US500 short: six of six confluences, four reads inside four minutes

Six of six confluences had cleared the short checklist before the Trend Agent ran a single evaluation on June 5. NYAD had collapsed from +984 to -777 overnight, a swing of roughly 1,760 breadth points. VIX was up 9.5 percent on the session while SPX was down nearly a full percent. The 60-minute, 15-minute, and 5-minute charts were all aligned below their fast and slow EMAs. The prior day's low at 7517.5 had already been broken through to the downside. The Macro Agent had logged lean_bear at 45 percent. And the dollar and yields were both surging after a hot NFP print. None of these readings is decisive on its own. Stacked, they form what the Macro Agent calls a structural breakdown regime, and the system knew it before the bell. About reported results. Every AI Trader publishes three take-profit targets (TP1, TP2, TP3) per trade. The broker closes 100 percent 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. What's worth the rest of the article, like our US30 short from earlier this week, is that the system still took four evaluations to enter. Not because the direction was in doubt, but because price was already 45 points below the 5-minute VWAP with the 5m RSI at 17. Chasing a vertical drop on a 6/6 short signal is a separate kind of mistake from being wrong about direction. The Trend Agent ran four reads inside four minutes, between 14:10 and 14:14 UTC, and watched confidence build 35 to 40 to 52 to 68 as the pullback the analysis required actually materialized. The entry came at 7516.8, the literal middle of the 7515 to 7522 pullback band the morning analysis had named. The position closed at TP3 one hour and thirty-four minutes later, plus 2.27R (TP3) on the full-potential arc and plus 1.04R (TP1) on the realized TP1 row.

The tape that made this trade necessary

The pre-open read on June 5 was the kind of tape that does the thinking for you. Non-farm payrolls had printed 172,000 against an 85,000 expectation, lighting the dollar and yields together. US 10-year yields were at 4.546 percent, above the prior day's high of 4.495 percent. DXY was at 99.73, also above prior day's high. The Macro Agent had US500 graded lean_bear at 45 percent confidence, and the cross-asset read had every variable pointing the same direction.

Breadth was the part that wasn't subtle. The NYAD line had closed yesterday at +984, a healthy positive print on a 5-day EMA of negative 218. Today it had collapsed to -777, an intraday swing of roughly 1,760 breadth points with no positive reading anywhere in the session's range. For a 500-stock index, that's the cleanest possible confirmation that selling is broad and not narrow leader-driven. Breadth and price were saying the same thing. The Trend Agent didn't have to look hard.

VIX was up 1.47 points on the session, a 9.5 percent move from 15.39 to 16.86. That kept the regime in the normal volatility band but expanding, with the 15-minute ATR jumping from 5.7 to 7.3. The point that matters: VIX was rising while SPX was falling. That's not divergence, that's confirmation. Together, the macro and breadth picture handed the Trend Agent a permission slip the second the NY session opened.

Inside that regime, the local structure was extended. The 5-minute chart was in a waterfall decline from 7541 down through 7501 in roughly thirty minutes of cash trading. RSI at 17.5 on the 15m, 19 on the 5m. Price was 45 points below the 5-minute VWAP. The 60m had MACD accelerating with a histogram at -5.68 and expanding. The prior day low at 7517.5, which had been support, had already broken to the downside and was now overhead resistance. The Trend Agent's own R-line sat right on top of it at 7517.8.

The setup the system flagged was a short on pullback to broken support. Wait for price to retrace back into the 7515 to 7522 zone. Then require a 5-minute bearish rejection candle inside the band, with the high of the candle staying below 7533 to keep the structural stop intact. Six of the six required confluences were already in place. The seventh requirement wasn't another confluence. It was the pullback itself.

The pullback to broken support, dynamic not dogmatic

The setup the Trend Agent flagged has a name among professional traders: a short on pullback to broken support inside a confirmed breakdown regime. It's one of the most teachable continuation patterns in structured discretionary trading, and it's worth spending a minute on, both because it makes the decision log readable and because it's a window into how the system grades entry timing when direction is already settled.

What the pattern is

Price has broken decisively below a structural support level on the higher timeframes. The level that was buying interest yesterday becomes selling interest today, because trapped longs from above and short covering from the breakdown have both repriced their reference points. After the break, price typically extends well below the level in the impulse leg, then retraces back into the broken shelf as a mean-reverting bounce or short-covering rally. A professional reading this setup does not sell the spike low. They wait for the retest of the broken level and require a bearish rejection candle inside the zone, with volume confirming distribution rather than absorption.

How professional traders actually use it

This is a staple of breakdown-continuation trading. Institutional desks and seasoned discretionary traders use it because the math favors the retest entry far more than chasing the impulse. Rough rule of thumb: breakdown levels are reclaimed roughly 40 to 50 percent of the time on a first retest, and rejected closer to 70 percent of the time when the retest prints a bearish candle on meaningful volume below a clear structural ceiling. That spread is the entire edge the patient short is pricing in.

The tell is the candle structure, not the price tag. A quiet retest with a body that closes near the highs of the test bar is incomplete: the level isn't proving rejection, price is just consolidating below it. A clear rejection with a bearish body and a higher-timeframe close back below the breached level is signal: real offers are stepping in, and the level has confirmed its new role as resistance.

Why it works

Broken support zones exist as future resistance because of resting orders that didn't fill on the way down. Sellers who missed the initial breakdown leave limit orders at the prior level. Trapped longs who held through the break add to those orders when price rallies back to break-even. When price returns to the zone, the first retest often gets faded but not decisively. If a second retest holds, or the first retest prints a clean rejection candle, the supply has proven structural rather than incidental.

It fails in the wrong regime. Range-bound markets and trend-exhausted markets turn pullbacks to broken support into V-reversals, where the broken level reclaims for one or two candles and the impulse reverses entirely. That's why the Macro Agent has to confirm the breakdown is regime-supported before the Trend Agent is allowed to size into this setup. On June 5 the Macro Agent had US500 graded lean_bear at 45 percent, breadth had collapsed by 1,760 points overnight, and VIX was rising into a falling tape. The breakdown wasn't a one-off. It was the regime.

How the system sees it

SkyAnalyst doesn't favor this strategy. That's the important part. On the same day the S&P 500 was setting up its short on broken support, our agents were watching a long-only window on USDJPY into intraday support, a separate Dow long on a pullback reclaim, and a different breakdown-fade on EURUSD. Each of those is a different pattern with different logic and a different edge.

The system reads the tape first and fits the pattern to what's actually there. It does not show up to the chart with a playbook and look for opportunities to run a preferred setup. That's the single biggest difference between how retail traders typically lose money, forcing their favorite pattern onto every chart, and how the system stays structurally honest: it has no favorite pattern. Every evaluation cycle re-reads the regime, re-scores the structure, and lets the confluence math decide which playbook applies, if any. Some sessions, none do. The system sits those out. The pullback to broken support worked on June 5 because June 5 was a breakdown-continuation day. The framework is dynamic, not dogmatic.

What this trade revealed about the system

We publish case studies because the interesting question isn't whether a single trade worked. Plenty of trades work for the wrong reasons, and plenty of clean reads end at the stop. The interesting question is what the trade reveals about how the system handles a specific kind of pressure. This one revealed two things worth naming.

Direction and timing are different problems

The morning analysis on June 5 had the direction settled before the first evaluation ran. Six of the six required confluences had already cleared. Breadth and macro and structure all agreed. A retail trader reading the same chart would have felt the pull to short at the spike low and would have eaten the snapback. The system did not feel that pull. It separated the question of "should we be short" from "where do we get in," answered the first one immediately at 35 percent confidence, and then ran three more evaluations watching the pullback fill before lifting confidence to 68 percent and acting. The discipline wasn't about whether. It was about where.

Four evaluations in four minutes is not the same shape as four evaluations in twenty

The 14:10, 14:11, 14:13, and 14:14 UTC reads happened on a clock that doesn't show up on the chart. The previous case study we published, a US30 long the day before, ran two evaluations a full minute apart while waiting for one trigger candle to print. This trade ran four evaluations on consecutive one-minute updates while the pullback itself was forming. Same number of reads. Completely different mechanism underneath. The first shape is the system waiting for a binary event. The second shape is the system pricing a continuous variable as it moves through a band.

The system did not wait three evaluations because it was uncertain. It waited because the price it wanted to short hadn't arrived yet. — From the post-trade review

The TP3 hit was structure, not luck

Once price rejected the broken shelf at 7517.5, the path to 7480 was already drawn in the higher timeframe levels. TP1 at 7500 was the round number and session low. TP2 at 7494 was the Trend Agent support. TP3 at 7480 was the 60m ATR 2x extension. The market closed the position at the literal TP3 level. That's not a tail outcome. That's the move running its allotted range in a structurally clean regime. We log the +1.04R (TP1) realized number to the track record because that's what the broker closed. We name the +2.27R (TP3) full-potential number because that's what the structure paid. Both are honest. Both are the same trade.

A note, before we move on

The case study we almost wrote about June 5 was a different one. There were three closed trades on the journal that day, and the obvious pick was the largest realized P&L. We chose the US500 short instead because the educational moment is structural, not performance-driven. A 6 of 6 confluence read on a breakdown day is the cleanest signal the system produces in any week. The temptation to chase those signals at the spike low is exactly the kind of mistake that turns a winning system into a losing trader. Documenting how the system separates the direction call from the entry call is the load-bearing lesson, not the dollar amount.

A reasonable question by now is whether a retail trader with ChatGPT and a clean data feed could reproduce this. They cannot, and not because of model quality. On June 5 the Macro Agent had written lean_bear at 45 percent to the shared state at 09:02 UTC and hadn't updated it since. The Trend Agent, on each of its four evaluations between 14:10 and 14:14, read that single value and used it to unlock the size it eventually took. If the Macro Agent had been chatting in prose about mixed signals, the Trend Agent would have had to interpret the tone. It doesn't, so it didn't. The coordination between the four agents through structured shared state is the product. That's what a chat interface can't simulate, and it's what this case study shows in practice.

The next case study will be a different shape. Different instrument, different regime, almost certainly a different number of evaluations. We file these as the position closes. See SkyAnalyst run your markets to follow along on your own setups.

— The SkyAnalyst Team