FLEXAI — LEARN
Hidden Signs
The economy whispers before it shouts. Below are the charts professionals watch for the first cracks — each one explained in plain words, with where it stands right now — and the reports and data libraries institutions study to learn how markets have actually behaved over the last 100+ years.
Snapshot: November 15, 2026 · Refreshed weekly · Every chart is drawn only from documented readings.
Part One — The Warning Dashboard
Think of the economy as a patient. Some charts are its temperature history, some are early-warning blood tests, one is a smoke alarm that only sounds once smoke is in the room, and some measure how jumpy the nurses are. Here is every gauge at a glance:
| Gauge | Latest | What it says |
|---|---|---|
| 10-year Treasury yield · Nov 13 | 5.15% | Back above 5% — expensive money again |
| Yield curve (10Y − 2Y) · Nov 13 | +0.27% | Normal shape — no inversion |
| Yield curve (10Y − 3M) · Nov 13 | +0.83% | Normal shape |
| NY Fed recession odds (12-mo) · from Oct spread | ≈ 12.4% | Near normal — was 70.9% at the peak |
| Sahm Rule (trigger 0.50) · Oct 2026 | 0.00 | Silent |
| Leading Economic Index · Sep 2026 | 99.6, flat | Stabilising after a long decline |
| ISM factory index (PMI) · Oct 2026 | 55.5 | Factories growing faster — costs still high |
| St. Louis stress index · wk of Nov 6 | −0.49 | Calmer than average |
| Kansas City stress index · Oct 2026 | −0.89 | Calm — a touch less calm than Sep |
| Chicago Fed conditions (NFCI) · wk of Nov 6 | −0.462 | Looser than average |
| High-yield credit spread · Nov 12 | 3.20% | Narrowed back — fear eased |
| Jobless claims (weekly) · wk ending Nov 7 | 200K | Still at/below 200K — 9th straight week |
| Housing starts · Sep 2026 | 1.221M | Rate-driven softening |
| CFO optimism survey · Q3 2026 | 60.3 / 100 | Steady — small firms strained |
The 10-Year Treasury Yield
FRED: DGS10What it means: The interest rate the US government pays to borrow money for 10 years. It is the world's most important number, because mortgages, company loans and the value of stocks are all priced off it. When it rises fast, borrowing gets expensive everywhere and everything wobbles.
Where it stands: It touched a 52-week high of 5.31% on Oct 5, eased back under 5% — 4.96% on Oct 22, 4.94% on Oct 30, 4.86% on Nov 6 — then jumped back above 5% in one week, to 5.15% on Nov 13 (U.S. Treasury daily par yield), its highest close since mid-October. The climb that came first: about 4.44% in late June, 4.68% in late August, roughly 5.26% (weekly average) in early October. In plain terms, the brief relief on borrowing costs is over for now.
The lesson: Yields at or above 5% were last normal before 2007; the 2020 low was 0.52%. History says the level alone has no magic trigger — what hurts is speed and cause. Rises driven by inflation fear (2022, and this episode) have been the damaging kind.
Weekly averages for the first three points, then daily readings (U.S. Treasury par yield). The Oct 5 daily print hit 5.31%.
Free source: FRED — DGS10 ↗
The Yield Curve — and the Recession Odds Model
FRED: T10Y2Y · T10Y3MWhat it means: Normally, lending money for 10 years pays more interest than lending it for 3 months or 2 years. When that flips — short-term pays more — it is called an “inversion,” and it means investors expect the central bank to cut rates later because the economy is weakening. It has been the most famous recession warning in finance.
Where it stands: Both curves keep their normal shape. The 10Y−2Y dipped to about +0.20 in late September, recovered toward +0.5 in early October, and sits at +0.27 on Nov 13 (10-year 5.15% minus 2-year 4.88%, U.S. Treasury par yields); the 10Y−3M sits at +0.83 (5.15% minus the 3-month bill's 4.32%). The New York Fed's model turns the 10Y−3M gap into recession odds: its November release, from the October spread, puts the odds at about 12.4% for the next 12 months — near a normal month's reading, and a touch lower than the prior month's 12.8%.
The lesson: The cautionary tale every beginner should know: inversions came before every US recession since the 1970s, and the model peaked at 70.9% after the deep 2022–24 inversion — yet no recession followed. That was the signal's first clear modern false positive. Learn the curve as “raises the odds,” never “schedules a recession.” Note: headlines use 10Y−2Y; the NY Fed model uses 10Y−3M. They are different charts.
10-year minus 2-year spread. Approximate late-Sep / early-Oct readings per FRED-based reporting.
Note: The ≈12.4% figure is the NY Fed model's November 2026 release, from the October 2026 spread (12.42%, as cited by secondary trackers of the NY Fed's monthly data file). Treat it as “about 12%.”
Free source: FRED — T10Y2Y ↗
The Sahm Rule — the Jobs Smoke Alarm
FRED: SAHMREALTIMEWhat it means: A smoke alarm for jobs. If unemployment has risen half a percentage point from its recent low (using 3-month averages), a recession had already begun every time that happened in the past. It was designed by economist Claudia Sahm to trigger early, so help can be sent fast — it says “a recession has likely started,” it does not forecast one.
Where it stands: Recent readings wobble around zero — May 0.10, Jun 0.07, Jul −0.03, Aug −0.07, Sep 0.00, Oct 0.00 — nowhere near the 0.50 trigger. October's jobs report (released Nov 6) is why it stayed at zero: unemployment actually ticked down to 4.1% (from 4.2%), and employers added 165,000 jobs — far more than the roughly 65,000 expected. (Negative values just mean unemployment is at or below its recent low.)
The lesson: The rule fired in mid-2024 as unemployment rose — and no recession followed, its first widely debated false positive; Sahm herself noted the post-pandemic job market was unusual. Today's reading says the jobs market is not in recession territory by this yardstick, but it sits closer to the line than a flat 0.00 suggests if unemployment keeps drifting up.
Free source: FRED — SAHMREALTIME ↗
The Leading Economic Index (LEI)
Conference Board LEIWhat it means: A single score built from 10 early-warning ingredients — factory hours, new orders, building permits, stock prices, credit conditions, what consumers expect, and more. When the score falls steadily for months, a recession has usually followed. Because it mixes 10 ingredients, no single weird number can fake the signal.
Where it stands: Flat in September after −0.1% in August. The 6-month change improved to +0.6% (Mar–Sep 2026), up from −1.1% in the prior six months — the direction is getting better, not worse. The October LEI is due Thu Nov 19, 2026.
The lesson: Steady LEI declines came before past recessions — but the LEI also fell for a long stretch in 2022–23 with no recession, the second member of this cycle's “famous signals that cried wolf” club. The recent flattening is why many economists call the current LEI message “caution lifted, growth slow” rather than alarm.
Six-month change in the index. Headline readings are free in the monthly release; the full database is a paid Conference Board product.
Free source: Conference Board — LEI release ↗
The Factory Index (ISM Manufacturing PMI)
ISM Report On BusinessWhat it means: Every month, factory purchasing managers are asked one simple question: are orders, production and hiring better or worse than last month? Above 50 means factories are growing; below 50 means shrinking. Subtler rule ISM teaches: for the whole economy, the line that matters over time is about 47.5, because services can grow while factories shrink.
Where it stands: A 10th straight month of factory expansion — and a faster one, up a full point from September's 54.5, after 10 months of contraction in late 2025. Inside the report: New Orders 55.5, Production 59.0 (the strongest since November 2021), Employment 52.4 — and the Prices index eased 2.7 points to 75.2, still very high. Factories are busy, and their costs are still surging, just a little less fast.
The lesson: Deep readings under 45 went with the 2008–09 and 2020 recessions; mild dips under 50 (2023, late 2025) repeatedly happened with no economy-wide recession. It is the first major reading on each new month and hard to fake. Today's warning inside this report is about prices, not activity.
Free source: ISM — Manufacturing Report ↗
The Financial Stress Indexes
FRED: STLFSI4 · KCFSI · NFCIWhat it means: One number that squeezes together interest rates, credit spreads and market jumpiness to answer: “How panicky is the financial system right now?” Zero means perfectly normal for history. Below zero = calmer than usual. Above zero = more stressed than usual. In the 2008 crisis, the St. Louis version peaked near +9.7.
Where it stands: All three sit below zero — calmer than average. St. Louis ticked up a little again (−0.52 → −0.49 in the week ending Nov 6) but stays firmly calm. Kansas City's monthly index reads −0.89 in October (Apr −0.72, May −0.88, Jun −0.76, Jul −0.84, Aug −0.95, Sep −0.99, Oct −0.89) — still the newest monthly reading, a touch less calm than September and still deep in calm territory. Chicago's gauge reads −0.462 (week of Nov 6), looser than average and essentially unchanged.
The lesson: In real crises these indexes don't drift — they explode (about +9.7 in Oct 2008, sharp spikes in Mar 2020). The teaching point: high yields are a price; stress indexes measure disorder. So far, 2026 has the first without the second.
Kansas City Fed index (KCFSI), monthly. All three stress gauges currently read below zero.
Free source: FRED — STLFSI4 ↗
High-Yield Credit Spreads — the Price of Fear
FRED: BAMLH0A0HYM2What it means: “Junk bonds” are loans to shakier companies. This chart shows how much extra interest those companies must pay, above safe government bonds. When investors get scared, that extra charge shoots up — often before anything else cracks. It is the market's real-time price of fear about company failures.
Where it stands: Spreads widened by almost a full percentage point in roughly two months — from the ~2.6% area (near the most relaxed levels investors ever charge) up to 3.48% on Nov 5 — then narrowed back through the bounce: 3.42% on Nov 6 and 3.20% on Nov 12 (the latest FRED reading). The fear that rose during the selloff eased again — and the level was never high.
The lesson: Credit panics took this gauge to 19.88% (Dec 2008) and 10.87% (Mar 2020); readings above ~8% have accompanied every recession and credit scare. At 3.20% it says: investors charge more for risk than in August, but are nowhere near pricing a crisis — nervous, not panicked. At the calm end, speed matters more than level: a fast doubling from quiet levels is exactly how past episodes started, which is why desks watch the pace.
August value is the approximate ~2.6% area reported by Reuters (Oct 20, 2026). FRED's version of this series shows roughly the last 3 years.
Free source: FRED — BAMLH0A0HYM2 ↗
Jobless Claims — the Fastest Layoff Detector
FRED: ICSAWhat it means: Every Thursday, the government counts how many people filed for unemployment benefits for the first time last week. It is the fastest layoff detector there is — one week old, not one month. Professionals watch the 4-week average, not one noisy week.
Where it stands: The 4-week average is 195,250 — the lowest since October 1969. Claims rose 9,000 from the prior week (the week ending Oct 31 was revised up, from 189,000 to 191,000) to exactly 200,000 — a ninth straight week at or below 200,000 (U.S. Department of Labor, released Nov 12). Recent weeks form a tight, low band — about 230K in early June, 198–207K through July–August, and 189–200K since mid-September.
The lesson: Before past recessions, claims climbed and stayed climbing — sustained moves above ~300K were the classic warning, and claims spiked to a record 6.14 million in April 2020. Today's pattern is the opposite. Hiring looked weak for months — a “low-hire, low-fire” market — but the October jobs report (Nov 6) showed +165,000 jobs, with August and September revised up by 84,000 combined, so the low-hire half of that label fits less well than it did a week ago. Claims say firms aren't cutting — and October says they added more than expected.
Thousands of first-time filings. Sustained readings above ~300K preceded past recessions.
Free source: FRED — ICSA ↗
Housing Starts — the Rate-Sensitive Canary
FRED: HOUSTWhat it means: How many new homes builders broke ground on this month (scaled to a yearly rate). A house is the most interest-sensitive big purchase in the economy, so building usually slows before the rest of the economy does — a classic early-warning chart. Its forward-looking cousin, building permits, is literally one of the LEI's 10 ingredients.
Where it stands: Down 5.2% from August's 1.288M and 7.4% below a year earlier. The 2026 path: May 1.182M, Jun 1.439M, Jul 1.369M, Aug 1.288M, Sep 1.221M — choppy, but trending down through the autumn as mortgage rates rose back above 7%. The October starts report is due Nov 18. The Census Bureau attaches wide error bars to monthly changes, so pros read the trend, not one print.
The lesson: Housing rolled over one to two years before the 2007–09 recession, and starts roughly halved from peak to trough. Today's decline is real but starts from a mid-range level and is rate-driven — an affordability squeeze, not the overbuilding collapse of 2006. A slowdown signal to watch, not (yet) a 2008 pattern.
Millions of homes, seasonally adjusted annual rate.
Free source: FRED — HOUST ↗
The Dashboards — and Asking the Bosses
FRED recession pages · The CFO SurveyWhat it means: Two free ways to see the whole board at once. FRED — the St. Louis Fed's free database — collects the canonical recession gauges on one page, and every FRED chart automatically shades past recessions in grey, so any indicator can be read against history at a glance. And every quarter, Duke University and the Federal Reserve ask about 500 chief financial officers — the people who sign the cheques — how optimistic they are and whether they plan to spend and hire.
Where it stands: CFO optimism about the US economy: 60.3 (Q2: 60.6) — steady, not euphoric. The crack inside: optimism fell and money constraints bit hardest at small firms — 20% said financial constraints stopped them covering costs or chasing opportunities, about double the large-firm rate — and monetary policy replaced inflation as CFOs' top concern.
The lesson: Historically, CFO optimism has sagged ahead of downturns. A flat 60 with a small-firm crack reads “watch,” not “alarm.” And the bigger lesson: the government's free website IS the dashboard — the paid products mostly add commentary.
Free source: Richmond Fed — The CFO Survey ↗
Part Two — The Reports Library
The publications and data libraries institutions actually study for market patterns and history — what each one is in plain words, where the latest edition stands, and whether you can honestly get it.
UBS Global Investment Returns Yearbook
What it is: A yearly book that answers: “Over the last 125 years, in every major country, how much did stocks, bonds and cash actually make after inflation?” — with the receipts. It is THE long-run history source: when academics say “since 1900, stocks returned X,” the number came from here.
Latest: 2026 edition (released Mar 3, 2026): $1 in US stocks in 1900 grew to $124,854 nominal — $3,296 after inflation. Worldwide, stocks returned 6.6%/yr after inflation vs 1.6% for bonds. And a surprise lesson: past “bubble” sectors (railroads in 1900, tech after 2000) still delivered strong multi-decade returns — bubbles and good long-run returns can coexist.
Why institutions respect it: It is the reference dataset for the long-run equity premium worldwide.
JPMorgan Guide to the Markets
What it is: A free picture book, updated every three months, showing how the stock market has usually behaved — what stocks earned, how often they fell, what things cost compared with history. About 80–100 charts: valuations and what followed them, returns by decade, earnings, interest rates, and “principles of investing” pages, like how cash vs stocks did over 1, 5, 10 and 20 years.
Latest: Latest: 4Q 2026 (data as of Sep 30, 2026). Current message: US stocks look expensive (forward P/E ~19×, CAPE ~40.9), so future returns will probably run lower than the last decade's; a 60/40 portfolio nobody rebalanced has drifted to roughly 74% stocks; and in the Guide's data there has never been a negative 20-year period for US stocks — the house argument for staying invested instead of timing.
Why institutions respect it: The most widely used advisor chart book in the industry — the same data strategists present to pension funds.
Shiller's Yale Data (CAPE)
What it is: A free spreadsheet from a Nobel-prize economist with 150+ years of US stock-market data (back to January 1871): prices, dividends, earnings, inflation, interest rates — plus his CAPE ratio, which divides today's price by 10 years of inflation-adjusted earnings to answer “is the market cheap or expensive versus its own history?”
Latest: Updated monthly through 2026. CAPE ≈ 40.9 — near the highest ever outside the 2000 dot-com peak (~44), versus a long-run average around 17. Translation: US stocks are priced for very good news. Historically, starting CAPEs this high were followed by below-average 10-year returns — not necessarily a crash. CAPE is a poor timing tool and a decent expectations tool.
Why institutions respect it: The longest clean US market dataset in existence, and the origin of the CAPE measure every strategist quotes.
Ken French Data Library
What it is: A free filing cabinet of stock-market pattern data. It lets anyone check claims like “small stocks beat big stocks” or “cheap stocks beat expensive ones” month by month back to 1926 — the raw returns behind the famous Fama–French factors (market, size, value, profitability, investment) and momentum.
Latest: Continuously updated (monthly through 2026). The durable beginner lesson from this dataset: the market itself explains most returns; the size and value premiums were large in the 20th century and much weaker after ~2000; momentum has been the most stubborn pattern of all.
Why institutions respect it: The shared public yardstick every factor claim in finance is measured against.
Damodaran's NYU Data
What it is: A famous valuation professor gives away the spreadsheet that shows, year by year since 1928, what $100 became in US stocks vs Treasury bonds vs cash — plus industry margins, multiples and risk data used worldwide for valuing companies.
Latest: Dataset updated January 2026 (through end-2025). The file's headline lesson: over 1928–2025 stocks compounded far ahead of bonds and bills — but with multi-year stretches (1929–32, 2000–02, and 2022 for bonds) where even the “safe” asset lost heavily. (Use the data download page — the returns sub-page carries an old footer date.)
Why institutions respect it: The default starting point for practitioners who need long-run US returns in one file.
SPIVA Scorecards
What it is: Every six months, S&P checks: did the professionals who charge fees actually beat a cheap index fund? It counts every fund — including funds that closed, so the losers can't quietly disappear — over 1, 3, 5, 10, 15 and 20 years. SPIVA stands for “S&P Indices Versus Active.”
Latest: Mid-Year 2026 scorecard: 67% of large-cap US funds lagged the S&P 500 in the first half of 2026 — and over 20 years, 92.6% underperformed. Companion “Persistence” scorecards show past winners mostly don't stay winners. The pattern hasn't changed in 24 years of scorecards — the most reliable pattern in all of fund management.
Why institutions respect it: The official scoreboard of the active-vs-index debate, counted without survivorship bias.
Note: Figures here are from secondary coverage citing the official scorecards — the publisher's page could not be opened directly during research. Re-check on the official page before quoting elsewhere.
Morningstar “Mind the Gap”
What it is: Morningstar checks whether investors in a fund actually made the money the fund claims — because people tend to pile in after the good years and leave after the bad ones. It compares each fund's official return with the return its investors actually earned, across roughly 23,000 US funds and ETFs.
Latest: 2026 edition (10 years to Dec 31, 2025): investors earned 8.7%/yr while their funds returned 9.9%/yr — a 1.2-point yearly gap, about 12% of the return given away to bad timing, roughly $3.8 trillion. The gap is smallest in boring allocation and index funds, largest in volatile sector funds — the kind of fund you pick changes how much your own behaviour costs you.
Why institutions respect it: The method and data logic are published openly — the behaviour-gap finding, with cleaner maths.
CFA Research Foundation — incl. “Exponential Wealth”
What it is: The investment profession's own free library. Its newest big book, Exponential Wealth (2026), pulls together 200+ years of “what did stocks and bonds actually earn” — 398 pages built from Roger Ibbotson's work — the modern, free replacement for the old paid yearbooks.
Latest: Current flagship: Exponential Wealth (2026). Its headline: across two centuries of wars, depressions and inflations, stocks' compounding advantage over bonds and bills is the single most robust fact in the data — and the post-1982 era of falling interest rates was a tailwind that cannot repeat from today's levels.
Why institutions respect it: The profession's own body publishing practitioner research with no sales agenda.
AQR Data Library & Papers
What it is: A big hedge fund gives away the data behind its best-known discoveries — like “stocks that have been going up tend to keep going up for a while” (momentum), cheap-vs-expensive (value), and defensive stocks — so anyone can test the claims. The datasets are labelled hypothetical long/short portfolios, not fund marketing.
Latest: Library live and updated monthly in 2026. Its best educational use: the same file that shows value investing's long-run premium also shows its lost decade (roughly 2017–2020) — the best free antidote to cherry-picked pattern claims.
Why institutions respect it: AQR's founders wrote many of the canonical factor papers themselves.
Research Affiliates — Asset Allocation Interactive
What it is: A research firm that publishes, free, its best guess of what each type of investment will earn over the next 10 years — and shows its maths: what you earn ≈ the income it pays + how fast it grows + whether it gets cheaper or pricier. The tool covers some 140–160 asset classes in 40+ countries.
Latest: 2026 view: US growth stocks expected to return only ~1.5%/yr nominal over 10 years (stretched valuations are a “hefty drain”), versus ~10–12%/yr for emerging-market value strategies. The firm's forecasts are explicit, dated, and later graded — a rarity.
Why institutions respect it: Arnott is one of the most-published practitioners in finance, and the forecasts show their work.
Stock Trader's Almanac
What it is: A yearly diary of the stock market's habits: which months, weeks and parts of the 4-year election cycle have usually been strong or weak since the 1950s and earlier. Home of the Best Six Months pattern (most gains historically landed Nov–Apr — “sell in May” is the folk version), the January Barometer (“as January goes, so goes the year”), the Santa Claus Rally, and the presidential-cycle sweet spot (midterm-year Q4 to pre-election Q2).
Latest: Current book: the 2027 edition — the 60th anniversary. The Almanac itself presents these as tendencies, not laws: the January Barometer's own advertised record is ~83% since 1950 with 12 major errors, and in any single year the calendar patterns can and do fail.
Why institutions respect it: Sixty years of continuous publication; every pattern is stated as a testable historical average, not a story.
SentimenTrader
What it is: A website that answers questions like: “The last 40 times investors were this scared, what did stocks do over the next 1, 3 and 12 months?” — with the full list of past cases, good and bad. 30,000+ data series (sentiment surveys, breadth, seasonality, volatility, macro), some with near a century of history, plus a backtest engine. It sells the historical evidence and makes no buy/sell recommendations — you decide.
Latest: Live in 2026, now including its DataEdge API (2,800+ indicators) for subscribers — data that used to be institutional-only. Best beginner framing: a library of base rates. Its real lesson is how wide the range of past outcomes is after any “signal” — much wider than social-media pattern posts suggest.
Why institutions respect it: Every claim is shown as a full historical distribution, including the failures — the anti-cherry-picking pattern library.
DALBAR — Quantitative Analysis of Investor Behavior (QAIB)
What it is: Each year DALBAR measures how much money ordinary investors lose by jumping in and out at the wrong times — buying after rises, selling after falls — compared with simply sitting still. It uses fund-flow data going back to 1985.
Latest: 2026 edition (32nd): the gap nearly vanished in 2025 — the average equity investor made 17.16% vs the S&P 500's 17.88% (a 0.72-point gap, third-smallest since 1985), after an unusually huge 8.48-point gap the year before. Bond investors still lagged badly (2.41% vs 7.30%). Teaching point: the gap is worst after volatile, scary years — exactly when beginners feel most like trading. (DALBAR's method has academic critics; pair it with Mind the Gap.)
Why institutions respect it: The most-quoted measurement of how investor timing destroys returns.
Apollo “The Daily Spark”
What it is: A top Wall Street economist sends one chart every weekday morning with a few sentences explaining why it matters — valuations, credit conditions, housing, AI spending. It is the easiest free way to see what professionals are looking at today.
Latest: Late-October 2026 Sparks argue the AI boom shows “no signs of payoff outside tech”: revenue per worker and profit margins are flat for the “S&P 493” (the index minus the Magnificent 7), with a warning that AI valuations face a “painful repricing” risk if returns don't arrive.
Why institutions respect it: Sløk is one of the most-followed economists on Wall Street; the Spark is the chart everyone forwards that day.
Vanguard Research & 10-Year Forecasts
What it is: One of the two biggest investment firms in the world publishes, free, what it thinks stocks and bonds will earn over the next 10 years — from its Vanguard Capital Markets Model — plus plain-English research on investor behaviour and investing principles.
Latest: Latest forecasts (Q3 2026): US stocks 4.8–6.8%/yr for 10 years, developed ex-US stocks 5.3–7.3%, US aggregate bonds 5.1–6.1%. Note the regime change: bonds now rival stocks in the forecast — a post-2022 world. The shared message with BlackRock below: expect less than the last decade, because prices start high.
Why institutions respect it: The largest asset managers publish their capital-market assumptions in the open; competitors benchmark against them.
BlackRock Investment Institute — Weekly Commentary
What it is: The other giant's free house view: a weekly commentary on what regime the market is in and where they are leaning — plus longer outlook work. Less pattern archaeology, more “what do history and valuations imply from here.”
Latest: Week of Oct 12, 2026: BlackRock downgraded US equities to neutral and upgraded Treasuries (overweight short-dated) “to lock in income” with the 10-year near 5.25% — described as managing risk, not changing core convictions.
Why institutions respect it: One of the two largest asset managers alive, publishing its positioning logic in the open.
The Academic Anomaly Literature — and Open Source Asset Pricing
What it is: Scientists collected every “secret pattern” ever published about stocks — about 200–300 of them — and checked what happened after the secret came out. On average the patterns kept only about half their power: once everyone knows, everyone trades on it, and the easy money disappears.
Latest: The key numbers: McLean & Pontiff (2016) found 97 published predictors earned 26% less out-of-sample and 58% less after publication. Harvey, Liu & Zhu catalogued the “factor zoo” (~300 claimed anomalies) and argued most are likely false discoveries from data-mining. Open Source Asset Pricing keeps free, replicated data and code for ~200 predictors, updated monthly. The honest summary of the whole literature: some patterns are real — all of them shrink.
Why institutions respect it: This is the peer-reviewed referee system for every pattern claim on this page.
Jordà–Schularick–Taylor Macrohistory Database
What it is: 150 years of “what did owning houses vs stocks vs bonds pay?” for 18 advanced economies, in one free file — annual returns on stocks, housing, bonds and cash since 1870, plus bank credit and crisis dates. The data behind the famous paper “The Rate of Return on Everything.”
Latest: Latest release: R.6. Continuously maintained academic infrastructure — no headline, just the deepest cross-country history available free.
Why institutions respect it: The companion dataset to one of the most-cited long-run return studies ever published.
Callan Periodic Table of Investment Returns
What it is: A one-page quilt: 9 types of investment ranked best-to-worst for every year. Its whole point is an anti-pattern — last year's winner is rarely next year's. The fastest visual cure for chasing whatever just went up.
Latest: Free on Callan's site, with monthly editions and a 12-table collection. Created in 1999 and still the chart advisors pin up to teach diversification.
Why institutions respect it: An institutional consultant's teaching chart, given away — its lesson is pattern humility.
JPMorgan Long-Term Capital Market Assumptions
What it is: Once a year, JPMorgan publishes its 10–15-year forecasts for what each type of investment is expected to earn and how bumpy the ride will be — the bank-world counterpart to the Vanguard and Research Affiliates forecasts in this library.
Latest: Annual cycle (29th edition for 2025; the 2025 USD edition forecast 6.4%/yr for a 60/40 portfolio). Useful for comparing how the big houses' long-run expectations line up — and where they disagree.
Why institutions respect it: One of the industry's benchmark long-horizon forecast publications.
SBBI Yearbook (Stocks, Bonds, Bills & Inflation) — retired
What it is: The granddaddy of “how much have US stocks made since 1926” tables — large stocks, small stocks, bonds, bills and inflation, the original “stocks beat bonds beat bills beat inflation” source. It changed owners several times (Ibbotson → Morningstar → Duff & Phelps → Kroll) and has effectively retired. Don't pay a reseller for a “current” edition; there isn't one. Its living, free successor is Exponential Wealth in the CFA entry above.
Latest: Final Kroll yearbook: 2023. Morningstar announced discontinuation of the underlying indices in early 2025; the data tradition continues inside Kroll's subscription Cost of Capital Navigator.
Why institutions respect it: For 40 years it was the default citation for US long-run returns and the small-stock effect.
Ned Davis Research (NDR)
What it is: A research shop that combines dozens of charts — sentiment, breadth, macro — into one disciplined verdict: “evidence says be bold / be careful.” Its decennial reviews set a decade against ~150 years of market history. Sold to big institutions, not the public.
Latest: No public headline is legitimately available — NDR does not publish its current stance free. That IS the finding: if a website claims to show “NDR's latest signal” free, it is at best stale and at worst a leak. Some NDR research reaches individuals indirectly through Interactive Brokers' client portal.
Why institutions respect it: Four decades of model-driven, non-storytelling calls; Ned Davis wrote the book on being right or making money.
BofA Global Research — “The Flow Show” & Fund Manager Survey
What it is: Bank of America asks the world's big money managers every month: “How much cash are you holding, what is everyone buying, what scares you?” When the answers get extreme — almost no cash left, everyone in the same trade — the crowd has often been about to be wrong. The weekly Flow Show tracks money flows and positioning, with BofA's Bull & Bear indicator.
Latest: October 2026 Fund Manager Survey (161 managers, $453B): cash 3.9% — inside Hartnett's ≤4% contrarian “sell” zone; Bull & Bear at 8.1, above his 8.0 sell trigger; most crowded trade: long global semiconductors (45%); biggest feared risk: a disorderly jump in bond yields (31%), ahead of an AI bubble (23%). Beginner translation: “everyone is already all-in on the same trades” — a reason for caution about the next surprise, not a timing signal by itself. Retail sees the survey 1–3 days later via Reuters/Dow Jones summaries; do not use leaked-PDF sites.
Why institutions respect it: The largest regular poll of the people who actually move the money; its extremes have marked several major turning points.
No public product page exists — see the survey via news summaries ↗
If you only study five — start here
The honest access map: most of the great history is free (JPMorgan's Guide, the UBS summary, Shiller, French, Damodaran, SPIVA, Mind the Gap, FRED). Seasonality costs a little (the Almanac). Two famous names — BofA Global Research and Ned Davis Research — cannot be bought by individuals at any price; you meet them second-hand, through news coverage, a few days late.
Sources: FRED (Federal Reserve Bank of St. Louis), the Federal Reserve Bank of New York, the Conference Board, ISM, the Richmond Fed, and the publishers linked on each card. Readings as of November 15, 2026; this page is refreshed weekly. Education only — not financial advice.