US30 Short: A Dow Bounce Fade Filled 28 Minutes Into a 30-Minute Window

Where the selling came from

When the analysis was written at 10:35 AM ET, breadth was one-sided. The NYAD proxy stood at -571 against a 5-day EMA of +144, under the prior day's low and at or near the session low. The analysis treated that as a fresh downside extreme and as a veto on longs. VIX was 19.84 against its own 5-day EMA of 18.30 and above the prior day's high, which the plan read as a wide-stop, risk-off tape.

Rates and the dollar were not the cause. The 10-year yield was below its 5-day EMA and DXY sat near or slightly below its own. The weakness was coming from breadth and risk sentiment, and the analysis argued that this kind of selling tends to keep Dow bounces sellable, because no cross-asset reversal is there to squeeze them. By 11:03 AM ET the Macro Agent's US indexes read named the source: Amazon and Intel earnings had extended a technology-led selloff, with Nasdaq futures down about 2.5%. It rated the group bear at 76 percent, while noting that with VIX at 19.16 this was not a full disorderly liquidation.

The Dow read softer than the group

That 76 percent described the index group. On the Dow itself the Macro Agent was less sure. The setup analysis carried a lean bear at 64 percent with tradeability at 55 out of 100, and the 11:03 AM ET row had that down to lean bear at 59 percent, tradeability unchanged. The bearish case was the break of the prior day's low at 52,184.8 and a breadth reading of -501. The counterargument was the Dow's relative resilience against the Nasdaq 100, which could mean rotation into value and cyclical names. The same row flagged that chasing weakness after the break was unattractive.

The plan had counted "Macro Agent bearish ≥60" as a met confluence at 64 percent. At 59 percent, less than three minutes before the fill, that line no longer held. None of the entry reads mentions it.

The Trend Agent stayed put

The Trend Agent had no such doubt. It rated the Dow bearish at 88 percent in a strong trend when the analysis was written, and bearish at 86 percent at 14:43 UTC, with no direction change across six decisions in four hours. It mapped resistance at 52184.8, support at 51945.4, VWAP at 52328.6 and invalidation at 52217.7. Its main caution was the short-term chart: the 5-minute bounce showed downside momentum waning after oversold conditions, and it called that bounce corrective while price stayed under 52184.8-52217.7.

An earlier plan on this automation shows how quickly the tape had moved. The 10:07 AM ET setup wanted to sell 52205-52220. Price never reached it, and the setup expired at 10:38 AM ET without a single read.

Bounce fade into a retracement. This is a pattern professional traders use when a market has just dropped hard and is too stretched to sell safely. Selling the low means a distant stop and a real chance of a squeeze. So they wait for the first bounce to retrace part of the drop, usually toward the 50 to 61.8 percent area, and sell if the bounce stalls there. The analysis put it plainly: "If neither trigger appears, no trade is better than chasing."

The zone and the trigger

The plan marked 52075-52105 as the sell zone, the 50-61.8% retracement of the 14:00-14:20 UTC selloff. The trigger came in two parts. After a bounce into the zone, it wanted "a 5m rejection candle / lower high", with 5-minute momentum rolling back down and price failing to hold above 52100-52110. The second part turned out to be the difficult one, because the bounce did not stop at that line.

The stop

The stop went to 52212, 5.7 points under the Trend Agent's 52217.7 invalidation. Any 5-minute close above it would kill the idea. The plan asked for structural stops rather than tight ones because volatility was elevated. From the fill at 52093, the stop was 119 points away.

The targets

TP1 at 51945 sat 0.4 points under the session low at the time, 51945.4. TP2 was 51800 and TP3 was 51680. The analysis put them at about 1.1R, 2.3R and 3.2R. From the actual fill they were worth 1.24R, 2.46R and 3.47R. The plan scored quality at 8.1 out of 10 and wrote its confluence total as 6 of 7, although its own checklist marks all seven items yes. We quote the total as written.

The same shape on other mornings

Two other October Dow shorts belong beside Friday's. Our October 23 US30 short sold a bounce into resistance, and the entry there turned on a single closed candle. The October 14 short sold a second leg lower after a CPI-morning low.

SkyAnalyst doesn't favor any single strategy. Last week it bought a Nasdaq 100 pullback to VWAP, and on Tuesday it sold a euro bounce into broken support. The plan follows the morning's chart and macro read, and on Friday both leaned bearish on the Dow, the macro read less firmly than the chart.

What this trade teaches

The 84 and the final 72 are easy to misread. A score on a WAIT read is confidence in waiting, so the fall from one to the other does not mean conviction in the short dropped. The 15:04 UTC read declined a provisional bar. A 5-minute bar ended at 15:05 UTC, and 50 seconds later the next read described the last candle as a strong rejection. The record does not timestamp the bar, but the timing fits a read that wanted a finished candle.

The entry still did not meet the trigger cleanly, and the read did not pretend otherwise. The plan accepted a rejection candle or a lower high, and it got the rejection without the lower high. The close near 52104 sat inside the 52100-52110 band rather than under it, after several candles above it. That is thinner confirmation than the October 23 fill, though more than the S&P 500 retracement short on October 22, where the entering read said no rejection candle had printed.

The trigger is not textbook-perfect because the bounce made a marginal higher high, but the rejection and broader trend alignment still give this short edge now. SkyAnalyst entry evaluation, 15:05 UTC

The clock nobody mentioned

The fill came with about a minute and a half left on the configured window, and none of the four reads refers to it. We see no sign that the deadline pushed the decision. What it controlled was whether a decision was available at all. Of the 25 setups this automation has run in October, 17 expired unfilled, 13 of them within 31 minutes of opening. Six of those 17 had reached their zone, but in four of them, the next setup on Friday included, the automation already held an open short for as long as price was there. Two were free to trigger and did not. A bounce fade needs the market to bounce and then stall, and a fixed window asks it to do that on schedule. A rejection that printed a few minutes later could have found this window closed.

Fast fills and slow fills

Eight of those 25 setups filled. The three that filled within a minute of the window opening all stopped out before reaching a target. The three slowest, at 8.7, 14.7 and 28.5 minutes, all reached at least TP1. The two in between split: one reached TP2 and one was stopped. Eight trades is too few to make a rule from, and we are not making one. We do want to keep tracking whether a fill that had to wait for its trigger behaves differently from a fill at first touch.

From the desk

Our ledger records +1.24R (TP1), booked at 12:46 PM ET, an hour and forty minutes after the fill. The hero panel shows +3.47R (TP3), where the market had traveled by 2:20 PM ET, three hours and fourteen minutes after the fill. The lowest price the monitor logged was 51664.2, 15.8 points beyond TP3. The space between those two R figures is the price of closing everything at TP1, and both stay on the page.

The trade had almost no heat. The worst price logged against the short was 52101.2, 8.2 points above the fill and 110.8 points short of the stop. The record does not say when it printed. Compare that with the October 23 short, which came within about ten points of its stop, or with Tuesday's euro short, which came within 1.9 pips. A late fill is not automatically a worse fill. On this trade, selling after the rejection had printed left the bounce with very little still to run.

The open position had one more effect. The next plan went live 44 seconds after the fill and wanted to sell a break lower at 52075-52085. Price reached that zone 31 seconds later. All 15 of its checks stopped at the same hard gate: "Already have 1 open short position(s) on US30." It never reached an AI read and it expired unfilled. One short on, no second short added.

This is one trade, not a proof. A flawed trigger accepted on the last read of a nearly spent window worked on a day when breadth, the Trend Agent and the tape all pointed the same way. The same habit on a mixed day is how a system pays for a stop. What we take from Friday is the record: the read named the flaw before the fill, and the position limit held while the trade ran.