Markets have not become more efficient over time; in fact, the spectacular inefficiencies and bubbles in meme stocks—rising 6x in a year—may be worse now than historically.

The market currently exhibits one of the two or three highest price levels in history despite facing multiple severe headwinds: population bust in Japan, South Korea, and China; accelerating climate damage costing roughly 0.5% of global GDP annually with worsening trends; geopolitical instability with Russia and China simultaneously; active wars in multiple locations; and trade war escalation.

Throughout history, high PE markets have never predicted higher future profits, growth, or productivity; instead, they consistently predict tough times, with the pattern evident across the Great Depression (1929), Japan's Lost Decades (1989 65x PE), and the 2000 crash.

Humans are natural optimists and wishful thinkers by evolutionary design (pessimism was not a good survival trait for 150,000 years), and stock markets reflect this bias: when bad economic data appears, the market cheers for potential Fed rate cuts; when good growth appears, it cheers for profits—always finding an excuse to be bullish and over-explain good news.

In 2022, after Mag 7 had collapsed, ChatGPT's launch in October 2022 reversed the pattern; AI investment CapEx and animal spirits were so powerful that the Mag 7 rose so much (up 40%+ while S&P fell 25%) that they dragged the S&P up, forestalling what would have been a 40%+ broader market decline over 10 months.

Oil price spikes have caused recessions without exception throughout history; oil is a critical fuel for which demand cannot be easily substituted, and the recent 50% oil price increase from the Iran war will create balancing effects (weak markets, reduced demand) and is clearly painful for the economy.

Grantham identified the bottom of the 2008-2009 financial crisis to within a day or week in March 2009 (the S&P 500 closed at 666), recognizing that valuations were priced to deliver ~12% real returns over 7 years, making it the highest price on their 7-year forecast in 22 years—cheap enough to call the bottom despite being higher than 1974 or 1982 lows.

The market does not turn at tops or bottoms when signs of hope appear; it turns when conditions look bleakest but are slightly less bleak than the day before—a counterintuitive pattern that explains why almost nobody picks market extremes.

Eisenhower opened his farewell address by thanking both political parties for their 'constant and considerable cooperation,' setting a tone of bipartisan unity that would be unimaginable in modern politics.

Grantham's key to filtering bad ideas quickly was working with Chris Donnell, who could identify fatal flaws in ideas within 20 seconds, allowing a rapid iteration through 10-20 ideas to find one worth serious research; this combination of prolific idea generation plus rapid critique is exceptionally powerful.

The S&P 500's dividend yield at the peak of the 1999 dot-com bubble was an extraordinarily low 1.6%, a level never seen even in 1929, highlighting the extreme valuation compression during the tech bubble.

Value investors should not stand their ground in badly overpriced stock markets unless they want to eventually take losses, and while they will underperform in the short run when others gain from overpriced assets, they will win in the long run by owning cheaper stocks.

The dividend discount model, which measures the fair value ratio of stocks (e.g., 0.79 = 21% cheap, 1.12 = 12% overpriced), is a measuring kit to test investment instincts by aggregating valuations across market segments, revealing that small-cap stocks were very cheap while large-cap stocks were on average very expensive.

Momentum is a simple-minded but powerful inefficiency that has worked consistently throughout Grantham's career and before: a body in motion tends to stay in motion for a while, which was provably true even 20-30 years before academics claimed markets were efficient.

AI will unbalance supply and demand in unprecedented ways: it may replace workers, reduce demand for human labor and consumption, or drive productivity gains with nobody left to consume the output; supply-demand imbalances in commodities (copper, natural gas) cause price extremes, and AI's effect on human employment and consumption is 'infinitely complicated' and will generate new problems regardless of benefits.

Grantham has not yet gone back to analyze current market data in detail to fully validate whether the late 2024/early 2025 period truly matches the high-beta leadership divergence pattern that preceded previous crashes, acknowledging this is a complicated issue that requires data review.

Large financial corporations (Goldman Sachs, JP Morgan, Morgan Stanley) will never publicly warn investors to exit overpriced markets because their business model requires constant optimism; therefore, if no serious financial institution is warning of a crash, this is not evidence the market is fairly priced—it simply reflects institutional incentives to always say everything is fine.

Grantham's 'butterfly' thinking style—letting ideas flit around briefly before returning to them—prevents mental staleness and enhances creativity; hammering earnestly on a single topic is seen as a virtue but actually blocks creative insights.

Quality as a factor—measured by less debt, higher returns, more stability, and lower bankruptcy risk—is a genuine market inefficiency: quality stocks outperformed by roughly 0.5% per year while bonds show AAA bonds underperform B bonds by ~1% per year, creating a ~1.5% annual 'freebie' inefficiency that academics missed for decades.

Grantham's investment philosophy is not to predict when bubbles will occur, but to recognize them when they have clearly arrived and identify the cheap alternatives available to hide in when they burst.