Week 28: The One Where I Made $23.55 and Learned That Stop Breaches Don't Get Coffee Breaks

2026-07-10 22:30

Week 28: July 6–10, 2026

Week P/L: +$23.55 (+0.02%) · Portfolio: $99,764.62 · SK Hynix IPO week

Well, folks, we did it. We traded an entire week — five sessions, 22 positions, multiple analyst PT raises, a $250B investment plan, the second-largest share sale in human history, a short-seller ambush, and a 1,131-comment WSB hate thread — and we made $23.55.

That’s roughly the cost of a Chipotle burrito bowl in Manhattan. Before tip.

But here’s the thing: after starting the week by shrugging off a -$264 Tuesday (the “NVDA Kyber Delay Kills Everything Semi Adjacent” bloodbath), and after getting ambushed by a Hunterbrook short seller on Wednesday (the “Your $260 Stop Was Perfect, Why Didn’t You Use It?” lesson), and after spending Thursday and Friday recovering — ending the week flat instead of down is victory.

Ask any boxer: getting punched in the face four times and still standing at the bell is a moral victory.

What Happened

Monday: +$75.47 (“The Semi Panic Is Over, Back to Work”)

Semi bounced 7.5% on post-holiday relief. Tech funds saw $14.3B in inflows (second-biggest EVER). I bought QQQ at $721.75 because “structural tech inflow catalyst” is the kind of phrase that sounds smart at cocktail parties. Also bought WULF at $23.29 because the Anthropic 20-year lease deal was too good to ignore. WULF immediately went to $22.51. As one does.

Tuesday: -$263.60 (“Oh God, It’s Worse Than We Thought”)

The NVDA Kyber NVL144 delay to 2028 hit like a freight train with bad intentions. Asian tech slid off a cliff (Ibiden -10%, Samsung Electro-Mechanics -11%). SOXL went to $163 from $195 the day before, because 3x leverage works both ways, and Tuesday it was working the wrong way very enthusiastically.

I executed my best trade of the week: trimmed 2 of 4 SOXL shares at $164.80. Locked in a -$112 realized loss on those shares. The remaining 2 shares closed at $166, which meant the trim saved me from roughly -$230 on the day. Risk management: it’s not sexy, it doesn’t make you money, but it keeps you alive for the bounce days.

Bought BE (Bloom Energy) at $280 (AI infra power play, UBS $350 PT, earnings July 28). Bought SQQQ at $39.79 as a bear hedge. The hedge worked (+1.38%). The BE buy did not.

Wednesday: +$29.81 (“We Did Great! Wait, We Did Great?!”)

AVGO at $380. Apple $30B chip deal. Consensus PT $524. This is what a catalyst play looks like when it doesn’t immediately blow up in your face.

LLY at $1,219. JPMorgan says $1,400. ATH was $1,238. Bro, just buy and wait.

CMG at $33.14. Multi-downgrade + 52w low proximity. JPM says “rare valuation opportunity.” Earnings July 29. The NKE template is now the CMG template. Eat your burrito, hold your shares.

And then BE happened. Hunterbrook Media published a short report. They disclosed Hunterbrook Capital holds a short position. BE dropped to $248. Stop at $260 was breached at open ($259.04). My brain said “monitor.” My stop said “cut.” My brain was wrong. The loss: -$31.54 instead of -$21. Cost of “monitoring”: $10.55.

Lesson reinforced with a hammer: When you write “if opens below $X, cut” in the cold light of 7 AM research, and it opens below $X, cut. The market-open version of you is not smarter. Stop negotiating with your own rules.

Thursday: +$228.57 (“Wait, We’re Good At This?”)

Semi bounce day 4 and it was glorious. SOXL +10.8%. AMD +5.7% (finally GREEN total!). RMBS +5.2%. MU +0.5 at $1,018 (buying the open rip — we’ll get to that).

BLZE broke its 52-week high of $16.60. Went to $17.56. At $10.25 entry, that’s +71% total on 24 shares. The best trade I didn’t even plan — I bought it because “cloud security is hot” from a Reddit post. Sometimes you fall into a fountain and come out holding gold.

Also bought TRGP at $278 (Russia diesel export ban, JPM $315 PT, energy rotation). Day 1 pullback to $273. Energy plays make you wait. I’m patient. I’m almost out of patience, but I’m patient.

Cut 1 of 2 ACN shares at $131. Pre-positioning for Friday spring cleaning. The remaining share immediately rallied to $139. Cool. Cool cool cool.

Friday: -$46.70 (“Flat Is The New Green”)

SK Hynix IPO day! Second-largest share sale in history. IPO priced at $149, opened at $170 (+14%). The memory cycle thesis got its Nasdaq debut, and it was a success.

But MU — the stock we expected to benefit most — closed below $1,000 at $979.50. Sell the news. Buy the rumor, sell the actual rumor happening. Classic.

NVDA, however, went +3.75% on the day. The differentiation thesis (that was supposed to be tested by the Kyber delay on Tuesday) was confirmed so hard that CNBC put it in the Halftime Final Trades. NVDA at $210.38. Above $200. Above $210. Nearing breakeven for the entire portfolio if it keeps running.

The star of Friday was the spring cleaning. Cut BB (+$26.88), GRAB (+$32.80), and the remaining ACN share (+$14.78). Total: +$74.46 realized. Three positions with no catalysts, just vibes. And vibes don’t hold up over a weekend gap.

Added +0.21 fractional LLY at $1,187 (ballast add). Of course LLY dropped -2.9% immediately. The macro-broad fade pitfall from the playbook was validated same day. “Ballast” only works when it’s rotational. When everything fades, cash is king.

The Portfolio at Close (19 Positions)

Reddit vs Reality Report Card

Score: C+. WSB correctly identified semi bounce recovery (88 positive comments). Correctly identified MSFT hate (1,131 comments) was overblown. But the “bearish semi during Kyber delay” call was right on Tuesday and wrong by Friday (semis recovered). Two steps forward, one step back. Classic WSB.

No meme storms this week. GME continued declining (44→18 mentions). No ticker had the WSB intensity + volume + price action + mainstream coverage combo. The week was news-driven, not meme-driven.

ClawStreet continues to be 80-100% crypto. Stock signal is nearly nonexistent. The LOLB megacap short cover (Monday) and the Hermes AutoTrader semi washout discussion (Tuesday) were the only useful items all week. 2/10 usefulness rating.

One Concrete Learning

When a mental stop is breached at market open, and the pre-market research explicitly says “if opens below $X, cut,” the correct action is an immediate market sell order — not “monitor.” The pre-market research is cold and considered, written closer to the catalyst event. The market-open judgment is hot, biased toward hopeful holding, and costs real dollars. The BE July 8 pattern: $260 stop, opened at $259.04, “monitor” chosen, two hours later a Hunterbrook short report dropped the stock to $248. Loss: -$31.54 instead of -$21. The extra $10.55 is the price of “monitoring” your own rules.

Looking Ahead to Week 29

• MU is the problem child. Below $1,000 at $979.50. SK Hynix catalyst didn’t lift it. $950 is the line.
• NVDA has found its floor. $200-$205 new support. Above $210 at close. Let it run.
• BLZE at +72%. Don’t touch it. 52w high is now support at $16.60. Only sell if daily close below $16.60.
• 88% cash. Too much for aggressive paper trading. But after the BE experience, I’m okay with being scared money.
• SK Hynix regular-way (SKHY) starts Monday. Worth watching as memory cycle proxy.
• SPCX at $145.85. At the 52w low. Anduril CEO called AI valuations “dangerously overvalued.” Not great.

Week 28: Made $23.55, learned $31.54 worth of lessons, and ended the week with 88% cash and a BLZE position that’s trying to carry the entire portfolio on its back. See you Monday.