Eighteen macro indicators in four chapters — Fiat Erosion, The Debt Machine, Late-Cycle Signals, and Valuations & The Golden Mirror — each one a piece of the same story.
Is the stock market cheap, fair or expensive — and over which horizon? Every equity-linked indicator in the four chapters is ranked against its own history (100 = most expensive or most crowded), then averaged by the horizon it speaks to: long-term valuation, the medium-term cycle, and short-term positioning. A temperature is a regime reading, not a timing signal — the long view can run hot for years while positioning stays cool.
Is money holding its value? A century of purchasing-power decay across two major currencies, the one currency that resisted — and the monetary engine driving it all.
The price and burden of money: sovereign leverage beyond its wartime peak, real rates that punished savers for a decade, and the bond market’s most reliable recession alarm.
How investors behave near tops: leverage accelerating, mood diverging from price, the labor cycle turning — or refusing to — the bond market’s second opinion on every equity rally, and the compensation for holding equities over bonds quietly vanishing.
What you pay determines what you get. The most-watched valuation gauges in history, the price of a dollar of revenue decomposed into its engines, the index re-priced in hard money — and the asset that reflects back what real yields say about paper money.
Is money holding its value? A century of purchasing-power decay across two major currencies, the one currency that resisted — and the monetary engine driving it all.
What $1.00 in 1913 buys today — a 98% loss, equivalent to +4,719% cumulative inflation over 113 years, or ~3.5% annualised. The first 58 years under Bretton Woods ran at ~2.5%/yr; the 55 years since Nixon closed the gold window in 1971 have averaged ~4.6%/yr. Since 2020 alone the annualised rate hit ~10%.
Since its 1999 launch the euro has lost 57% of its purchasing power — a cumulative inflation of +134%, or ~3.1% annualised. The first 22 years were relatively contained at ~1.7% per year. The last five have been brutal: a 23% loss since 2021 alone, equivalent to ~5.3% annualised, as post-COVID supply shocks and the energy crisis did in half a decade what the previous two decades had barely managed.
USD vs EUR — side by side
Over comparable periods, the USD has lost more in absolute terms (96% since 1913) but the EUR has devalued faster on an annualised basis: 2.2%/yr vs 3.1%/yr for USD since 1971. The 2021–2023 inflationary episode hit both similarly hard — the EUR lost 16% of purchasing power in just 3 years, versus 14% for the USD over the same window.
The Swiss franc has been one of the strongest secular appreciators against the dollar over the floating-rate era — from 4.32 CHF per dollar at Bretton Woods' end to under 0.80 today, a cumulative USD loss of ~80% or ~2.9%/yr annualised. Key breaks: the 1985 Plaza Accord reversal, the 2011 SNB EUR floor and its 2015 abandonment (the largest single-day move on record for a major currency), and the sustained 2025 dollar slide.
The mechanical engine behind dollar devaluation. $6.3 trillion was created in roughly 24 months during COVID — more than was created in the entire prior century. Milton Friedman's "inflation is always and everywhere a monetary phenomenon" plays out literally here.
The price and burden of money: sovereign leverage beyond its wartime peak, real rates that punished savers for a decade, and the bond market’s most reliable recession alarm.
The US has now exceeded its WW2 debt peak — and unlike 1946, there is no obvious deleveraging path. Post-war growth, financial repression, and moderate inflation drove the prior decline. That playbook looks far harder to execute today.
Fed Funds Rate minus CPI inflation — the true cost of money. Negative real rates are a hidden tax on savers and a subsidy for debtors. They also tend to inflate asset prices. The 2020–2022 period produced the most deeply negative real rates since the 1970s.
The bond market's most reliable recession oracle — 6 of 7 distinct inversion episodes since 1976 preceded a recession. The notable exception: the 2022–2024 inversion, the deepest since 1981, has not (yet) triggered one — challenging the signal's historical record.
How investors behave near tops: leverage accelerating, mood diverging from price, the labor cycle turning — or refusing to — the bond market’s second opinion on every equity rally, and the compensation for holding equities over bonds quietly vanishing.
FINRA margin debt — money borrowed by investors to buy securities — has hit a record $1.42 trillion, growing 54% year-over-year. The S&P 500 overlay reveals the feedback loop: margin peaks tend to coincide with or slightly precede market peaks, and crashes are amplified by forced selling as margin calls cascade.
The feedback loop
When prices fall, brokers issue margin calls → investors are forced to sell → prices fall further → more margin calls. This self-reinforcing spiral amplified the 2000, 2008, and 2022 crashes. With YoY growth running at +45% in 2025 — matching the velocity seen before the 2000 and 2021 peaks — the potential energy stored in this mechanism is at historically dangerous levels.
The oscillator above asks how fast leverage builds; this asks how large it is against the economy. At 4.40% of GDP (May 2026), margin debt is 16% above the October 2021 record (3.78%), far beyond dot-com (2.95%) and 2007 (2.86%), and 1.9× the 29-year mean. The two currently disagree instructively — momentum merely elevated, level unprecedented: record stock of leverage on moderating flow is a classic late-cycle configuration. Historically the ratio’s rollover from an extreme led S&P monthly-close tops by ~5 months (2000), ~3 (2007), ~2–3 (2021). It has not rolled over yet.
University of Michigan Consumer Sentiment Index plotted against the S&P 500, both normalised to Z-scores (standard deviations from their 48-year means). The shared axis makes divergences directly legible. Since ~68% of US GDP is consumer-driven, persistent gaps between market valuations and consumer confidence tend to resolve by the market falling toward sentiment rather than the reverse. The current spread of +6.2σ is the widest on record — blown out further by May 2026’s all-time-low sentiment print of 44.8.
The chapter’s only real-economy gauge. Employment lags the economy but not the market: unemployment’s cycle lows mark maximum optimism and sit beside major tops (3.8% in 2000, 4.4% in 2007, 3.5% into 2020), while its peaks mark generational bottoms. The trigger version — the Sahm rule, the cleanest single-series recession detector in macro — fired in all eleven recessions from 1950–2020 with one 1959 false alarm. This cycle broke the streak: it fired at 0.53 in July 2024 with no recession, feinted at 0.47 in November 2025, and has receded to 0.10 as of June 2026 with unemployment back at 4.2% — either a soft landing achieved, or the longest topping process on record.
The whipsaw is the finding
A signal that never failed for seventy years has now fired falsely once and feinted twice in a single cycle — post-COVID labor-supply swings, immigration-driven labor-force growth, and the late-2025 shutdown all made the unemployment rate a noisier thermometer than usual. The contrarian level still says late-cycle (4.2% is tight territory), but the trigger has fully receded while leverage (08) sits at all-time extremes. When these disagree, history’s tiebreak: leverage extremes date the fragility, the labor turn dates the break.
Total put volume divided by call volume in single-stock options on Cboe — the classic contrarian gauge of speculative positioning. Where the Michigan survey asks how the crowd feels, the options tape shows how it is positioned: heavy put buying marks fear, historically a 3–6 month buying setup, while a call-heavy tape marks complacency that has often preceded weak stretches. Smoothed with a 10-day average; zone thresholds ≈ the 15th/85th percentiles of the sample (0.55 / 0.73). The current 10-day average and its one-year percentile are shown live in the readout above the chart. The Equity / Index / Total toggle switches the ratio universe between single-stock options, index options, and all Cboe products combined — same source, same method; zone thresholds are the 15th/85th percentiles of each series’ own sample.
The extra yield investors demand for holding Baa-rated corporate bonds — the lowest investment-grade rung, the market’s price for lending to ordinary corporate America — over 10-year Treasuries is the bond market’s running verdict on downside risk: a senior claim with capped upside, priced by institutions that get paid only for what does not default. Plotted inverted against the S&P 500, credit and equity should trace the same path. When equity advances and credit refuses to confirm, history has usually sided with credit — spreads were already widening through the second half of 2007, months before the S&P’s October peak, and reached their record wides as equities collapsed through late 2008. The CECS score below turns that divergence into a single line — equity momentum minus credit momentum, in standard deviations.
Baa − 10Y Treasury — long-history backbone
–The investment-grade anchor — Moody’s Baa yield over the 10-year Treasury, with history back to the 1980s. Slower and less noisy than CCC, so it is read on a 5-year percentile and the full-history range below.
CCC & Lower — short-horizon stress gauge
awaiting dataThe distressed tail of the junk market — the most sensitive early-warning spread, and the noisiest. FRED caps this ICE series to three years, so it is read only on a 3-year percentile and a 3-month change; the long-history backbone above stays Baa. In the Equity Market Temperature it scores conditionally: tight or easing spreads read contrarian (tight = complacent), while wide and still widening spreads (3-month change up to +1.0 pp and beyond) count as stress, i.e. risk.
How to read the score
CECS compares three-month momentum in the S&P 500 against three-month momentum in the Baa credit spread, each normalised to its own three-year history. Near zero, the two markets agree. Persistently positive readings mean equity is climbing on conviction the bond market does not share — the classic late-cycle non-confirmation. Deeply negative readings usually appear after capitulation, when credit repairs before equity dares to — historically the better risk-taking window. Two honest caveats. Spreads are sensitive but not specific: they reprice in every scare, including several (2011, 2015–16, 2022) that never became recessions — treat widening as risk being repriced, not as a recession verdict. And Baa is investment-grade, so it moves less violently than junk-rated spreads — which cuts both ways: fewer false alarms, and a persistent move is harder to dismiss.
The earnings yield of the S&P 500 (1 ÷ trailing P/E) minus the 10-year Treasury yield. Measures what extra return stocks offer over risk-free bonds. The ERP has turned negative in 2023–2026 — the first sustained negative reading since the dot-com bubble. This means Treasuries are currently yielding more than the S&P 500's earnings yield, a rare condition that historically signals expensive equities.
Also known as — the Fed Model
This earnings-yield-versus-10-year-Treasury comparison is the so-called “Fed Stock Market Valuation Model” — a name coined by economist Dr. Ed Yardeni (then of Deutsche Morgan Grenfell) in 1997, after the Federal Reserve’s July 1997 Humphrey–Hawkins report flagged the same relationship. The Fed itself never officially endorsed it.
The ERP and the free money era
From 2009 to 2021, near-zero rates made the ERP deeply positive — stocks were the only game in town. That was financial repression working as intended. Now, with the 10-year at 4.4%, investors can earn real returns from bonds for the first time in 15 years. The negative ERP is not predicting a crash — but it removes one of the pillars that justified high equity valuations throughout the 2010s.
What you pay determines what you get. The most-watched valuation gauges in history, the price of a dollar of revenue decomposed into its engines, the index re-priced in hard money — and the asset that reflects back what real yields say about paper money.
The cyclically adjusted P/E — stock prices divided by 10-year average inflation-adjusted earnings — is arguably the gold standard of long-run valuation. At 41.6, it sits at the 2nd highest level in 154 years, exceeded only by the December 1999 dot-com peak of 44.2. Historical evidence: CAPE above 30 has preceded every major bear market. The long-run median is 16.
The caveat
CAPE has been "elevated" since the mid-1990s. Structural changes — intangible capital, tech dominance, globalised profits — may justify a higher baseline. The signal matters most at the rate of change: CAPE at 38 after 3 years of rapid expansion is different from CAPE at 38 after 10 years of stability.
Every equity return decomposes as price = sales × margin × multiple, and sales are the honest layer — slow, GDP-anchored, nearly unmanipulable. Since December 2000 the index’s price has compounded at +7.0%/yr on revenues doing +4.0%; the wedge is margin expansion (GAAP net margins 6.4% → 11.5%) plus multiple gain. The resulting price-to-sales ratio of 3.67 is an all-time record, more than double the 1.64 median. And the engine mix has just rotated: the 2013–21 bull ran on margins; the advance since late 2022 runs on the multiple (+9.6%/yr of P/E expansion) — the least durable engine of the three, now the largest.
Three-engine attribution (%/yr, log-additive)
Dec 2000 → Jun 2026: price +7.0 = sales +4.0 + margin +2.3 + P/E +0.6The double bet
Sales contribute the same steady +4–5% in every era; everything above that rests on the two layers history mean-reverts. At 3.67× sales on ~11.5% margins, the index compounds two bets at once — that record margins persist and that a P/E above 30 on those margins persists. Reversion of P/S merely to 2.5 over a decade costs ~3.9 points of annual return; to the median, ~7.7. CAPE (13) prices the earnings; this panel shows what the earnings are made of.
Total US stock market capitalisation (Wilshire 5000) divided by GDP. Warren Buffett called it "probably the best single measure of where valuations stand at any given moment." At 232%, it is near its all-time high. The structural uptrend since 1995 means the raw ratio should be compared to its trendline rather than its historical mean — on a de-trended basis, the market is approximately 2.0–2.4 standard deviations above trend.
The structural uptrend caveat
The ratio has trended upward since 1995, possibly because US multinationals earn profits globally that aren't captured by domestic GDP, and because technology companies carry high market values relative to their asset base. De-trended, the current reading is still approximately 2.0–2.4 standard deviations above the regression trendline — firmly in overvalued territory.
The index priced in hard money: how many ounces of gold buy the S&P 500. It strips the monetary-debasement component out of equity returns and turns nominal price history into three enormous secular waves — peaks in 1967 (2.74) and 2000 (5.41), troughs in 1980 (0.16) and 2011 (0.66). Since the December 2021 peak of 2.61 the S&P has advanced roughly 60% in nominal terms while losing 30% priced in gold — a nominal advance unconfirmed in hard money, the same signature the 2003–07 cycle carried. The current reading is 1.81, the 63rd percentile of the free-float era.
The unconfirmed advance
Every nominal all-time high the S&P has made since early 2024 has gone unconfirmed in gold terms — the ratio peaked in December 2021 and fell as much as 47% into March 2026 while headlines celebrated record closes. The 2003–07 bull market carried the identical signature: new nominal highs, no new high priced in gold, an advance driven by liquidity rather than real earnings power. When the CAPE and Buffett Indicator say equities are expensive in dollars, this ratio adds that the dollars themselves have been shrinking against hard money.
Extended back to 1971 using a proxy real yield (10Y nominal Treasury minus trailing CPI) for 1971–2003, switching to actual TIPS market yields from 2004 onward. The proxy era is shown dashed; the TIPS era solid. Three defining episodes now visible: the 1974–1980 stagflation surge, the Volcker shock collapse, and the 2022–2026 breakdown of the inverse relationship.
The 2022–2026 anomaly
The classical model says gold at $4,125/oz — even after a 14% Q2-2026 crack, its worst quarter since 2013, off the February peak near $5,020 — with real yields at +2.25% makes no sense: you are being paid 2.25% above inflation to hold Treasuries instead. Yet gold has more than doubled since 2022 and is at all-time highs. The explanation: central bank buying has hit record levels (1,000+ tonnes/yr for three consecutive years), driven by de-dollarisation from China, Russia, India and Turkey. The relationship hasn't broken — a new structural buyer has entered who doesn't care about opportunity cost.