SPY drops 1.05% as just 5 of 63 setups broke out and 16 failed
Market recap, bot performance, and scanner analysis for Friday, July 17.
The UnxEdge breakout scanner only found 63 setups today, and that scarcity mattered because the tape was hostile enough that clean wedge pattern breakouts were rare but the short-side names that did trigger paid hard.
With SPY off 1.05% and QQQ down 1.54%, this was another session where weak leadership in growth kept pressure on anything tied to high multiple tech, especially semis. Retail is now ROTATING IN week over week while semis keep ROTATING OUT, and that showed up directly in the scanner with failed longs and cleaner bear follow-through in names like AMAT and MDB.
There was no major economic print to blame, and that matters. On a clear macro calendar, the market had no outside excuse, so this looked more like internal rotation and risk reduction than headline shock. The news flow was mostly noise, from thematic ETF launches to crypto chatter and Nasdaq noncompliance notices, which meant price had to stand on its own. It did, and the message was simple: money kept moving away from semis and broad growth beta, while defensive relative strength stayed concentrated in retail and energy. For context, here is yesterday's debrief.
Market Snapshot
Thin opportunity set, ugly resolution quality
The scanner leaned bearish from the start with 37 bear setups against 26 bulls, and the resolution count backed that bias up. Five breakouts against sixteen failures is not a broad momentum tape. It is a selective tape where chasing average-looking longs gets punished. Setup grades and trigger levels still matter here, and traders should see live setups in the scanner instead of forcing second-tier names.
Yesterday's Pick
OKTA bull failed fast
No spin needed. OKTA did not survive the growth unwind and stopped out. In a market where QQQ was getting sold and semis stayed weak, the burden of proof for bullish software names was higher than the chart alone suggested.
Arxe Pick of the Day
ROKU had the cleanest internal edge, even in a bad tape
ROKU earned the nod because it paired the highest pressure reading at 70 with the largest usable backtest sample at 14 trades, while still holding one of the best win-rate and profit-factor profiles in the group. It was not a great market for longs, but among imperfect options, this was the most statistically coherent one.
The problem is that a good setup inside a bad index tape is still fighting math. With growth under pressure and semis dragging sentiment, the bullish case needed clean follow-through that never really broadened. If you want the active grades and breakout levels behind selections like this, see live setups in the scanner.
ROKU scanner chart – July 17, 2026 | UnxEdge
Today's separate Arxe swing pick was BX BULL (HIGH), but neither system traded it. That likely came down to pre-filter rejection or the setup failing contract and trigger requirements before a valid entry could print. When a setup does not satisfy execution rules, skipping it is correct, even if the headline pick looks attractive on paper.
Bot Performance
Paused, and that is the honest answer
Wex and Xcel are in training mode while strategy is being refined. The scanner and Arxe intelligence remain fully active. When trading resumes, every trade will be published here. You can watch Wex and Xcel trade live in the Edge Lab when execution is back on.
Missed Trades
The bots sat out a day that favored downside momentum
Zero breakouts traded, three missed, +7.49R left on the table. Two of those were full TP3 runners. That is the kind of gap that forces a review of filters, especially on short setups tied to sector-wide weakness.
Breakout Scorecard
More failure than expansion
Notable names that did break included AMAT bear A-, BIDU bear B, LCID bull B+, MDB bear A-, and ROKU bull B+. The split tells the story: a few names worked, but there was no broad-based lift. Bears had cleaner index alignment than bulls.
Watch Tomorrow
The unresolved A- names worth stalking next
This is the list that matters. These are the higher-grade names still coiling, and each needs a real trigger, not anticipation. For updated live levels, see live setups in the scanner.
If semis remain a drag, AMD and AVGO are the most obvious continuation names. If software weakness broadens, CRWD becomes more interesting. The key is patience. This tape has been punishing early entries.