How a 5-1 Model Vote Turned SOXL's Deep-Discount Bounce Into +16.7%
View original analysis report →
The receipt
Called SOXL long on 2026-07-30 at 10:04 KST. Entry zone 86-91 (midpoint 88.5), stop 82.5, T1 101.53, RR 2.17 on paper. The actual fill printed at 93.7, above the planned zone, and the position closed at 109.36 - a 16.7 percent move captured. Realized P&L: 6223.90.
The setup
At call time the 1d was fully inverted - price sitting beneath ema20, ema50, and ema200 - with a textbook down channel of lower highs at 302, 272, 165.75 and lower lows at 157.56, 150.08, 116.47. Nothing about it looked safe. The SOX index had fallen more than 20 percent off its June high into a bear market, and 82 percent of respondents in BofA's fund manager survey called long semiconductors the most crowded trade on record. The only defense still standing was the 12h ema200 at 84.94. That single line was holding up the entire long thesis.
The signals
Five of the six perspectives went long; only Macro/Trend argued short. But the long-side confidences sat at 41 (Elliott), 47 (Wyckoff), 52 (ICT), 54 (Divergence), and 58 (News) - all low, which made this a weak majority rather than a strong consensus, and the final confidence of 52 reflected exactly that. The 1h RSI traced a regular bullish divergence, climbing from 24.5 to 27.8 to 29.0 across lows of 115.45, 101.53, and 92.41 even as price kept printing new lows, with the MACD histogram rising alongside it from -3.73 to -2.16 to -1.34. The 1d RSI, by contrast, made a new low right along with price - no divergence there, meaning there was fuel for a bounce but no evidence of a trend change. Wyckoff read the 12h bar - a record 212.5M-volume drop from 129.70 to 92.41 that closed back up at 97.05 - as a Selling Climax candidate. The most ironic detail: the single dissenting vote, Macro/Trend, had marked 108-116 as its own short re-entry zone, and this trade closed at 109.36, right inside that exact band.
What would have killed it
Invalidation was a decisive 12h close below 84.94. Break that and 78.67 opens up, and the ICT discount-long, Wyckoff Selling Climax, Elliott wave-5-exhaustion, and divergence-bounce theses all collapse at once. In practice price never got near that line and bounced straight away, but the fill itself deserves an honest note: the planned limit zone was 86-91, and the actual entry printed at 93.7, above it. On top of that, this was a countertrend trade taken at low confidence (52), with the 1d trend and the macro backdrop both pointing the other way, and AMD earnings on 8/4 plus NVIDIA earnings on 8/28 sitting inside the bounce window the entire time.
The lesson
5-1 is not automatically a strong signal. When five perspectives agree on direction but each carries confidence stuck in the 40s and 50s, that is agreement on direction, not agreement on conviction. What made this one work was a deep-discount entry price paired with a single, explicit invalidation line at 84.94 that offset the weak conviction. Count the votes, but also read where the one dissenting vote is pointing - it often marks a level worth respecting either way.