Methodology — Every Formula, Source and Threshold
Every series we read, every formula we apply and every threshold we score against is listed below. If a reading here disagrees with someone else's, you can find out why.
Where the numbers come from
| Indicator | Group | Source series | Thresholds |
|---|---|---|---|
| Buffett Indicator | Crash risk | Yahoo Finance ^DWCF (Dow Jones U.S. Total Stock Market) / FRED GDP, quarterly | Negative >150 • Warning ≥120 |
| Yield Curve | Crash risk | FRED T10Y2Y, daily | Negative ≤-0.3 • Warning ≤0.2 |
| Shiller PE Ratio | Crash risk | multpl.com Shiller PE by month | Negative >35 • Warning ≥25 |
| US Dollar Index (DXY) | Growth | Yahoo Finance DX-Y.NYB, daily | Negative >1% (3-month change) • Warning ≥-1% (3-month change) |
| US Money Supply (M2) | Growth | FRED M2SL, monthly | Negative >15% (3-month change) • Warning ≥10% (3-month change) |
| Real Interest Rates | Growth | FRED DFII10 (10-Year TIPS constant maturity), daily | Negative ≤-1 • Warning ≤1 |
Scroll for thresholds →
Risk Page Scales
The risk pages score their readings the same way, against cutoffs fixed in advance. These set the label on a reading. They do not feed the crash-risk or growth scores above.
| Scale | What is measured | Cutoffs |
|---|---|---|
| Chokepoint tanker transits | Current week of tanker transits against the 2019 to November 2023 baseline | Negative ≤-50% • Warning ≤-20% |
| Brent crude | Three-month price change — a rise is the alarm, a fall means the shock has not arrived | Negative ≥30% • Warning ≥10% |
| US days of cover | Commercial crude plus the SPR over one day of refinery runs — how long the barrels already held can feed US refineries | Negative ≤35 days • Warning ≤45 days |
| AI builder leverage | Long-term debt divided by one year of capital spending | Negative ≥2x • Warning ≥1x |
| Credit spread gap | CCC option-adjusted spread minus AAA | Negative ≥8 pp • Warning ≥5 pp |
| Japan's Treasury holdings | Change from the largest holding Japan has reported | Negative ≤-10% • Warning ≤-5% |
| Japan's holdings over a year | Change in Japanese holdings of US Treasuries over twelve months | Negative ≤-8% • Warning ≤-3% |
| Federal interest bill | Interest as a share of federal current expenditures (BEA), against the highest quarter in the series | Negative ≥15% • Warning ≥12% |
| Yen per dollar | How weak the yen is — the weaker it gets, the more pressure to sell reserves to defend it | Negative ≥160 • Warning ≥150 |
| Japan's monthly change | Month-over-month change in the holdings, in USD millions — a fall is market value, net selling, or both | Negative ≤-50000m • Warning ≤0m |
| 30-year Treasury yield | What every future refinancing of the federal debt will cost | Negative ≥5% • Warning ≥4.5% |
Scroll for cutoffs →
How each value is computed
Buffett Indicator
The Dow Jones U.S. Total Stock Market index level is treated as one index point per $1bn of market capitalisation, then divided by the most recent quarterly GDP figure published before that trading day, expressed as a percentage. Trading days on which the index has no close are skipped, not carried as zero. Other providers pick their own market-cap source and GDP vintage, so a reading a few points away from this one is normal.
Yield Curve
Read directly from the Federal Reserve series T10Y2Y: the 10-year minus 2-year Treasury constant maturity spread. Nothing is transformed.
Shiller PE (CAPE)
Taken monthly from multpl.com, which publishes Robert Shiller's cyclically adjusted price-to-earnings series. If the page layout changes and no rows parse, the indicator reports an error rather than a stale or invented value.
US Dollar Index, M2 and Real Interest Rates
DXY comes from Yahoo Finance (DX-Y.NYB). M2 comes from the Federal Reserve series M2SL. The real interest rate is the Federal Reserve series DFII10, the 10-year Treasury Inflation-Protected Securities constant maturity yield. It matches the nominal series: DGS10 minus T10YIE equals DFII10.
How the composite is scored
Each indicator scores as positive, warning or negative against the fixed thresholds above. Crash risk is the share of the three crash indicators reading negative. Growth probability is the share of the three growth indicators reading positive. Both are plain counts. There is no weighting and no model fitting.
Growth probability above 50% with crash risk at or below 50% reads as GROWTH POSITIVE. Crash risk above 50% with growth probability at or below 50% reads as CRASH RISKY. Anything else reads as NEUTRAL MARKET.
Each computed answer is held for an hour, and the calls to the sources beneath it for fifteen minutes. Each page then rebuilds on its own hourly schedule, so a reading can be a little over two hours old. Every series here updates daily at best, and most of them monthly or quarterly, so that delay changes no number. One snapshot is stored per day, so the history shown is a recorded series, not a recomputation.
What this does not do
These are valuation and condition gauges, not timing signals. The Buffett Indicator and CAPE have both stayed in expensive territory for years at a stretch while markets kept climbing. An elevated reading says future returns start from a worse price and the market has less cushion when something breaks. It does not say when. Nothing on this site is investment advice.
Found a number you think is wrong? Check the data changelog . Every correction is listed there with the date it shipped.