M2 + dollar + inflation + Treasury flows + 401k bid + rates as ONE macro baseline — is the market running ahead of its fuel?hub401kinflationfundamentalsdecoupbaselineloading…
Is the market running ahead of its macro fuel?
The premise (printed below, always): stocks drift up by default — 401k
paychecks and money printing are a standing, price-insensitive bid. The oscillator
is the headline: for each index, divergence D = market impulse (z) − macro
baseline B (z), in σ units. z-score (jargon kept): how unusual a value is
vs its own history, in standard deviations — 0 = typical, +2 = rarer than ~97% of readings.
Above zero = the market is accelerating harder than the macro fuel justifies; below = it
is lagging its fuel; zero = moving exactly as the flows justify. Shaded episodes = the
divergence stayed beyond its 95th (or below its 5th) percentile for 5+ straight trading
days — start→stop ranges, never one-day dots. The price band below it (flush,
the TradingView indicator arrangement inverted — the oscillator is the star) shows what
SPX and NAS100 actually did, indexed to the visible window start; real closes in the
tooltip. The price band also carries the THICK modeled fair value line per index —
where price “should” sit if it only followed the macro-fuel LEVELS (expanding
no-look-ahead fit, ±1σ/±2σ corridor; modeled, never a forecast —
the audit panel below the comparison table shows the fit, its signs and its stability).
Click a line to identify it alone AND get a plain-words explainer of what it measures;
click empty space to dismiss; shift-click for the all-series dateline (x-synced across
every chart on this page). Descriptive only — this page never says what price does
next.
What is the macro engine doing? — every lever, week by week
The mandatory tape+line pair. Tape: one row per lever (plus the
baseline B under YOUR weights, and the two indexes as labeled reference rows), columns =
weeks, each cell = how much that lever’s z-score (its reading vs its own
multi-year normal) moved that week. Monthly levers print roughly one column in four; the
hatched cells between are carried, never faked into weekly noise. Colour modes
re-label the ramp: vs its own history (default) · all rows pooled on one σ
scale · rank within the week · vs SPX · vs NAS100 · own
min–max. Click any week’s header to sort by it. Lines (flush below):
the same baseline + all 11 lever z-series as individually-toggleable normalized lines vs
the SPX/NAS100 baselines (solid, dark, never muted) — z-score view by default, stretch-to-fit
and %-change as modes. The composite never hides its parts.
Select allDeselect allweakerstronger
each cell vs that row’s own history — deep green = an unusually supportive week for that lever, deep red = an unusually draining one, pale = a normal week.
Does the market actually follow the fuel? — rolling correlation, fast & slow
Correlation (jargon kept): how tightly two things move together,
from +1 (in lockstep) through 0 (no relationship) to −1 (opposite). Here: the
week-to-week CHANGE in the macro baseline B vs each index’s weekly % return, over a
trailing 52-week window (the slow, stable read) and a 13-week window (the
fast read that catches regime turns first). High = the market has been tracking its macro
fuel; near zero = price has been marching to its own drum; negative = moving against the
fuel. Weight-dependent — drag a slider and this recomputes. n = the number of weekly pairs
in each window (printed on the chart). The comparison inputs are now DRAWN on the chart:
the black baseline B fuel line (own scale, weekly steps of the same recomputed B
the correlations use) with SPX/NAS100 price behind (own scales) and the faint
weekly ΔB lane at the bottom (the literal series each correlation pairs with
that week’s return) — fuel + price are the inputs, the colored lines are how tightly
they’ve moved together at two speeds. Click any line for a plain-words explainer of
what it measures. Descriptive only.
How does each window compare? — market vs baseline vs divergence
One row per index per window (week / month / quarter / year / 4 years).
Market return = price at the window end ÷ price at the start − 1.
Baseline move = how much the macro-fuel gauge B changed across the window, in σ
units. Divergence move = how much the market-vs-fuel gap D changed. Divergence
now = today’s gap with its percentile vs the trailing 4 years (the value AND where
it sits in its own history, in one cell). Correlation = the window’s weekly
(ΔB, return) pairs; windows with fewer than 8 weekly pairs are greyed — one or
five weeks of pairs cannot mean anything, and saying so beats faking it. Click any column
header to sort; every column is shaded on its own green↔red scale. Recomputes when you
move a weight slider.
computing…
Do the manual weights match what history fitted? — the fair-value model, audited
The thick fair-value line in the price band comes from a
regression (jargon kept): for every day, “given where the fuel levels sat —
M2, the dollar, CPI, the Fed’s Treasury book, foreign custody, the 401k bid, rates —
where did price USUALLY sit?” Coefficients are fitted freely, then each
one’s SIGN is audited against its economic prior (the owner’s own examples: a
growing 401k bid should push fair value UP, Treasury selling should pull it DOWN) — a
wrong sign is flagged in red, never silently corrected, because with near-collinear
trending levels a flipped sign usually means another factor absorbed the shared trend.
The era columns refit the same model on three thirds of the window: a β that flips
sign between eras is unstable and says so. Fitted β as weights: the
“Weights” chip in the top bar can apply these fitted β’s to the
whole page’s composite (sliders stay; dragging one returns to manual). Everything
here describes the historical fit — nothing predicts.
waiting for the fair-value fit…
What is going on right now — in plain words
Every sentence below is picked from ONE served copy table (5 percentile
bands × 11 levers, 3×5 gauge bands — divergence, baseline, fair value — 5 named
cross-metric combinations, plus the click-to-explain classes), keyed to the LIVE
reading’s band — the same computed percentiles the
charts draw, never a hardcoded twin. The boring middle gets a sentence too: a quiet gauge
says so instead of going silent. Everything is descriptive — what is happening and the
mechanics behind it, never what price does next. The gauge sentences re-cut when you move
a weight slider.
loading…
What this isapps2/baseline · one macro-fuel BASELINE (M2 · dollar · inflation · Treasury flows · 401k bid · rates) vs the market’s own acceleration, scored against its trailing 4-year history. The premise is structural updrift; the measurement is the divergence.
The model, honestlyeach lever is transformed at its NATIVE cadence (weekly M2, monthly CPI, daily rates — never z-scored on forward-filled values), z-scored vs its own trailing window, then mixed as Σw·z ÷ Σ|w| with YOUR signed weights. Equal-magnitude defaults are the honest prior — the sliders are the weighting mechanism, not a fitted claim.
Signs live in the weightstransforms measure the RAW economic direction (dollar strengthening = positive z); the owner’s support-signs sit in the DEFAULT WEIGHTS (DXY at −1.00). Dragging any slider past zero INVERTS that lever — the deliberate directionality test.
Alignmentnative-cadence z-values are step-forward-filled onto the SPX trading-day grid for mixing and display only; all timestamps floored to the day before any join. A weekly lever inside a 1-week window shows its last print as a held step — never a blank.