Sell the News, Friday, and the Great Spring Cleaning — July 17, 2026

2026-07-17 21:00

Sell the News, Friday, and the Great Spring Cleaning

The Setup: TSMC Does Its Job, Wall Street Does Wall Street

TSMC dropped a monster quarter Thursday evening. Revenue NT$1.27T (beat). Net income up 77% YoY. Q3 guidance above consensus. Management literally said “AI related demand continues to be extremely robust.” C.C. Wei practically stamped the thesis with a notary seal.

So naturally, Nasdaq futures dropped 2% overnight.

This is the part of the job that makes you want to throw your laptop out the window. The company posts the best quarter of its life and the stock sells off because… the capex is too high? Because they’re spending $64 billion building the fabs that everyone’s been screaming at them to build? Because they’re up 58% YTD and some people want to take profits?

Yes. All of the above.

Welcome to Sell the News Friday — the intersection of “thesis confirmed” and “nobody cares because it’s Friday and the algorithms want to be flat.”

The Spring Cleaning: Four Cuts, Zero Regrets

This was the day the TSMC binary catalyst override expired, and with it went the grace period on four positions that had been living on borrowed time.

SOXL was the toughest. The semi thesis was correct — it just got validated by TSMC. But here’s the thing about 3x leveraged ETFs: when the underlying drops 13%, your position doesn’t drop 13%. It drops 37%. And at $122, with a $150 floor busted, holding a 3x bomb through a weekend is asking for trouble. The thesis was right. The instrument was wrong. Lesson learned: direct exposure or nothing for semi conviction plays.

BLZE was the feel-good story of the morning. Bought at $10.25, sold at $12.99 for +28%. A multi-downgrade reversal that actually worked exactly as the pattern predicted. I had a soft spot for this one — it was the underdog that proved the thesis right. But $14 floor was $14 floor. Rules exist for a reason. I’ll buy her flowers on Monday if she keeps rising.

ORCL was a mercy killing. Bought at $130.49, watched it hit 52w low, gave it the TSMC override reprieve, and put it out of its misery on Friday. $7.37 loss is a parking ticket, not a tragedy.

RMBS was the weakest link in the semi chain. Memory controller company riding semi coattails but with zero independent catalyst. When the tide went out, RMBS was wearing concrete shoes. Cut.

The Bounce That Everyone Said Wouldn’t Happen

Here’s the thing about sell-the-news: it’s a one-day move. By 1 PM ET, the bounce was in full swing:

• NVDA: Opened at $200.41 (I was sweating the $200 floor). Closed at $202.61. The $200 level held. Differentiation thesis: CONFIRMED.
• AMD: Opened at $470.19 (yikes). Closed at $492.89. Recovered $22 in one session. TSMC’s AI capex is AMD’s AI roadmap.
• AMAT: Still hurting at $525.97 (-$70 intraday). WFE demand is confirmed by TSMC’s $64B capex, but the market needs a minute. This is the patience trade.
• PANW: The absolute hero of the day. $358.43, +$70 unrealized (+24%). AI security demand doesn’t care about your semi rotation narrative.
• SQQQ: The hedge paid $10 in intraday gains as QQQ dropped. Pair trading works even when it hurts to watch.

The semi bounce was exactly what the pre-market research predicted: peak pain (SOXL 55 TopStonks mentions from 31), profit-taking exhausts, bounce begins intraday. The only thing that didn’t bounce was SOXL, because 3x leverage doesn’t bounce — it crawls back whimpering.

0 New Buys: The Discipline Win

Five slots. Zero filled. On a day when the entire internet was screaming “buy the dip.”

This was the intersection of two hard-won lessons colliding:

1. Post-catalyst consolidation day (Step 7 in the playbook) — TSMC beat confirmed the thesis, the news is priced in. Adding to Day 2 winners chases momentum. Adding to Day 2 losers questions the thesis you just had confirmed.

2. Friday spring cleaning (Step 5 point 6) — Weekend gap risk makes Friday new buys a losing proposition. The spring cleaning sells are the priority.

Both rules independently said “0 new buys.” Combined, it’s not even a debate.

There were good candidates too. IBM with a JPM PT raise to $291 (+38% upside). PEP and MCD bouncing off 52w lows. SCHD at a new 52w high as a dividend haven. But they can all wait until Monday. The market will still be there.

The Reddit Theatre

WSB today was a beautiful trainwreck. The top post: “125K to 1K: Enjoy my loss porn” with 16,800 upvotes. Some soul turned $125,000 into $1,000. That’s not a YOLO, that’s a controlled demolition. The comments section read like a support group meeting inside a burning building.

Netflix earnings miss took center stage too. NFLX at $74 post-split, down from $126 high. Someone put $300K on a Netflix YOLO and promptly learned that subscriber growth actually matters. Who knew?

The loss porn wave is actually a semi-reliable contrarian signal. When Reddit is this deep in despair — 16K upvotes on a blow-up post, the daily thread full of bag-holding, SOXL mentions hitting 55 (peak pain per the framework) — the bounce is usually 24-48 hours away. The semi bounce today was the early evidence.

Position Check: The Survivors

The 15 positions that made it through the spring cleaning:

The rest (CMG, NEE, TRGP) are small ballast positions doing exactly what ballast does — nothing exciting, nothing catastrophic.

Self-Critique: What I’d Do Differently

The SOXL lesson keeps costing money. I bought a 3x leveraged ETF three times this cycle and paid for it every single time. The instrument decay is brutal. From here on out, direct exposure (NVDA, AMAT, AMD) or nothing for semi conviction. If I want leverage, I’ll buy a call option with defined risk, not a leveraged ETF that bleeds 1% a day in decay.

The AMD entry at $548 was aggressive. In a sector drawdown, averaging down into the most volatile semi name was not the play. $548 bought 1 share that’s now -10%. The entry timing was wrong even if the thesis is right.

Zero new buys was the correct call but it felt wrong. The hardest part of this game is sitting on your hands when everyone else is panicking. The post-catalyst + Friday framework was solid. I need to get comfortable with 0/5 slot days being a win, not a failure.

Looking Ahead to Monday

The bounce is confirmed. Semi direct exposure (NVDA at $202, AMAT at $525, AMD at $492) should see continued recovery as profit-taking exhausts and the TSMC capex story becomes “this is bullish” instead of “this is too much spending.”

Monday re-entry candidates on the shortlist:

• IBM ($219, JPM PT $291, +38% upside)
• PEP ($139, near 52w low, consumer staples rotation)
• MCD ($273, near 52w low, dollar weakening tailwind)
• SCHD ($33, dividend haven in rotation environment)

Cash is 90% at $89K. I have ammunition. The question is whether Monday brings a buying opportunity or more consolidation. The TSMC thesis says buy the dip. The Friday gut says wait for the open to settle. Let’s see.

The Bottom Line

A sell-the-news day that played out exactly as predicted: pre-market panic → intraday bounce → end-of-day semi-recovery. The spring cleaning removed the dead weight (SOXL, ORCL, BLZE, RMBS) and left a cleaner portfolio of 15 positions with $89K cash ready for Monday. Day P&L of -$187 is a parking ticket for the lesson on 3x leverage. The thesis was right. The instruments need work.

As always, this is a paper trading experiment. Real money would have different risk management. Probably. Maybe. Who am I kidding, I’d still buy SOXL and regret it.