Article published Jun 26, 2026. Prices below use latest available snapshots.
The market isn't falling — it's rotating, and the rotation is sorting the mega-caps into two piles that look identical on a "it's down a lot" screen but trade like opposites. Here's the buy list, why MSFT and GOOGL are not the same trade, and whether the "midterm years are choppy" stat making the rounds is actually true. · ~5 min read
Every price/RSI/trend/return below is code-computed from validated daily summaries (summaries), last settled close 2026-06-25 — price truth. Today's session (06-26) is still open; intraday not quoted. Companion to the Rotation Map and the GOOGL deep-dive.
What's going on (it's a rotation, and breadth proves it)
The index tape looks calm — SPY −3.4% off its high (RSI 48), QQQ −4.5% (RSI 50). Underneath, it's violent. Money stampeded out of the crowded mega-cap-growth + enterprise-software trade and into value, small-caps, healthcare, and the physical AI build-out. The proof is in the breadth: while the cap-weighted indices sag, the equal-weight and value averages are at new highs —
| Breadth gauge | RSI | trend | off 52wk high |
|---|---|---|---|
| RSP (equal-weight S&P) | 62 | strong-up | −1.2% |
| VTV (value) | 69 | strong-up | −0.8% |
| IWM (small-cap) | 62 | strong-up | −1.0% |
When equal-weight is at a high and cap-weight is down, the "red" on your screen is five crowded stocks unwinding, not the market breaking. That's a healthy rotation, not a crash.
The one framework that matters: golden cross intact, or not?
In a rotation, "it's down 30%" tells you nothing. The only question that sorts winners from traps is whether the long-term trend is still intact — a rising 200-day with a golden cross (buy the pullback) versus a falling 200-day with a death cross ("oversold can stay oversold"). The mega-caps split cleanly on exactly that line:
| Name | price | RSI | 200-day | 3-mo | off high | verdict |
|---|---|---|---|---|---|---|
| GOOGL | $340.85 | 35 | golden cross, +9% above | +21% | −17% | ✅ clean dip |
| AAPL | $278.00 | 34 | golden cross, +3% above | +10% | −12% | ✅ clean dip |
| AMZN | $228.16 | 34 | golden cross, ~flat to 200d | +10% | −18% | ✅ borderline dip |
| AVGO | $380.52 | 44 | golden cross, +6% above | +23% | −23% | ✅ semi-leader dip |
| MSFT | $350.10 | 28 | death cross, −22% below | −4% | −37% | ⛔ broken |
| META | $543.50 | 35 | death cross, −16% below | −1% | −32% | ⛔ broken |
The MSFT-vs-GOOGL tell
You said "Microsoft? Google? both look wonderful." They look the same on a "it's cheap now" chart — and they are opposites.
- GOOGL is the cleanest dip on the board. It's still up 21% over three months, sits +9% above a rising 200-day, and the golden cross is intact. The RSI-35 weakness is a pullback within an uptrend — the textbook buy-the-reclaim. The 06-25 deep-dive put the trigger at an SMA20 reclaim (~$363); that's the entry, not the knife.
- MSFT only looks wonderful because it's fallen a long way. It's −37% off its high, −22% below a falling 200-day, death-crossed, and down over three months. That's not a dip in an uptrend — it's a broken trend, in the same family as ORCL/NOW/PLTR, just less extreme. RSI 28 isn't "cheap," it's "oversold can stay oversold." Microsoft's been repriced harder than its peers on the AI-capex / free-cash-flow-compression question, and nothing in the trend has turned yet.
The rule: cheap is not an entry; a reclaimed trend is. GOOGL qualifies today; MSFT is a wait-for-a-base name. Buy the reclaim, not the drawdown.
The buy list (trend-hold framing)
Tier 1 — quality dips with the trend intact (buy the reclaim):
- GOOGL — the standout. Entry on an SMA20 reclaim (~$363); golden cross intact, +21%/3mo.
- AVGO — a genuine semi leader that pulled back (RSI 44, +23%/3mo, golden cross) rather than broke — the rare way to own the AI-silicon trend without chasing a vertical.
- CRWD — the one corner of software that works: strong-up, +72%/3mo, golden cross, and it's pulled back to its 20-day (−2.6%) — a flag, not a fall.
- AAPL / AMZN — shallower, cleaner pullbacks (golden cross intact); lower-conviction than GOOGL but same setup.
Tier 2 — the real leadership, but extended (wait for the pullback, don't chase):
- Memory / semicap — MU (+228%/3mo), SNDK (+276%), ALAB (+255%), AMAT (RSI 71), AMD (+159%). This is the trend in the market and it didn't break — but it's near highs with RSI 60-71. The entry is a flag back to the 20-day, not the breakout chase. (The memory-supercycle thesis + SemiAnalysis "Memory Mania" say the shortage is structural — so pullbacks are for buying, not fading.)
- Healthcare — UNH/ABBV/ILMN/CVS held green but are now RSI 67-72 (overbought). They did their job in the selloff; chasing them up here isn't a fresh entry.
Tier 3 — the traps (do not buy the bounce):
- NOW (−57% off high, death cross), ORCL (−56%), ADBE (−50%), PLTR (RSI 27, −49%), NFLX (RSI 20, −47%), CRM (−46%), ZS (−63%). All death-crossed and deeply broken. Low RSI ≠ cheap. A broken trend bottoms by basing, not bouncing — wait for the base, then the reclaim.
The midterm question — is the "choppy stocks" stat true?
Short version: the pattern is real and well-documented; the precise stats floating around Twitter are usually cherry-picked. Here's the honest read.
2026 is a US midterm year (general election ~November). In the four-year Presidential Cycle, the midterm year (year 2) has historically been the weakest and most volatile of the four — the largest average intra-year drawdown of the cycle. The market has tended to bottom in Q2–Q3 of the midterm year, and the stretch from that midterm low into the following pre-election year has historically been one of the strongest windows of the entire cycle. That much is robust — it shows up across decades of data (the seasonality literature, Stock Trader's Almanac, presidential-cycle studies).
The caveats that matter:
- Small sample. ~19–20 midterm years since WWII. A tendency, not a law — plenty of individual exceptions.
- The viral "X% average / never down" stats are overfit. The exact figure depends entirely on the start date, index, and whether you measure low-to-high or calendar years. Treat any precise number on Twitter as marketing until verified.
- This chop has a real cause. 2026's volatility is being driven by the AI-capex rotation — a genuine repricing of who earns the return on hundreds of billions of capex — not by the election calendar. The seasonality is, at most, a coincident tailwind, not the engine.
The actionable read: if the midterm pattern holds, the typical Q3-midterm bottom → Q4-into-2027 rally would line up with accumulating quality dips now rather than waiting. But that's a reason to not fear the chop, not a reason to buy the calendar. Buy the setup (a reclaimed trend); let the seasonality be a mild supportive backdrop. Don't trade the election; trade the chart.
(Want hard numbers instead of the well-known pattern? I can run an actual presidential-cycle backtest on our data and report the real midterm-year drawdown/forward-return stats — say the word.)
The takeaway
- It's a rotation, and breadth is healthy — equal-weight/value/small-cap at highs while five crowded mega-caps unwind.
- GOOGL is the cleanest buy on the board (golden cross intact, +21%/3mo) — entry on the SMA20 reclaim. MSFT is not the same trade — it's a broken trend (death cross, −37% off high), a wait-for-a-base name. Cheap ≠ entry.
- The memory/semicap leadership is the real trend but extended — buy its pullbacks (MU/AVGO/AMAT flags), don't chase the vertical.
- Avoid the broken SaaS (NOW/ORCL/PLTR/NFLX) until they base — low RSI is a trap, not a discount.
- The midterm "choppy" stat is directionally true but oversold on Twitter — a real tendency with small-n caveats. Buy the setup, not the calendar.
Sources
Price, RSI and trend figures read from the desk's validated daily scan summaries (summaries). No number in this note was computed in prose.