The Last Five Shorts All Lost. This One Was a Long.
SkyAnalyst AI journal entry: NAS100 Long on Sep 17, 2026 closed +1.85R on TP2. Full workspace view, decision log, and AI reasoning, unedited.
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. That’s what makes the system auditable, and it’s what this case study will show, step by step, on a specific setup the trend agent almost passed on.
Last week this desk sold the Nasdaq, the S&P, the Dow and Cable ten times and won eight of them. Every trade was a short, and we wrote at the time that the tape gave nothing else. On Monday it took four more shorts. All four stopped out. On Thursday it bought the Nasdaq at 29,395 against a stop at 29,295 and closed at 29,580 the following afternoon, a hundred and eighty-five points for +1.85R. That is the whole story of the last eight sessions in three paragraphs: a regime that paid, a regime that stopped paying, and one trade on the other side of it.
Every input reversed
The setup on September 17 is most usefully read against September 10, because the same instruments were saying the opposite thing.
Last Thursday the ten-year was printing fresh five-day highs on its way above 4.92 percent, and that was the single strongest bearish signal available for a duration-sensitive index. On September 17 it had climbed further, to a five-day high of 5.025 percent, and then reversed hard: down to 4.947 percent, below the five-day EMA of 4.966, with an intraday low of 4.943 after touching 5.008. Not a lower absolute level than last week, a lower direction. Yields had stopped making highs and started retreating from them.
VIX did the same thing more plainly. Last week it ran from the mid fifteens to eighteen. On September 17 it sat at 15.73, below its five-day EMA of 16.61, below the prior day's low of 16.40, and well below the prior close of 17.72. That is not a drift, it is a collapse in hedging demand.
Brent had broken past 105 last week and added inflationary pressure on top of the rates story. It was at 103.05 against a prior close of 105.61, falling hard, which removes that pressure and reinforces the yield decline rather than fighting it.
DXY at 100.12 was marginally above its five-day EMA of 100.01, fading from a session high of 100.38 and below the prior close. Neutralised rather than supportive, which is worth naming honestly: it was the one input not confirming.
The morning's data was strong. Philly Fed manufacturing at 37.8 against a 31.3 forecast, jobless claims at 196K against 207K. In a higher-for-longer environment strong data would normally push yields up. Yields fell anyway, which suggests the market had already priced the Fed's stance and was rotating into growth on its own terms.
Trending Pullback Continuation. The pattern is the most ordinary one in the book. What earns it an article is the direction, because this desk had not taken a winning long since September 3 and had spent the intervening two weeks almost exclusively short.
What the agents were saying
Both agreed, which had not happened on the long side in a fortnight. The Macro Agent read lean_bull at 76 percent, citing the Nasdaq's outperformance, flat the prior session while the Dow dropped 1.21 percent, and rotation into AI and semiconductor names. The Trend Agent read BULLISH at 78 percent in a TRENDING regime with invalidation at 29,313.9, and it had produced four consecutive bullish calls with zero direction changes across four hours.
The 60-minute structure backed them: price at 29,432 well above the fast EMA at 29,240 and the slow at 29,159, a clean bullish stack, with RSI at 67.8 and MACD firmly above zero.
The gap that refused to fill
Price had gapped roughly 180 points above the prior day's high of 29,252 and then did not come back for it.
That distinction did most of the work. A large gap that fades is a liquidity event; a large gap that consolidates near its highs for two hours is buying that has not finished. Each hour the gap stayed open weakened the gap-fill thesis and strengthened the continuation one, which is why the setup was a pullback entry rather than a fade.
Where the entry sat
The five-minute chart had pulled back from 29,457.5 to a session low of 29,313.9 and then built an ascending series of higher lows, 29,313.9 to 29,347 to above 29,400, riding the five-minute EMA9 at roughly 29,399. A fresh five-minute MACD crossover had just turned the histogram positive.
Professional traders: the stop went at 29,295, below the session low and below the Trend Agent's 29,313.9 invalidation rather than above it, which is the correct side and a distinction worth being pedantic about. A hundred point band against a 60-minute ATR of 75.9 clears the one-ATR minimum with room.
SkyAnalyst does not favor any single strategy, and it does not carry a directional view between sessions either. It reads the tape first. Two weeks of shorts did not make this a short desk, and one long does not make it a long desk.
NAS100 NY AM Session Analysis: Thursday, Sep 17, 2026
Market Environment Summary
The NAS100 is trading in a distinctly favorable cross-asset environment this morning. 10Y Treasury yields have dropped sharply from yesterday's close of 5.020% to 4.947%, now sitting below the 5-day EMA (4.966%) and pulling away from the 5-day high of 5.025%. This yield retreat is the single most powerful bullish signal for the Nasdaq, which is the most duration-sensitive US equity index. The move is notable: yields touched 5.008% today but have since reversed hard, establishing an intraday bearish yield structure that directly supports NAS100 upside.
VIX has collapsed to 15.73, below its 5-day EMA (16.61), below yesterday's low (16.40), and below yesterday's close (17.72). This represents a significant risk-on shift. DXY at 100.12 sits marginally above its 5-day EMA (100.01) but has faded from its 100.38 session high and trades well below yesterday's close of 100.318, effectively neutralized as a headwind. Oil (Brent) is falling hard (103.05 vs 105.61 close), which reduces inflation fears and further supports the yield decline narrative.
This morning's economic data was unambiguously positive: Philly Fed Manufacturing at 37.8 (vs 31.3 forecast) and Unemployment Claims at 196K (vs 207K forecast), both released at 8:30 AM ET, well behind us. Strong data in a higher-for-longer environment would normally be a yield headwind, but yields are falling anyway, suggesting the market has already priced in the Fed's stance and is now rotating into growth/tech on secular AI tailwinds. The Macro Agent confirms this with a lean_bull bias at 76% confidence, citing NAS100's clear outperformance (flat yesterday while Dow dropped 1.21%) and rotation into AI/semiconductor names (NVIDIA +0.8%).
NAS100 has gapped above yesterday's high (29,252) and is now trading at ~29,432, a +460 point move from yesterday's close of 28,971. The gap was substantial (~280 points above yesterday's high), but rather than filling, price has consolidated near highs, indicating genuine buying pressure rather than gap-fade mechanics. The Trend Agent reads BULLISH at 78% confidence in a TRENDING regime with invalidation at 29,313.9, four consecutive bullish calls with zero direction changes in 4 hours.
Directional Bias: Bullish Volatility: Normal (VIX 15.73, declining; 60m ATR ~76 pts, contracting)
Step-by-Step Confluence Assessment
1. 10Y Yield Assessment
- Current: 4.947%, below 5-day EMA (4.966%)
- 5-day high: 5.025% (yesterday), yields are retreating, NOT making new highs
- Intraday trajectory: touched 5.008% then reversed to 4.943% low, now at 4.947%
- Verdict: ✅ Yields falling = bullish for NAS100. Default bias: LONG.
2. Macro Regime & Cross-Asset
- Macro Agent: lean_bull, 76% confidence, citing tech leadership and AI rotation, ✅
- Rate factors in macro: Fed rate hike is "priced in" (partially), no yield alarm, supportive, not a headwind
- VIX: 15.73, below 5-day EMA (16.61), below yesterday's low, ✅ risk-on
- DXY: 100.12, barely above 5-day EMA (100.01), fading from highs, neutral/slight headwind, not confirming bearish
- Cross-asset verdict: VIX strongly confirms bullish; DXY neutral. No maximum-conviction short signal present.
3. Trend Structure & Key Levels
- Trend Agent: BULLISH, 78% confidence, TRENDING regime, invalidation 29,313.9
- 60m EMA alignment: Price (29,432) >> EMA fast (29,240) >> EMA slow (29,159), clean bullish stack ✅
- 60m RSI: 67.8, strong but not overbought (was 73.8 one candle ago, pulled back)
- 60m MACD: Line 82.87, histogram 26.85 (strong), above signal and zero, ✅
- 60m VWAP: 29,228, price ~200 pts above, extended but in trending regime
- Daily context: Price above yesterday's high (29,252), above 5-day EMA (29,188), above yesterday's close (28,971)
- Pre-market gap: ~180 pts above yesterday's high, large gap, but NOT filling (consolidated near highs for 2+ hours), gap fill thesis is weakening with each passing hour
- Both agents agree (Macro lean_bull 76%, Trend BULLISH 78%) with yields supportive, strongest setup foundation confirmed
4. Lower-Timeframe Entry Analysis
15-minute structure:
- EMA fast (29,361) > EMA slow (29,277), bullish stack ✅
- RSI: 64.3, above 50, supporting longs ✅
- MACD: Line 50.95, histogram -0.09 (flattening near zero after pullback), momentum pausing, not reversing
- Price consolidated in a 29,314-29,460 range over the last 5 candles, healthy consolidation after impulse
5-minute structure:
- EMA fast (29,399) > EMA slow (29,364), bullish ✅
- RSI: 61.5, above 50, neutral-bullish ✅
- MACD: Line 12.8, histogram just turned positive (+1.77), fresh bullish crossover emerging ✅
- Price has formed a clear higher low: 29,313.9 → 29,347 → 29,400+, ascending support structure
- 5m EMA9 at ~29,399 is acting as dynamic pullback support, price is riding it higher
- Key Fibonacci levels (5m): Using swing 29,313.9 (NY session low) to 29,457.5 (session high): 38.2% = 29,402.6, 23.6% = 29,423.6, price is bouncing between these levels, holding above 38.2%
Entry type identified: This is a trending pullback continuation setup, price pulled back from 29,460 to 29,314, found support, and is now grinding higher with a fresh 5m MACD bullish crossover. The ideal entry is on a pullback to the 5m EMA9/Fibonacci support cluster.
5. Confluence Gate Scoring
| # | Factor | Status | Score |
|---|---|---|---|
| (i) | 10Y yield direction supports long | Yields falling below 5d EMA | ✅ |
| (ii) | Macro Agent aligns (≥60%, rate factors) | lean_bull 76%, growth rotation cited | ✅ |
| (iii) | Trend Agent aligns (≥60%) | BULLISH 78% | ✅ |
| (iv) | 60m EMA stack confirms | Price > Fast > Slow, clean bullish | ✅ |
| (v) | Price at structural level with 5m reaction | At Fib 23.6%-38.2% pullback zone with MACD cross | ✅ |
| (vi) | 15m RSI >50 with MACD histogram expanding | RSI 64.3 ✅; MACD hist flat at -0.09 (not expanding) | ⚠️ Partial |
| (vii) | No high-impact events within 30 min | Next events already passed (8:30 AM), calendar clear | ✅ |
Score: 6 of 7 = High (7.5-8.5)
The only partial miss is the 15m MACD histogram, which has flattened near zero rather than expanding, consistent with consolidation before the next leg rather than deterioration. The 5m MACD has just turned positive, suggesting the 15m will follow shortly.
6. Risk Management Calibration
- 60m ATR: 75.9 pts, minimum stop width = 76 pts (1x ATR)
- VIX declining, range compressing → can use tighter structural stops within ATR constraint
- Structural stop level: Below NY session low at 29,313.9, add 15 pts NAS100 overshoot buffer = 29,298
- Trend Agent invalidation: 29,313.9, our stop at 29,298 is below this, which is correct (stop below invalidation, not above)
- Stop distance from entry zone (29,405-29,420): ~110-120 pts, exceeds 1x ATR ✅
- R:R assessment:
- TP1 at session high 29,460 = ~45 pts = 0.4R, too close, not sufficient alone
- TP1 at 29,530 (60m ATR upper band) = ~115 pts = ~1.0R ✅
- TP2 at 29,580 (1.5x ATR extension) = ~170 pts = ~1.5R ✅
- TP3 at 29,700 (round number + measured move from today's range) = ~290 pts = ~2.5R
Wait, let me reassess. The entry needs to be optimized. If we wait for a pullback to the 29,380-29,395 zone (5m EMA9 + Fib 38.2%), the stop distance to 29,298 becomes ~85-95 pts, and the targets become:
- TP1 at 29,460 (session high) = ~70 pts = 0.8R, still below 1R
- TP1 at 29,500 (round psychological + upper VWAP band area) = ~110 pts = ~1.2R ✅
- TP2 at 29,580 = ~190 pts = ~2.1R ✅
- TP3 at 29,700 = ~310 pts = ~3.4R
This profile works. TP1 at 29,500 achieves 1.2R at a psychological level above the session high (which serves as a breakout acceleration point). TP2 and TP3 are strong with both agents bullish and yields supportive.
NAS100 divergence check: US30 (Dow) at 51,786 is within yesterday's range and below its 5-day EMA, NAS100 is significantly outperforming. ADD (breadth) current reading is 1,001, positive but off highs (was 1,624 earlier). This confirms sector rotation into tech/growth, not a broad-market rally. This is flagged as sector rotation risk but is actually supportive of the NAS100-specific long thesis, money is flowing INTO NAS100 constituents.
Trade Setup
Setup #1: NAS100 LONG, Pullback Continuation
- Bias: Bullish trending continuation on yield decline + tech rotation + multi-timeframe alignment
- Entry Zone: 29,380-29,400 (5m EMA9 cluster at ~29,399 + Fib 38.2% at 29,403 from NY session low-to-high swing)
- Entry Trigger: 5m candle closes above 29,400 with body above EMA9 after touching the 29,380-29,400 zone, OR price holds 29,400 for two consecutive 5m candles with RSI above 55. If price does not pull back to this zone and instead breaks 29,460 directly, entry shifts to a breakout retest of 29,450-29,460.
- Stop Loss: 29,295 (15 pts below NY session low of 29,313.9 and below Trend Agent invalidation, accounting for NAS100 overshoot tendency; ~100 pts from mid-entry = 1.3x 60m ATR)
- Targets:
- TP1 = 29,500 (psychological round number + breakout extension above session high 29,460)
- TP2 = 29,580 (60m ATR upper band projection + London session extension)
- TP3 = 29,700 (measured move target: today's range ~505 pts projected from consolidation base at 29,314)
- R-Multiples: TP1 = 1.0R-1.2R, TP2 = 1.8R-2.1R, TP3 = 2.9R-3.4R (varies with exact fill)
- Partial Management: Secure 50% at TP1, trail stop to entry on remainder, secure 30% at TP2, let 20% run to TP3 with trail below each 15m swing low
- Quality Score: 7.5/10
- Confidence: Medium-High, Six of seven confluence factors align. The primary caution is that price is ~200 pts above VWAP (extended), and the 15m MACD histogram has stalled. The setup relies on trend continuation rather than mean-reversion, which is appropriate given the TRENDING regime call but demands patience for the pullback entry. If 29,313.9 breaks, the bullish thesis is invalid, respect the stop without exception.
Key Risks & Watchpoints
- If price breaks below 29,314 (NY session low): Invalidates the trend structure. No longs. A break of this level with expanding volume could trigger a gap-fill toward 29,252 (yesterday's high) and potentially 28,971 (yesterday's close).
- If yields reverse above 5.00%: Reassess long bias immediately. A close above 5.025% (5-day high) would negate the bullish yield thesis, flatten any remaining position.
- Sector rotation risk: NAS100 is outperforming while Dow underperforms and breadth (ADD) is fading from highs. If ADD turns negative while NAS100 remains elevated, this is a distribution signal, tighten stops to 29,370.
- Session timing: We are 34 minutes into the NY session. The first 60-90 minute window remains active, this is the highest-probability entry window. If the pullback doesn't materialize by 11:15 AM ET, the setup quality degrades as we enter the midday lull.
- Position sizing: Given normal volatility (VIX 15.73) with a ~100 pt stop, standard 1% equity risk. The setup is clean but the extension from VWAP warrants conservative sizing, do not exceed 1% risk on this entry.
Decision log
14:47 UTC, 68 percent, ENTER. One evaluation. Yields retreating below their five-day EMA after failing at 5.025, VIX at 15.73 and collapsing, Brent falling, both agents bullish at 76 and 78 percent, and a clean 60-minute bullish stack with price riding the five-minute EMA9 after a higher low. Six of seven confluences cleared. The one that did not is worth stating: the 15-minute MACD histogram had flattened to -0.09 rather than expanding, which is consolidation rather than deterioration but is not confirmation either, and the 68 reflects it. The Risk Agent placed the stop at 29,295, below both the session low and the Trend Agent's invalidation, and bought 29,395.
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 |
|---|---|---|
| Stop hit (invalidated) | -1R | −$2,000 |
| TP1 hitActual | +1.05R | +$2,100 |
| TP2 hit | +1.85R | +$3,700 |
| TP3 hit (max potential) (not tracked) | +0R | +$0 |
All six agents combined.
What this trade teaches
The useful thing here is not the trade, it is the sequence around it.
This desk published ten short case studies in a single week and every one of them was correct. It then took four more shorts on Monday and lost all four, took this long on Thursday and won, and lost another short on Friday. The macro inputs that made the first ten work had reversed by the time of the last five, and the record shows the desk was slower to turn than the tape was.
That is worth publishing precisely because it is not flattering. A system that reads the tape rather than carrying a view should turn when the tape turns, and four stop-outs in one session is what it looks like when the turn is recognised a beat late.
The second point is the one we are obliged to make against ourselves. Eight days ago we published a recap arguing that an eighty percent week on ten trades proved nothing, because the sample was too small to distinguish skill from variance. That argument does not get suspended when the small sample happens to support a more interesting story. One winning long is one trade. It is the first evidence of a turn, not the confirmation of one.
From the desk
September peaked at +8.87R on the tenth, after a Thursday where four shorts opened inside forty-eight minutes and all four paid. It sits at +4.92R now.
That is a 3.95R drawdown across six trades, and the composition matters more than the number. Five of the six were shorts and all five lost. The one that was not a short is this trade. The short book is not broken and we are not switching it off on five trades, but the tape that made it work through the first half of the month is visibly not the tape in front of it now.
The NAS100 book itself has now taken seven September trades: four shorts, three longs, and a record of +3.46R on the conservative baseline the recaps use. It lost 2.13R across four trades in August and kept its allocation because four trades was not a sample. That decision continues to look better than it did, and four plus seven is still not a sample.
What happens next is a question about the regime rather than about this position. If yields keep retreating and VIX keeps falling, the long setups will keep clearing the gate and the desk will keep taking them. If the last two sessions were a pause inside a rate shock rather than the end of one, the shorts will come back. The system does not need to know which, and neither do we, because it re-reads the tape every session and carries nothing between them.
At a Glance
What this teaches about AI-driven trading
The desk lost four shorts on Monday. Was that a mistake?
It was the cost of recognising a turn a session late, which is a real cost and worth naming. Each of the four cleared its gate against the conditions present at the time and each stopped out at exactly 1R, the designed size. A system that reads each session fresh will sometimes be trading the previous regime on the day it ends. The alternative, carrying a directional view, fails worse and more often.
What does +1.85R mean in dollars?
R is the trade's risk unit, the distance from entry to stop, here 100 points. On a $100,000 account risking 2% per trade, 1R is $2,000, so +1.85R is roughly $3,700. Reporting in R rather than dollars keeps results comparable across account sizes and across instruments whose point values differ.
Why buy a market that had already gapped 180 points?
Because the gap did not fill. Price consolidated near its highs for over two hours instead of fading, which distinguishes genuine buying from a liquidity event. The entry was still a pullback rather than a chase: price had retraced to the five-minute EMA9 and built a higher low before the position went on.
Does this mean SkyAnalyst has turned bullish?
No. The system holds no directional view between sessions. It scored a long on September 17 because yields were retreating, VIX was collapsing and both agents read bullish, and it scored a short on September 18 on a different instrument under different conditions. Two weeks of shorts did not make it a short desk and one long does not make it a long desk.
How do you count R on this trade?
This case study reports full potential, the R distance to the furthest target price reached, here the second target at 29,580. Our weekly, monthly and year-to-date recaps use a stricter TP1 baseline crediting only the first target, so this trade enters those totals at +1.05R. The two numbers are deliberately different and we never mix them.
Run your markets with SkyAnalyst
21-day free trial. No credit card. Full access to the Trend Agent, Macro Agent, and six-factor confluence scoring.
Trading involves substantial risk of loss. Past performance is not indicative of future results. The analysis shown was produced by an AI model operating on SkyAnalyst’s live trading infrastructure; it is shared for educational and research purposes only and is not financial advice. About reported results. Every AI Trader publishes three take-profit targets (TP1, TP2, TP3) per trade. The broker closes 100% of the position at TP1, so two distinct R-multiples appear in this article. The hero R-multiple is the full-potential R: where the market actually traveled (the highest take-profit hit, or the stop loss) before the setup was invalidated or exhausted. The realized R, shown on the TP1 row of the simulated returns panel, is TP1’s R (or -1R on a stop out). The realized R is what we log to our running track record. Both numbers are honest. Showing both is what lets readers see the full arc of the move and the conservative ledger entry it produced. Simulated returns in this article are calculated against a hypothetical $100,000 account at 2% risk per trade (1R = $2,000). These are educational reference figures and do not reflect any specific account or broker execution. Your actual result depends on your position size, your risk parameters, and live market conditions.
From the SkyAnalyst Journal
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.
The Week Our Own Selection Gate Refused to Pick a Loss
This report normally tears down the best-graded loss of the week. This week it could not: neither loss graded above C+, because almost nothing did. Two stop-outs, minus 2R, and a 1.91 percent drawdown.
Four Shorts in Forty-Eight Minutes, on One Reading
Between 14:20 and 15:08 the desk sold Cable, the Nasdaq, the S&P and the Dow. One macro read, four instruments, four stops, four winners. This is the Dow leg, and the question is what that correlation costs.