overshoot. The exit was clean. Tight structural stops on borderline-grade setups produce fast clean losses when the read is wrong, which is what we want over a chasing exit that lets the loss expand.
Claude vs GPT: who led the week.
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.
- US30-1.0R · 1 trade
Same signals, same risk framework, different foundation model.
Six instruments, six stories.
EURUSD: no trades this week. The dollar bloc lean produced no setup the Trend Agent could grade above the gate floor on any of the five sessions.
All EURUSD this week →GBPUSD: no trades this week. Cable's volatility profile kept the 5m structure unsettled enough that no setup cleared the gate.
All GBPUSD this week →US30: one long, one stop-out (-1.00R). The Monday opening setup cleared at C+ and stopped within 90 minutes when the structural support did not hold.
All US30 this week →NAS100: no trades this week. The index chart did not produce a setup above the gate floor on any session.
All NAS100 this week →USDJPY: one long, one winner. The Thursday carry trade ran to TP1 cleanly on yield support. Setup grade B.
All USDJPY this week →US500: two trades, two winners. The Wednesday and Friday entries ran to TP1 in the same sessions. Setup grades B and B+.
All US500 this week →Loss of the week: US30 Long · -1R
Losses worth learning from
US30 Long, Apr 27, -1.00R (C+ grade)
What was right: the structural support cluster was a setup the system has traded successfully many times. The Trend Agent's confidence at C+ reflected honest grading of a setup that cleared the gate but did not exceed it. The Risk Agent's stop placement was tight to the structural invalidation. The Macro Agent's regime read was supportive for the long direction.
What was wrong: the structural support did not hold on the first test. The chart produced a clean rejection at the support level (the entry trigger), then failed the level on a subsequent retest. The trade's invalidation was the level failing the second test rather than the third. A B-grade version of this setup would have required a more decisive holding pattern on the first test before entry.
What we would do the same: take it again. The setup was inside the catalogue, the grade was honest, and the position size was correct for the C+ classification. A single stop-out at C+ is the cost of including borderline grades. The discipline lies in the stop holding when the read is wrong, and the next-day discipline of not chasing the loss with a revenge entry. Both held this week.
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
From the desk
A 3-to-1 win-loss week with the loss landing first is what the system is designed to handle. The discipline lives in the Tuesday non-entry: a discretionary trader chases the morning's stop with afternoon urgency; a systematic gate does not.
A $100,000 simulated account at 2 percent risk sits at $126,071 (static) or $127,598 (compounded) through May 4. The week's give-back from the single US30 stop is approximately $2,000 of that figure. The compounded balance now leads static by $1,527, which is the largest compound payoff we have published this year and the first visible payoff of the running-balance risk discipline at a real sample size.
The 58.67 percent win rate at 75 trades sits inside the long-run target band and is no longer dominated by sample noise. We are running the calibration we built. The Weekly Losses article exists to put the one loss on the page at the same prominence as the three wins. Selective reporting is what funds blow up; comprehensive reporting is what compounds.
What we're tuning
The post-trade review for this week produced no methodology changes. The C+ Monday loss is statistically routine. The discipline of refusing Tuesday entries on a single-stop start is the existing system behavior, not a new tuning. The compounded-versus-static gap at 75 trades is the expected payoff of the position-sizing rule, not a result of changing it.
One observation worth logging: the week's three winners on Wednesday, Thursday, and Friday all came at B-or-better grades. The system's recent tendency on multi-trade weeks has been to cluster B-grade entries together rather than spread C+ entries across more sessions. We are tracking whether this is a structural feature (the macro regime producing better setups in clusters) or a sample artifact. No tuning action yet.
What the numbers actually mean
A 58.67 percent win rate across 75 trades is the cleanest signal we have published yet. Position-trading research consistently treats samples above 50 trades as the threshold where the underlying win rate becomes a real estimate rather than sample noise. Van Tharp's <em>Trade Your Way to Financial Freedom</em> explicitly walks through the convergence math: a system with a 58 percent underlying win rate will produce 75-trade samples within plus or minus 5 to 7 percentage points of the underlying mean roughly 90 percent of the time. Our observed 58.67 sits comfortably inside that band.
What is interesting at this sample size is the static-versus-compounded gap. The compounded balance of $127,598 is $1,527 ahead of the static $126,071. That is a 1.21 percent dollar lead from compounding, after the system has cleared 75 trades. The math works like this: each trade's dollar risk under compounding is 2 percent of the running balance, not 2 percent of the original $100,000. After 75 trades, the running balance has grown enough that each subsequent winner contributes a slightly larger dollar return than the fixed-risk path would have produced.
Across systematic-trading research, the consistent finding is that compounded dollar returns diverge from static returns slowly at first and meaningfully later. Jack Schwager's <em>Market Wizards</em> interviews repeatedly cite traders who report compound-vs-static gaps becoming visible at the 50-trade mark and obvious at the 100-trade mark. Our gap at 75 trades is exactly in the middle of that progression. The discipline of maintaining a 2-percent-of-running-balance risk rule across an entire trading year is what produces the gap; the gap is the visible payoff of that discipline.
At a Glance
Drawdown questions
How does the system prevent revenge trading after a stop-out?
Structurally rather than psychologically. The Trend Agent's gate threshold does not adjust based on recent P&L. The Macro Agent's regime read does not become more permissive after a loss. The Risk Agent's position scalar does not increase to compensate. After a stop-out, the system runs scheduled evaluations exactly as it would have if the stop had not been hit. Refusing the next setup is the default, not the exception.
Why does the compounded balance pull ahead of the static balance over time?
Because each trade's dollar risk under compounding is 2 percent of the running balance, not 2 percent of the original starting balance. After winning trades grow the balance, each subsequent winner risks slightly more dollars and returns slightly more on a per-R basis. The gap widens as the sample grows. At 75 trades, the gap is $1,527; at 200 trades, the gap will be meaningfully larger.
What does a C+ grade mean for setup expectancy?
C+ is the lowest grade the gate accepts. Setup grade is a deterministic output of the Trend Agent's confluence scoring: A and B grades produce higher per-trade expectancy than C+ grades. The gate accepts C+ because their long-run expectancy is still positive over a large sample, even though the per-trade variance is higher. A single C+ stop-out is statistically routine; the gate accepts the higher variance as the cost of including borderline grades in the catalogue.
When does a single loss on Monday matter more than a single loss on Friday?
Statistically, never. A -1R stop-out is a -1R stop-out regardless of day-of-week. Psychologically, a Monday loss tests the operator's discipline more than a Friday loss because the operator has the full week ahead to "make it back". The system does not have that bias because it is not the operator. The day-of-week is independent of the gate's calibration; the operator's experience of the week is not.
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: +13.04R YTD across 75 trades, see stats strip.
From the SkyAnalyst Journal
US500 Long Waits for the Conditional Reclaim Before Buying the Dip
GPT-5.5 refused four times before entering US500 long at 7487.2. The Trend Agent required a reclaim of the opening-range breakdown zone, not the VWAP touch. TP1 booked +1.15R.
Eleven stops, one tape: the week structure refused to confirm
Eleven losses, nine R given back, a peak-to-trough drawdown of 10.81 percent and a longest losing streak of four. The honest portfolio view: what each stop taught us, and what the curve says about a week the structure refused to confirm.
May 18-24, 2026: A Seven-Eleven Week That Closed -2.82R Across Claude and GPT
Eighteen trades, seven winners, eleven losers, -2.82R net at TP1 baseline. Claude opened Monday with two early wins, GPT carried the index side mid-week, and a Friday cluster netted both sides back toward flat without crossing it.