The five small losses hiding inside a winning week
The week made plus 3.02R, and it still produced five losing trades. We publish them the same way we publish the losses of a red week, because a track record tha
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.
This is a losses report on a week that made money. The system closed the week up plus 3.02R across twelve trades, and it still lost five of them, so here they are. We write this the same way we write the losses report on a red week, because the only version of transparency that means anything is the one that does not wait for a bad week to show you the bad trades. The proportion is the whole point, so we will state it plainly. Through Jul 27, 2026 the system has banked plus 29.27R since the January 12 inception, and on a 100,000 dollar account at two percent risk that static balance sits at 158,553 dollars. This week's five losses gave back 5R gross, about 10,000 dollars of simulated equity at the moment they were taken, though the seven winners more than covered it. A 5R give-back inside a plus 29.27R record is not a drawdown that means anything on its own. It is the ordinary friction of a system that loses roughly four trades in ten by design. What follows is the anatomy of five losing trades that happened to land in a winning week, and the statistics that explain why five losses out of twelve is exactly what a 59 percent win rate produces.
Act I — the lone early loss
The week's losses did not arrive in a heap. The first was solitary: a Monday US500 short that stopped at 1R while the trades around it were winning. It is the kind of loss that barely registers in the moment, one red trade in a green sequence, and it is worth naming precisely because it is so ordinary. A single stop inside a run of winners is not a warning. It is the baseline noise of any system that does not win every trade.
Act II — the long that leaned the wrong way
The middle loss was the only long among the five. On Wednesday the system took a US30 long, reasoning much as it did on the Nasdaq long that day, that intraday strength could outrun the macro headwind. On the Nasdaq that read paid; on the Dow it did not, and the long stopped at 1R. Same logic, same session, different instrument, opposite result. That is not an inconsistency; it is what happens when you take probabilistic bets across correlated but distinct markets.
Act III — the Friday give-back
The only real cluster came at the end. Two Friday shorts, on US30 and NAS100, stopped back to back for a combined 2R, pulling the account from its Thursday high near 110,055 dollars down to about 106,055. That was the deepest the week's drawdown got, 3.63 percent measured from the peak, and it was entirely orderly: two flat, identical losses taken at their planned risk with no attempt to widen a stop or add size to defend the high.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Jul 20 | 15:04 UTC | US500 | Short | Claude Opus 4.7 | US500 SHORT — VWAP Breakdown Continuation | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 22 | 14:17 UTC | US30 | Long | GPT-5.5 | US30 Long Pullback Retest | C | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 22 | 15:26 UTC | GBPUSD | Short | GPT-5.5 | GBPUSD SHORT | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 24 | 14:08 UTC | US30 | Short | GPT-5.5 | US30 SHORT | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - |
| Jul 24 | 15:19 UTC | NAS100 | Short | Claude Opus 4.7 | NAS100 SHORT — Aggressive Resistance Cluster Rejection | 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 in the losses was that they were marginal, not sloppy. None of the five carried a setup grade of B or better; they were the C-grade and C-plus entries that cleared the filter but never had the conviction behind them that the week's winners did. That is the grading system working: the trades that failed this week were disproportionately the ones the grade had already flagged as the weakest, and the trades that graded highest, like the B-plus US30 fade and the B Nasdaq short, were the ones that paid.
The one shape worth naming is timing. The two Friday losses were both late-session continuation shorts, taken after the bulk of the week's directional move had already happened. A continuation entry into an already-extended move is inherently lower quality than a fresh one, which is exactly why both graded C-plus, and it is the closest thing to a common thread the five losses share.
Decision highlights
The Friday give-back was allowed to happen at full discipline. After the account reached its plus 5R high on Thursday, two more setups cleared the filter on Friday and both stopped at a flat 1R. The system did not widen stops or size up to protect the weekly high, which is the single most common way a good week becomes a mediocre one. It took the two losses exactly as planned.
The US30 long was the only loss that leaned against the week's grain. It was taken on the same intraday-strength logic as the Nasdaq long that paid the same day, and it stopped because the Dow lacked the sector-specific bid the Nasdaq had. The judgment worth noting is that the system did not let one instrument's winning read force the correlated trade; it sized and stopped the Dow long on its own merits, and its own merits ran out.
Every one of the five losses was a marginal-grade setup. None graded B or better, and that is not a coincidence: the trades that failed this week were the ones the grading system had already scored lowest. The decision that matters here is one the system makes before entry, when it assigns a grade that turns out, across a large sample, to correlate with whether the trade works.
Six instruments, six stories.
EURUSD: No losses this window. Both EURUSD trades were winning shorts, outside the scope of this report.
All EURUSD this week →GBPUSD: One loss, minus 1R. A single short that stopped, against two winners in the same book on the week.
All GBPUSD this week →US30: Two losses, minus 2.0R combined, the only book to lose more than once. A Wednesday long and a Friday short both stopped.
All US30 this week →NAS100: One loss, minus 1R. A late Friday continuation short that stopped, against two Nasdaq winners earlier in the week.
All NAS100 this week →USDJPY: No losses this window. The pair stayed outside our setup criteria the entire week.
All USDJPY this week →US500: One loss, minus 1R. A solitary Monday short that stopped without following through.
All US500 this week →Loss of the week: NAS100 Short · -1R
Losses worth learning from
NAS100 short, Friday's late continuation
The loss of the week, and instructive precisely because it looks so similar to a winner. A day earlier the system had shorted the Nasdaq into a fresh, high-conviction breakdown and it ran to its third target. This Friday short was the same instrument and the same direction, but it arrived after the big move had largely played out, which is why it graded C-plus rather than B. What was right was the process: a defined setup, a flat 1R risk, a stop that did its job. What was wrong was timing, taking a continuation entry into a move that had already spent most of its energy. What we would do the same is take it at reduced conviction and let the stop settle the question, which is exactly what happened. A late continuation that stops at 1R is not an error; it is the lower-probability tail of a pattern whose fresh version had paid handsomely the day before.
US30 short, the second Friday stop
The other half of the Friday give-back, and a near-twin of the first. A US30 continuation short taken late in the session after the week's index weakness had largely resolved, graded C-plus for the same reason: the easy part of the move was gone. What was right was the risk discipline, a flat 1R sized identically to every other trade regardless of how the week had gone. What was wrong was chasing a continuation that the tape had already mostly delivered. What we would do differently is not the entry itself but the weighting we give late-session continuation setups, which we address below. Paired with the Nasdaq short, this loss is the clearest argument of the week that a continuation entry is only as good as how much move is left when it triggers.
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 | -5R | −$10,000 |
All six agents combined.
From the desk
A note, before we move on.
We publish a losses report on a winning week for the same reason we publish one on a losing week: because the alternative is to let you see the losses only when we are forced to. A track record that reports its red trades only in its red weeks is curating, and curation is the one thing a track record cannot survive. So here are five losing trades from a week that made plus 3.02R, in the same detail and the same voice we would give them if the week had gone the other way.
Put the give-back in dollars and it stays in proportion. On a 100,000 dollar account at two percent risk, the static balance sits at 158,553 dollars, and this week's five stops removed roughly 10,000 dollars of that at the moment they were taken, a 3.63 percent dip from the Thursday peak that the week's winners more than reversed by the close. The compounded version of the same record, where each trade risks two percent of the growing balance rather than a fixed amount, sits higher at 173,098 dollars. The gap between those two figures is what disciplined sizing builds over 163 trades, and a five-loss week narrows it by a few days at most.
The reason we can write about five losses this calmly is that the risk policy made the worst case knowable before any of them happened. Five stops were always going to cost 5R, no more, and the day they landed we did nothing differently than the day the winners did. That steadiness is the product, and it is just as visible in a green week's losses as in a red one's.
What we're tuning
There is nothing structural to fix in a winning week's losses, and we are wary of inventing lessons just because a report exists to be written. Each of the five losses was a graded setup stopped at its planned 1R, and the flat risk policy did exactly what it is supposed to do on both the winning and losing trades.
The one genuine observation is the one the two Friday losses keep pointing at: late-session continuation entries. Both were continuation shorts taken after the bulk of the week's directional move had already happened, and both graded C-plus for that reason. We do not currently apply an explicit penalty to a continuation setup that arrives late in an already-extended move, and the pairing of Friday's tired shorts against Thursday's fresh, profitable one is a clean argument that we could. It is a refinement to weigh, not a flaw to fix, and it would have changed a green week only at the margins.
What the numbers actually mean
The first thing to understand about five losses in a twelve-trade week is that it is not just normal, it is the expectation. The system's win rate sits at 58.9 percent, which means it loses roughly 41 percent of the time, and 41 percent of twelve is about five. This week did not underperform its loss rate or overperform it; it landed almost exactly on it. Jack Schwager's Market Wizards interviews return to this point again and again: even the traders with the best long-run records lose a large fraction of their individual trades, and the ones who survive are the ones who made peace with that arithmetic rather than fighting it.
The second thing is why we report every loss in R. Van Tharp's framework in Trade Your Way to Financial Freedom reframes each trade as a multiple of the amount risked, which is what lets five scattered stops become a single knowable number: minus 5R, no more, because the risk per trade is fixed in advance. The system's average reward target sits near 1.15R and its win rate near 59 percent, which is a positive expectancy, and a positive-expectancy system still produces five-loss weeks constantly. Expectancy is a statement about the average of many trades, never a promise about any twelve of them.
The third thing is proportion, and it cuts against reading too much into this report. The longest losing streak of the week was two, which is about as benign as a losing streak gets. The peak-to-trough drawdown was 3.63 percent on a 2 percent risk model, which is well inside the range this strategy generates in ordinary operation. Set against a 163-trade record and a plus 29.27R year, a 5R give-back that the same week's winners more than covered is not a signal about the edge. It is the texture of running one.
At a Glance
Drawdown questions
Why publish a losses report on a week that made money?
Because transparency only counts if it does not wait for a bad week. A system that shows you its losing trades only when the weekly total is negative is curating its record. We publish the losses of a green week in the same detail as a red one, here five stops inside a plus 3.02R week, so the track record reports the same way regardless of the outcome. That consistency is what makes it a record rather than a highlight reel.
Is five losses in twelve trades a bad week?
No, it is almost exactly the expected rate. The system wins about 59 percent of its trades, so it loses about 41 percent, and 41 percent of twelve is roughly five. This week landed right on its long-run loss rate. Combined with a positive expectancy, that loss rate still produces a profitable week most of the time, which is what happened here: seven winners outweighed the five losses for a net plus 3.02R.
What was the drawdown, and does it matter?
The peak-to-trough drawdown was 3.63 percent, measured on a 2 percent risk model from the Thursday high after two Friday stops. It is a small, ordinary dip well inside the range this strategy generates in normal operation, and the longest losing streak behind it was just two trades. Against a 163-trade record and a plus 29.27R year, a give-back that size is texture, not a signal about the edge.
Why were all five losses low-grade setups?
Because the trades that failed this week were disproportionately the ones the grading system had already scored lowest. None of the five graded B or better; they were the C and C-plus entries that cleared the filter without the conviction the week's winners had. That is the grade doing its job: over a large sample, the highest-graded setups win more often than the marginal ones, and this week the marginal ones were where the losses landed.
Two of the losses were late Friday continuation shorts. What is the lesson?
That a continuation entry is only as good as how much of the move is left when it triggers. Both Friday shorts took the same continuation logic that had paid on a fresh, high-conviction Nasdaq short the day before, but they arrived after the week's directional move had largely played out, which is why both graded C-plus and both stopped. The refinement we are weighing is an explicit penalty for continuation setups that arrive late in an already-extended move.
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: +29.27R YTD across 163 trades, see stats strip.
From the SkyAnalyst Journal
The week we traded the Nasdaq both ways and won them both
Twelve trades, seven winners, a net plus 3.02R. The standout was the Nasdaq, which we shorted with the yields on Thursday two days after buying it against them, and both trades paid.
The Nasdaq short the macro demanded, days after the long
Two days ago the system bought the Nasdaq against rising yields. Today it sold the same index with them, as yields accelerated, crude spiked, and the VIX jumped. The relief rally failed and the short ran to TP3.
The Dow short the market breadth called before price did
A Dow short where the tell was not the chart but the market internals. Negative breadth kept the bias short, price failed at the open, and the fade ran to TP2 for a full-potential 1.5R.