Market temperature
Twenty indicators, one number. Each series is reduced to where it sits in its own history, flipped where a low reading is the dangerous end, and averaged. It tells you how stretched the present is — not what happens next.
Every indicator, ranked by how stretched it is
100 = the top of that series' own historyHow the score is built
Three steps, in this order. First, each indicator is converted to a percentile: the share of its own observations since the start of its record that sit below today's reading. A buffett indicator at the 88th percentile means valuations have been higher only twelve percent of the time since 1950 — it says nothing about the level itself, which is the point, because a debt-to-GDP ratio and a credit spread cannot be averaged in their raw units.
Second, the percentile is flipped for series where the low end is the dangerous end. The yield curve, bank credit growth, M2 growth, reserves and the reverse repo balance are all signals that deteriorate as they fall, so their score becomes 100 minus the percentile. Without this step the composite would be incoherent: a late-cycle economy typically combines expensive valuations with deteriorating liquidity, and the two would cancel out in a naive average.
Third, the twenty scores are averaged inside their group, and the four group scores are averaged with equal weight. The order matters. Liquidity and macro has eight members while valuation has two, so a flat average would let the plumbing outvote what you pay for earnings three to one. Grouping first keeps each theme at 25 percent of the total and keeps the headline number from being a referendum on the Fed's balance sheet.
What each group measures
Valuation covers market capitalisation against GDP and the real 10-year yield. Both answer the same question from different ends: what a dollar of future cash flow costs today, once you strip out inflation. This is the slowest-moving group and the one that stays stretched for years without anything breaking.
Recession and stress covers the Sahm rule, the St. Louis Fed financial stress index, realised Treasury volatility, the Chicago Fed credit subindex, margin debt and bank lending standards. These are the fastest-moving inputs and the ones that actually confirm a turning point, usually months after the market has already moved.
Liquidity and macro covers both yield curve spreads, M2 growth, reserves, the broad dollar, breakeven inflation, the credit impulse and reverse repo usage. This is the mechanical layer: how much money exists, what it costs, and whether it is flowing into credit or sitting at the central bank.
Real economy and fiscal covers CPI, industrial production, unemployment and federal debt against GDP. The slowest group of the four — a quarter can pass before a change here shows up — but it is the one that decides whether the other three were describing a slowdown or a recession.
Why direction matters more than level
A single indicator read the wrong way is worse than no indicator. A credit impulse at the 8th percentile is not calm, it is a credit contraction; a reverse repo balance at the 5th percentile is not cheap money, it is the buffer that absorbed excess reserves having been spent. Both are near the bottom of their records and both are warning signs, which is exactly why the scoring flips them.
The same logic runs the other way for the curve. A deeply inverted 10Y–2Y spread sits near the bottom of its range and scores high on risk, because every recession since the 1970s was preceded by one. The raw percentile would call that a low reading and the composite would read it as calm.
What a high reading does and does not mean
It does not mean sell. The 1990s spent seven years above 70 without a recession, and the score has spent long stretches in the sixties while equities compounded. Percentiles are a description of the present, not a probability statement about the next twelve months, and nothing on this page is a forecast.
It does mean the set of conditions is unusual. When fourteen of twenty readings sit above their medians at once, the configuration has few historical precedents, and the useful question is which group is doing the work. Expensive valuations with calm credit is a different situation from calm valuations with tightening standards, and the group scores are there to separate them.
Three limits are worth stating plainly. Quarterly inputs — margin debt, debt to GDP, the buffett indicator, lending standards — are published two to three months after the period they describe, so the composite always contains readings that are older than they look. Percentiles depend on where each series starts: one that begins in 1982 carries a different notion of normal from one that begins in 1960. And the weights are equal by choice, not by evidence, because there is no defensible way to estimate them without overfitting to a handful of cycles.
Where the data comes from
Every input is a U.S. federal release — Federal Reserve, Treasury, BEA or BLS — pulled through FRED and rebuilt twice a day. Nothing here is licensed from a vendor, which is why there is no high-yield spread, no VIX and no Shiller CAPE: their terms forbid redistribution. Where a restricted series would have been the obvious choice, the page uses a public-domain equivalent and says so on the detail page.
The numbers on this dashboard are calculated from the same free view as the detail pages. On daily and weekly series the most recent window is reserved for members, so the composite you see here is computed through the cut-off date rather than through today. Members see the same score on current data.
Questions
What is the market temperature?
It is the equal-weighted average of twenty indicators, each first reduced to its own historical percentile and then flipped where a low reading is the risky end. The result runs from 0 to 100: 0 means every series sits at the bottom of its own history, 100 means every series is at the top.
Does a high reading mean a crash is coming?
No. The score describes where conditions are today relative to their own history. It contains no timing information, and several of its inputs are released with a lag of a month or a quarter, so it is a description of the present rather than a forecast of the next twelve months.
Why are some indicators scored as riskier when they fall?
Because direction is not uniform. An inverted yield curve, contracting bank credit and shrinking reserves are all stress signals at the low end of their ranges, so their percentiles are flipped. Without that flip the average would cancel itself out, since a late-cycle economy usually combines expensive valuations with deteriorating liquidity.
Is the dashboard free?
Yes, the whole page is free, including every group score and every individual reading. The numbers are calculated from the same free view as the detail pages, so the most recent window on daily and weekly series is withheld here exactly as it is there.
Important
Not investment advice. This page is published for information and education only. It is not a recommendation, solicitation, or offer to buy or sell any security. A composite score is a summary of historical positions, not a prediction. Market data may be delayed, incomplete, or revised. Past performance does not guarantee future results. Do your own research or consult a licensed financial professional before acting on anything you read here.