Insights

Nvidia's Aug 26 print — $96.2B revenue, +106% — was traded as the macro event of the week. Concentration measured with real data: a 44-point cap-weight vs equal-weight gap since 2023, big index days 2.4× likelier after Nvidia reports, and the 69–89% drawdowns that followed when past eras crowned IBM, Cisco and Intel.

On Aug 25, 2026 all four strategies read the same risk-on regime score — 28.8 — and sealed books from 28.7% to 81.9% deployed, with the aggressive book suddenly the most defensive. Real sealed data on where the disagreement is designed, where it is earned stock by stock, and why we publish the gap.

SK Hynix approved 54.3 trillion won ($38.1B) for two new memory fabs on Aug 7, 2026. Neither ships a wafer before 2028. Charted with real Micron price history: two prior memory capex supercycles, both ended in drawdowns over −48% — and the open question inside every AI memory stock today.

The S&P 500 closed at a record 7,757.64 on Aug 7, 2026 — the same week nonfarm payrolls printed −23,000. Real FRED and Tiingo data on two precedents, 1995 and 2007, where the same setup looked identical for months before the paths split.

A 1% annual fee sounds trivial. Run $10,000 through 30 years of compounding at 10% versus 9% and it quietly erases $41,817 — about a quarter of the final balance. The real math: a fee isn't a percent of your money, it's roughly a tenth of your return, and time makes that gap explode.

Chips sold $791.7B in 2025 — yet memory spent July in a bear market while AI accelerators printed record quarters. Our first industry deep dive maps the five businesses hiding inside "semiconductors," then opens the sealed record: a chip sleeve swinging 0% → 8.7% in 48 hours, this morning's scores for all 11 chip names, and the honest ledger of what July's chop cost.

Gold never moves our regime dial — it's ballast, not a signal. How four strategies size their gold sleeve, the execution leak our own sealed record exposed (14 of 18 July days at 0%), and the structural-hold fix.

Nineteen earnings reports from one megacap, charted with real data: reactions from −26.4% to +23.3%, ten up and nine down — a miss that soared, beats that sank. The one lever you actually control.

It sits behind your mortgage, your stocks, and every valuation on Wall Street. A plain-language explainer on the 10-year Treasury yield — the market's discount rate — and why its move from 1% to 4.7% quietly re-rated everything.

Someone always has a signal that marked every top since 1987. A 100% hit rate isn't the flex it looks like — it's the biggest tell that a backtest is fitting the past, not predicting the future. How we guard against it — including the upgrade we just killed.

Everyone fears a high VIX. Across 8,400+ trading days since 1993, higher fear has meant a higher median forward return — but a wider range in both directions. Why our engine weights the VIX continuously instead of drawing a line in the sand.

He ran Fidelity's Magellan at ~29% a year for 13 years — and still warned that more money is lost preparing for corrections than in the corrections themselves. A thought experiment on the quiet cost of waiting for the dip.

A routine BOJ rate hike detonated the yen carry trade. On Aug 5, 2024 the Nikkei fell 12.4% — worst since 1987 — the VIX hit 65.73 intraday, the S&P dropped 3%. Eleven sessions later it was all back. Charted with real USD/JPY and S&P 500 data — plus where the yen stands in 2026, back near multi-decade lows with the BOJ at 1%.

A U.S.-only view starts the day blind to the biggest thing that happened overnight. How our engine reads the just-closed Asian and European sessions — from the news wire, not a day-old proxy.

The Nasdaq peaked at 5,048.62 in March 2000 and fell about 77% by October 2002 — roughly $5 trillion erased, and 15 years to recover. A history lesson on how a bull market hides the difference between skill and a rising tide.

You'll see it next to every return on our site: α. Alpha — the part of a return the market doesn't explain — charted with real data: the same 500 S&P stocks, cap-weighted vs equal-weighted, diverged ~30 points in five years, and the sign flipped along the way. Why a rising tide isn't skill, and why we lead with α.

When the engine shows a stock a 75, what does that mean? A no-black-box breakdown of the AlphaScore — the four 1–10 reads (technical, sentiment, macro, catalyst) that average into every number.

The Dow fell 22.6% in a single day — no bank failed, no headline explained it. Portfolio insurance made the machines sell into the selling. A history lesson on crowded strategies, market structure, and why circuit breakers exist.

Howard Marks built a career on one idea: you can't predict, you can prepare. A thought experiment on the pendulum of greed and fear, risk as permanent loss, and why "where are we now?" beats "where are we going?"

Same 55% edge: bet 10% (Kelly) and grow 3.5×, bet 20% and gain nothing, bet 30% and lose almost everything — charted with math and a virtual simulation, plus the sizing recipe our strategies publish daily.

Korea fell 7.9% — and our sealed record said "Asia +0.35%." The look-ahead-bias rule that made the engine read Friday on a Monday, the same-day disclosure, the overnight fix, and why the miss barely mattered.

Korea plunged 7.9% on Monday; U.S. stocks rebounded Tuesday. A plain-language explainer on market regimes — the posture-control tool that decides how much to risk before what to buy.

Korea's market fell 7.9% today. A crash-day thought experiment built on Munger's public principles — equanimity through 50% declines, inversion, temperament, and the discipline of waiting.

Korea's KOSPI fell 7.9% in a single session on July 13, 2026. Three Asian crashes charted through one U.S. lens — real S&P 500 data: −7.2% (1997), −4.1% (2015), −2.9% (2024) the next day, higher 60 sessions later all three times — and why semiconductor-driven contagion behaves differently.