apps2 · gdp — how stretched is the market versus the economy?

How large, profitable, owned, levered, concentrated, and fiscally stretched is the market versus the economy? Each tile is one official ratio. The big number is the latest printed reading (a quarterly FRED print, or a monthly FINRA print) — held months after a print are not treated as new information. The percentile is this reading versus that series’ own expanding history (how unusual it is, not a forecast). Descriptive only. Nothing here says what price does next.

The ratios, together Each line is that ratio in z-score versus its own history, so a 200% cap/GDP and a 4% margin/GDP can sit on one axis. The S&P 500 and NAS100 are the dark baselines (price, their own scale). Click a line to isolate it. A spike that shows up in several ratios at once is the historically expensive cluster — 2000, 2021, and the present. Quarterly prints are held as steps; that is the data, not a smooth interpolation.

Does any of this actually mark subsequent returns? In-sample. For every month we have a finished 5-year or 10-year window, that month’s ratio is dropped into a quintile of the whole sample, and we record what the S&P 500 price did next (split-adjusted close; dividends not included). Q1 = the cheapest 20% of that ratio; Q5 = the richest. This is association on history, not a trading result, and it carries no costs. A relationship that exists at a 10-year horizon is not a timer.

What this is apps2/gdp. Official series, documented unit conversions. Mag 7 earnings are SEC 10-Q GAAP net income, not prices. Federal debt/GDP is FRED GFDEGDQ188S as published. The page refuses Wilshire-index-points-as-dollars and refuses to invent a daily market-cap from a quarterly print.

Is the economy flashing extreme weakness or strength right now? modeled A composite built from the owner’s ask (GDP, earnings, debt burden, the economy-level price ratio, real M2, foreign Treasury holdings + the factors that belong with them). Three dials: Direction (is the economy weakening?), Fragility (would a downturn be amplified?), Trigger (does the market confirm?). Each input is a robust z-score versus its own last decade; weakness = negative direction, amplified by agreement and fragility, halved when the market does not confirm. The score shown is a percentile of its own 50-year history (in-sample, recomputed each build). EXTREME is a visible 4-gate checklist, never a hidden AND. Descriptive only — nothing here says what price does next.

Weakness & strength scores, 1974 → now — with the S&P 500 behind them The red line is the weakness score, the green line the strength score (each a percentile of its own full history, 0–100). The S&P 500 rides faintly behind on its own scale for context. Shaded columns mark months where all four extreme gates passed (red = weakness, green = strength). The thin grey band along the bottom is coverage — how many of the 22 sub-signals were alive that month; ticks mark each sub-signal’s birth. Click for the full readout; the dateline syncs with every chart on the page.

What is driving it? — every sub-signal, tape + lines The composite never hides its components. TOP — the stretch tape: one row per sub-signal, one column per month; red = weak/fragile versus that signal’s own decade, blue = strong. Click any column to sort rows by that month (click again to flip; Reset order restores blocks). BOTTOM — the same components as lines, each selectable, with SPX/NAS100 baselines. Sub-signals are already z-scores, so σ units are comparable across every row.

Has a high weakness score marked anything? In-sample, same rules as the ratio table above: every month is dropped into a quintile of the whole weakness-score history, and we record what the S&P 500 price did over the following 6 and 12 months (dividends not included). Below it, every episode where all four gates passed — dates, length, and what followed. n is small and shown. History, not a forecast: note 2009-03 fired at the bottom.

How the composite is built The formula as built, every vendor series with its verified start date and the publication lag actually applied (no look-ahead), the v1 tunables, the two documented substitutions, and the record-first store. Provenance: owner ask 2026-08-20; framework adapted from an external AI’s answer, improved by the orchestrator; every constant is in SPEC.md.