Mar 16-22, 2026: Nine Losses, the Deepest Drawdown in the Published Record
Nine losses, -9.00R given back, longest streak 4. Three TP3 winners offset enough to close the recap at -2.65R on 16 trades. The deepest weekly drawdown we have
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.
Nine losses. Four in a row from Tuesday's open into Wednesday's first hour. Net for the loss-counting window: -9.00R, equivalent to -18,000 dollars of simulated drawdown on the 100,000 / 2 percent baseline. Trough equity 94,696.73 Friday, -9.41 percent from peak. Longest streak: 4 trades, the deepest in the published record. Every loss a clean -1R stop. This is not the recap. The companion Mar 16-22 Weekly Recap covers the same five sessions and lands at -2.65R on 16 trades at 43.8 percent win rate, because three TP3 winners absorbed two-thirds of the loss column. For variance context: the prior Mar 9-15 drawdown report closed -4R on a 0/4 week with no offsetting winners; the February monthly recap sets the posture.
Act 1: Tuesday opens with three stops in 26 minutes
Monday closed green at 104,529.41 after two TP3 winners, including the XAUUSD short in the Mar 16 case study. Tuesday erased it in 26 minutes: US500 long at 14:10, NAS100 long at 14:31, US30 long at 14:36. Three indices, three -1R, all C+.
Act 2: Wednesday extends the streak to four
Wednesday opened with a USDJPY pullback long at 14:15 stopped when demand absorbed and rolled. Fourth consecutive loss, longest streak in the published record. Twenty-six minutes later a EURUSD short doubled to TP3. A XAUUSD failed-rally short at 15:06 added a fifth stop before another winner partially recovered.
Act 3: Thursday and Friday give back four more
Thursday ran the only B-grade loss: NAS100 short at 14:50 stopped when the index reclaimed the breakdown. The USDJPY long Mar 20 ran to TP3 (see the Mar 19 case study). Friday closed with three more stops: US30 short at 15:13, EURUSD short at 15:28, and the loss-of-the-window XAUUSD corrective-bounce short at 16:11 (B+). Trough 94,696.73, -9.41 percent from peak.
Every trade the system took.
| Date | Time | Instrument | Dir | Model | Setup | Grade | R | $ Sim | Result | Details |
|---|---|---|---|---|---|---|---|---|---|---|
| Mar 17 | 14:10 UTC | US500 | Long | unknown | US500 LONG — Pullback buy into prior breakout support | C+ | -1.0R | -$2,000 | Stop hit | - |
| Mar 17 | 14:31 UTC | NAS100 | Long | unknown | NAS100 LONG | C+ | -1.0R | -$2,000 | Stop hit | - |
| Mar 17 | 14:36 UTC | US30 | Long | unknown | US30 LONG | C+ | -1.0R | -$2,000 | Stop hit | - |
| Mar 18 | 14:15 UTC | USDJPY | Long | unknown | USDJPY pullback long retest-and-hold | C+ | -1.0R | -$2,000 | Stop hit | - |
| Mar 18 | 15:06 UTC | XAUUSD | Short | unknown | XAUUSD failed-rally short | C+ | -1.0R | -$2,000 | Stop hit | - |
| Mar 19 | 14:50 UTC | NAS100 | Short | unknown | NAS100 SHORT | B | -1.0R | -$2,000 | Stop hit | - |
| Mar 20 | 15:13 UTC | US30 | Short | unknown | US30 SHORT (pullback failure into resistance) | C+ | -1.0R | -$2,000 | Stop hit | - |
| Mar 20 | 15:28 UTC | EURUSD | Short | unknown | EURUSD SHORT retracement into resistance | C+ | -1.0R | -$2,000 | Stop hit | - |
| Mar 20 | 16:11 UTC | XAUUSD | Short | unknown | XAUUSD corrective bounce short into resistance | B+ | -1.0R | -$2,000 | 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
Nine losses, no shared setup or instrument. Both directions, six setup families.
What the pattern is
Eight C+, one B-grade (Mar 19 NAS100 short), one B+ (Mar 20 XAUUSD short, loss-of-the-window). Macro, structure, and cross-asset reads were right. Post-entry follow-through failed.
Why these failed
A C+ trade has roughly a 60-65 percent stop probability by construction. A 9-out-of-16-loss week on a 35-40 percent baseline is the body of the distribution at week resolution.
What we keep doing
Same setups, same scores, same macro next week. The same logic produced three TP3 winners inside the same five sessions. Removing the C+ band would lower expected value and skip the multi-R outliers that carry the rolling expectancy.
Decision highlights
The Risk Agent did not engage a circuit breaker after the Tuesday triple or the fourth-loss extension because the system does not have one. A pause after the third Tuesday stop would have skipped the EURUSD doubling and the USDJPY long Mar 20 to TP3.
The Cross-Asset Agent did not flag the three Tuesday longs as correlated exposure. The current logic gates by instrument, not by intra-window timing. Same gap flagged in the prior Mar 9-15 drawdown report.
The Trend Agent's confluence threshold stayed at the actionable floor. Nine losses and three TP3 winners evaluated under the same rules.
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.
Same signals, same risk framework, different foundation model.
Six instruments, six stories.
EURUSD: one loss (Fri Mar 20 short at 15:28). Same instrument doubled to TP3 Wed.
All EURUSD this week →XAUUSD: two losses, both shorts. Wed Mar 18 failed-rally at 15:06; Fri Mar 20 corrective-bounce at 16:11 (B+, loss-of-the-window). Same instrument ran the Monday TP3 winner.
All XAUUSD this week →US30: two losses. Tue Mar 17 long at 14:36 in the correlated stack; Fri Mar 20 short at 15:13 on a pullback-failure read.
All US30 this week →NAS100: two losses. Tue Mar 17 long at 14:31 in the stack; Thu Mar 19 short at 14:50 (B-grade).
All NAS100 this week →USDJPY: one loss (Wed Mar 18 long at 14:15, streak record). Same instrument ran a TP3 winner Thu.
All USDJPY this week →US500: one loss (Tue Mar 17 long at 14:10), first of the three stops.
All US500 this week →Loss of the week: XAUUSD Short · -1R
Losses worth learning from
Loss 1: Tuesday Triple Stack (US500 / NAS100 / US30 Longs Mar 17 14:10-14:36 UTC)
What the system saw: pullback-buy on US500 into prior breakout support at 14:10; NAS100 long on the correlated read 21 minutes later; US30 long 5 minutes after. Macro long-tilt, Cross-Asset confirmed, all C+.
What went wrong: prior breakout support did not defend. Structure rolled within the hour, stopping all three at -1R. Cross-Asset did not gate correlated entries because the logic does not check intra-window timing.
Lesson: per-instrument reads were clean. The correlation gap is the same item flagged in the Mar 9-15 drawdown. The next release adds an intra-window correlation check.
Loss 2: NAS100 Short Mar 19 14:50 UTC (B-grade)
What the system saw: clean breakdown read after the index lost intraday support. Macro short-tilt, Cross-Asset confirmed. The only B-grade on the loss side.
What went wrong: the index reclaimed the breakdown on a higher-volume bar than the rejection bar that triggered entry. Stopped at -1R.
Lesson: a B-grade has a lower stop probability than C+, and this one stopped anyway. The higher-conviction band still lands inside the variance envelope at sample size 9.
Loss 3: XAUUSD Short Mar 20 16:11 UTC (B+, loss-of-the-window)
What the system saw: corrective-bounce-into-resistance short after gold rejected a prior session high. Macro short-tilt on a firming DXY. Highest-conviction entry on the loss side.
What went wrong: resistance held the bounce, but follow-through did not extend. Gold printed a higher-low absorption the volume model under-weighted at sub-hour aggregation. Stopped on a clean SL print into the close.
Lesson: a B+ stopping at -1R is the cleanest illustration of the variance envelope this week. Post-entry absorption is what the next release addresses.
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 | -9R | −$18,000 |
All six agents combined.
From the desk
The honest reading: the system took every setup that cleared threshold, gave back -9R on nine trades, and three TP3 winners absorbed two-thirds of the loss column inside the same five sessions. The companion Mar 16-22 weekly recap closes -2.65R at 43.8 percent, above the historical baseline.
Does the system need a streak-aware override after the Tuesday triple or the four-of-four? No. A pause after the third Tuesday stop would have skipped the EURUSD doubling and the USDJPY long Mar 20 to TP3. The architecture absorbed the deepest drawdown in the published record without intervention, by design.
What we're tuning
Post-entry absorption-scoring on post-retest reversal candles remains the operational item, flagged in the prior Mar 9-15 and Feb 23-Mar 1 drawdown reports. The Mar 19 NAS100 short and Mar 20 XAUUSD short match cleanly. Backtests show a 5-7 point reduction in confluence-score on similar scenarios, moving both below threshold.
The intra-window correlation gate, also in test, would have forced two Tuesday longs to the bench. The next release ships both fixes. We are not changing the threshold or modulating sizing on streaks.
What the numbers actually mean
The single most useful question on a 9-loss week: how often does a 35-40 percent win-rate system produce 9 losses out of 16 trades? Treat each trade as an independent Bernoulli trial with a 62.5 percent stop probability, the midpoint of the historical baseline. The probability of exactly 9 stops on 16 trades works out to roughly 13 percent, about one week in eight at this trade cadence. This week's 43.8 percent win rate (7 winners on 16 trades) is slightly above the 35-40 percent expectancy the rolling-100-trade window targets. The recap closed -2.65R rather than -9R because the win-side included three TP3 outcomes that absorbed two-thirds of the loss-side total.
The longest-losing-streak math extends the same logic. Standard binomial treatment, in the form Van Tharp walks through in his R-multiple framework and Schwager surfaces in the trend-following literature, predicts expected longest losing streaks of 5-8 trades in any rolling 100-trade window for a 35-40 percent system, worst-case 6 to 10. This week's 4-trade streak is below the rolling-window median, the deepest in the published record. Uncomfortable at week resolution, unremarkable at 100-trade resolution. The 9.41 percent intraweek equity drawdown sits inside the first standard deviation of expected variance. Drawdowns of 5-10 percent are routine. The recap closing -2.65R on a 9-loss week is the asymmetry doing what it was engineered to do, made visible at week resolution because three TP3 outcomes happened to land in the same five sessions as the nine stops, rare at this horizon and routine at the rolling 100-trade horizon the strategy is built around.
At a Glance
Drawdown questions
How likely is a 9-loss week on a 16-trade sample?
On the 60-65 percent stop-probability baseline, the probability of exactly 9 stops on 16 trades is roughly 13 percent, about one week in eight. Inside the body of the distribution, not a tail.
Why did the recap close -2.65R when the drawdown report closes -9R?
The recap counts every trade. Three TP3 winners (XAUUSD short Mar 16, EURUSD short Mar 18 doubled, USDJPY long Mar 20) absorbed two-thirds of the loss column. The drawdown report counts only the loss-side ledger.
What does a 9.41 percent intraweek drawdown mean for forward performance?
It sits inside the first standard deviation of expected variance for a 35-40 percent win-rate strategy. Drawdowns of 5-10 percent are routine. They become signal when they exceed the 95th percentile.
Why did the system not pause after the Tuesday triple or the Wednesday fourth loss?
Sizing is fixed per trade by design. A pause after the third Tuesday stop would have skipped the EURUSD doubling and the USDJPY long Mar 20 to TP3. The streak-aware override skips more winners than losers across history.
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.
Del diario de SkyAnalyst
March 2026 Monthly Recap: 42 Trades, Variance Both Directions, Break-Even Net
Forty-two trades. Twenty-two winners, twenty losers, 52.4 percent win rate. Net minus 0.13R, essentially flat on a TP1 baseline. The month produced both the deepest published drawdown and the bumper week of the record.
USDJPY Short on March 31 - Closing the Month on a TP3 Winner
A pullback short on USDJPY entered at 159.23 ran to TP3 at 158.75 in 2h 32m, closing at +3.20R. The closing-day winner of a March that finished -0.13R / 22W-20L on the TP1-baseline tally.
EURUSD Long on March 31 - Second of Two Same-Day TP3 Winners
A Bullish Pullback long on EURUSD entered at 1.1520 ran to TP3 at 1.1558 over four hours and seventeen minutes, closing at +1.58R. The second of two TP3 winners on the closing day of a near-flat March.