Every One of This Week's Six Losses Cost Exactly 1R.
Six stop-outs, a four-loss streak on Monday, and a 9.9% drawdown on the simulated account. Not one of them cost more than the 1R it was sized for, and that is t
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.
There are two ways to read a week with six losses in it, and only one of them is useful. The first is to ask what went wrong with each entry. We did that, and the answer is unsatisfying: not much. Every one of the six cleared its confirmation trigger, sat inside its entry zone, and stopped at a level that was set before the position existed. Each of them cost exactly 1R. The gross damage was 6.0R, which is 12,000 dollars on a simulated 100,000 dollar account at 2% risk, and the week netted to minus 4.95R after its single winner. The second reading is about how six losses arrived rather than why each one did. Four of them opened on Monday inside sixty-six minutes. That is the longest losing streak of the year and it is also, on inspection, close to a single position taken four times. Against the year to date ledger of +28.75R across 220 trades, which puts that same simulated account at $157,519.50, a 9.9% drawdown week is a thing that happens. Four correlated entries in one hour is a thing we chose.
Monday: four stop-outs in two hours and twenty-five minutes
The first entry filled at 14:17 UTC and the last of the four stopped at 16:42. EURUSD short, NAS100 short, US500 short, GBPUSD short, in that order, each one clearing its own confluence gate on its own instrument.
Every one of them was short risk. The two index positions needed equities to fall and the two currency positions needed the dollar to rise, and in a risk-on session those are the same bet with different tickers on it. Equities rose and the dollar fell, so all four failed together.
Four consecutive losses is the longest streak the system has produced this year. It is worth saying plainly that the streak is not four independent pieces of evidence about the system. It is closer to one.
What did not happen
No loss exceeded 1R. No stop was moved. No position was averaged into. The worst single trade of the week cost exactly what the worst single trade of a good week costs.
This is the boring half of the report and it is the half that matters most over a year. A system that loses six times and gives back 6.0R is behaving. A system that loses six times and gives back 9R has a sizing problem that the win rate will eventually be asked to cover, and it will not be able to.
The drawdown curve makes the same point visually. Four even steps down on Monday, one step back up on Thursday, two even steps down on Friday. Peak to trough 9.9% on the simulated account.
Friday: the dollar trade again
Cable short at 14:26, euro short at 14:47. The first had already stopped by the time the second was twenty-four minutes old, so the two overlapped rather than followed each other, and both of them were long dollar.
Two of the three teardowns below are those Friday trades and the third is Monday's Cable short. That is not a coincidence of selection. Four of the week's six losses were currency shorts and every one of them was the same directional bet on the dollar.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Sep 14 | 14:17 UTC | EURUSD | Short | EURUSD Sell-the-Pullback Short | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - | |
| Sep 14 | 14:42 UTC | NAS100 | Short | NAS100 SHORT, VWAP Rejection / Fibonacci Resistance Fade | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - | |
| Sep 14 | 15:16 UTC | US500 | Short | US500 SHORT, VWAP Rejection / Lower-High Sell | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - | |
| Sep 14 | 15:23 UTC | GBPUSD | Short | GBPUSD SHORT | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - | |
| Sep 18 | 14:26 UTC | GBPUSD | Short | GBPUSD Short Retracement Fade | C+ | -1.0R(SL) | -$2,000(SL) | Stop hit | - | |
| Sep 18 | 14:47 UTC | EURUSD | Short | EURUSD Short Bearish Continuation | 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 is correlation, and the useful version of that observation is more specific than the word.
Four of the six losses were shorts in EURUSD and GBPUSD, two on Monday and two on Friday. Both pairs are quoted against the dollar, so a short in either is a long dollar position. The system does not see it that way. It evaluates EURUSD and GBPUSD as two instruments, scores each against its own charts, and sizes each at 1R because each cleared its own gate.
On Monday the same thing happened across asset classes rather than within one. Short NAS100 and short US500 are one bet on US equities falling. Short EURUSD and short GBPUSD are one bet on the dollar rising. In a risk-on session those two bets are themselves correlated, because risk-on lifts equities and sells the dollar at the same time. Four positions, two apparent bets, one actual exposure.
The distribution of losses across the instrument grid says the same thing without any interpretation. EURUSD gave back 2.0R across two trades, GBPUSD 2.0R across two, NAS100 and US500 1.0R each on one apiece. The concentration is in the currency books because that is where the same trade was taken twice.
Decision highlights
Sizing Monday's four entries independently is the decision the week turns on. Each of the four cleared its own gate against its own charts and each was risked at 1R, which is correct behavior for four unrelated trades and wrong behavior for four expressions of one exposure. The Risk Agent sizes per position and enforces portfolio limits on gross exposure, but it has no concept of directional correlation between open positions, so four short-risk entries read to it as diversification.
Not intervening once the streak was running is the decision we would repeat. After the third Monday stop-out there was an obvious temptation to pull the fourth entry, which was already working through its gate at the time. Cutting a qualified setup because the previous three lost is how a system starts trading its own equity curve instead of the tape, and over a year that costs more than the 1R it saves here.
Publishing this week at all is a decision worth naming, because the quality gate on this report tried to stop it. Every trade in the window graded C+, below the B threshold the selector normally requires before it will build teardowns, on the reasoning that low-grade losses teach less. We overrode it. A week where the gate finds nothing instructive in six consecutive losses is exactly the week a reader deserves to see.
Six instruments, six stories.
Two losses, Monday and Friday, both shorts, 2.0R given back. Tied with Cable as the largest single contributor to the week's drawdown.
All EURUSD this week →Two losses, Monday and Friday, both shorts, 2.0R given back. The Friday entry stopped twenty-four minutes after filling, the fastest of the six.
All GBPUSD this week →No losses this week because there were no trades. The Dow stayed outside our setup criteria for all five sessions.
All US30 this week →One loss on Monday for 1.0R. The same book produced the week's only winner on Thursday, which is why it finished the window nearly flat rather than negative.
All NAS100 this week →No losses this week because there were no trades. Nothing in the yen cleared the confluence gate.
All USDJPY this week →One loss on Monday for 1.0R, stopped twenty-two minutes after filling. The single fastest failure of the week.
All US500 this week →No losses this week because there were no trades. USDCAD is the newest book and did not produce a qualifying setup.
All USDCAD this week →Loss of the week: EURUSD Short · -1R
Losses worth learning from
EURUSD Short, Friday, -1R, grade C+
The stop on this trade was seven and a half pips from entry. That is the detail worth pulling out.
Entry at 1.1464, stop at 1.14713, first target at 1.14514. The reward side was fine, 12.6 pips to the first target against 7.3 of risk for roughly 1.7R, and the structure behind it was real. What a band that narrow does not survive is ordinary noise. EURUSD routinely traverses seven pips on no information at all, so a stop placed that close is being asked to distinguish between the thesis failing and the pair breathing.
It lasted an hour and forty-one minutes, which is long enough to say it was not run over immediately. Price went the other way and kept going.
We will not claim the stop was wrong. It sat beyond the structure the setup was built on, and widening it after the fact is hindsight. What we will say is that a 7.3 pip band on a major is the tightest risk the system has taken in weeks, and it deserved a smaller position than a standard one.
GBPUSD Short, Friday, -1R, grade C+
This is the one with a genuine problem in it, and it is not the entry.
Entry at 1.33483, stop at 1.3363, first target at 1.33352. Run those numbers: 14.7 pips of risk against 13.1 pips to the first target. The first target sat at 0.89R, which means that on the conservative baseline our recaps use, a fully successful trade would have booked less than the loss it risked.
That is a setup that needed the second target to justify itself, and nothing in the entry logic requires it to get there. Five evaluations ran before the trigger printed. All five were disciplined about the entry and none of them asked whether the target ladder made the trade worth taking.
It stopped twenty-four minutes later, so the question never came due. It will come due eventually, and the fix belongs in the gate rather than in the teardown: a setup whose first target sits below 1R should have to clear a higher bar, not the same one.
GBPUSD Short, Monday, -1R, grade C+
The same book, the same direction, four days earlier, and this one was properly constructed.
Entry at 1.34806, stop at 1.34955, first target at 1.34655. Risk of 14.9 pips against 15.1 to the first target, so almost exactly 1R at TP1 with a second target beyond it. That is the ladder Friday's trade should have had.
The patience was there too. Eight consecutive evaluations declined to enter, each one recording that price was inside the sell zone but that no five minute candle had closed back below it with the rejection the setup required. On the eighth the confirmation printed and the position went on. An hour later it was stopped.
There is nothing here to fix. The trade was built correctly, waited correctly, triggered correctly and lost, in a session where every short lost. Filing it under process error would be a lie, and the honest filing is that this is what the losing tail of a working distribution looks like from the inside.
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 | -6R | −$12,000 |
All six agents combined.
From the desk
From the desk
Six losses, 6.0R gross, 12,000 dollars on the simulated account, and a 9.9% peak to trough drawdown. After the week's single winner the net is minus 4.95R and 9,900 dollars. We publish the losses article in weeks like this one on the same schedule we publish it in quiet weeks, because a drawdown report that only appears after good weeks is marketing.
The year to date ledger stands at +28.75R across 220 trades at a 56.82% win rate. On a static 100,000 dollar account at 2% risk that is $157,519.50. Compounding the same sequence, where each position is sized against the balance it actually has, gives $169,638.84. The roughly 12,000 dollar gap between those figures is not extra edge, it is the same R banked in a different order, and it survives a week like this precisely because no single position was allowed to be larger than the rest. A 9.9% drawdown on fixed fractional risk is an inconvenience. The same six losses at discretionary sizing, where the fourth trade of a losing Monday is often the biggest one, is how accounts end.
One note on scope. The year to date figure above is sealed at the August close by design and does not yet include September. September is open and currently sits at +3.92R, down from +8.87R on the tenth, and this week accounts for most of that round trip.
What we're tuning
We are not touching the entry gate, the stop placement rules, or the confirmation triggers. Six losses where every entry satisfied every rule is not evidence that the rules are loose, and retuning a system that is +28.75R on the year against the week it just had is the most reliable way to destroy it.
Two things this week did raise. The first is directional correlation between open positions: the Risk Agent limits gross exposure but has no concept of four positions all being long dollar or all being short risk, and we are looking at whether the fourth correlated entry in a session should have to clear a higher bar than the first. The second is narrower and came out of Friday's Cable trade, whose first target sat at 0.89R, below the risk it took. A setup that cannot pay 1R at its first target is a different animal from one that can, and the gate currently treats them identically. Neither change ships on the strength of one week.
What the numbers actually mean
A four-loss streak feels like evidence. It is worth doing the arithmetic on how much evidence it actually is.
Across 220 trades since inception the system has won 125 and lost 95, a 56.82% win rate, for a net of +28.75R. That works out to roughly 0.13R per trade taken, which is the number that matters and the one nobody feels. What a reader feels is the sequence, and the sequence contains streaks by construction. With a 43.18% chance of any given trade losing, the chance of four consecutive losses starting at any particular trade is about 3.5%. Over 220 trades you would expect somewhere around four separate four-loss runs. We have now had one. The streak is not early evidence of decay, it is slightly overdue.
This is the framework Van Tharp built the R-multiple around, and the point of expressing every result in units of risk rather than dollars is precisely that it makes the distribution legible. A system is not a win rate. It is a distribution of R-multiples with an expectancy attached, and the same expectancy can be delivered by a 40% win rate with large winners or a 57% win rate with modest ones. Jack Schwager's interviews across two decades of Market Wizards keep returning to the same observation from the other direction: the traders who survive are not the ones who avoid losing runs, they are the ones whose losing runs are boring.
One caveat on the numbers in the panel above. The average winner figure for this window is drawn from a single winning trade, because there was only one, and a one-trade average is not a target. The 56.82% win rate and the 220-trade sample are the figures with any weight behind them.
At a Glance
Drawdown questions
Six losses in one week. Has the system stopped working?
Nothing in this week's data says so. Every entry cleared its confirmation trigger and every loss was exactly the 1R it was sized for. Across 220 trades the system runs a 56.82% win rate, which means roughly 43% of trades lose, which means four-loss streaks are expected several times a year rather than never. One has now happened.
Why publish the losses at all?
Because a track record that only shows the winners is not a track record. We publish this report on a fixed schedule whenever a week contains at least one loss, so the reader sees the same distribution we do. The quality gate on this particular report actually tried to skip it, on the grounds that every trade graded C+ and low-grade losses teach less. We overrode that.
If four trades were really one bet, why did the system take four positions?
Because each instrument is evaluated independently and each of the four cleared its own gate. The Risk Agent enforces limits on gross exposure but has no model of directional correlation between open positions, so four short-risk entries look to it like four separate trades rather than one exposure taken four times. That is the honest answer and it is also the criticism.
Was the four-loss streak the worst drawdown of the year?
No. The 9.9% peak to trough on the simulated account is the deepest single-week drawdown so far and the four-loss run is the longest losing streak so far, but the year as a whole remains at +28.75R and the account has never been below its starting balance.
Why does this article say minus 6R when the weekly recap for the same week says minus 4.95R?
They measure different things. This report covers the losing trades only, so it totals 6.0R across six losses. The recap covers every trade in the window, so it nets those six losses against the one winner and arrives at minus 4.95R. Both numbers appear in this article for that reason, and neither is the other one adjusted.
Trade with the system that publishes its drawdowns.
Subscribers receive every signal, winners and losers alike, 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: +28.75R YTD across 220 trades, see stats strip.
From the SkyAnalyst Journal
The Regime Flipped on Monday. We Kept Selling Until Friday.
Seven trades, one winner, minus 4.95R. Every loss was a short, the average short died sixty-seven minutes after it was filled, and the only trade that went the other way was the only trade that paid.
The Last Six Shorts All Lost. This One Was a Long.
SkyAnalyst spent last week selling ten times and winning eight. Then the tape turned, six straight shorts stopped out, and the first trade that worked again was a NAS100 long for +1.85R.
Eight of Ten Winners, +5.43R, and an 80 Percent Strike Rate
Ten trades, eight winners, +5.43R and an 80 percent strike rate, the best week the desk has had this month. The setup grades underneath it are a reason to enjoy it without extrapolating from it.