Method
The ETF-price model is the interim published proxy; the stock-value model is the one intended to measure wealth composition and has no results yet. Method version: CCI weekly v4 BTC-Friday-21UTC experimental.
Formula
CCI(t) = Σ wᵢ · ln( pᵢ(t) / pᵢ(base) ). Each input price is set relative to its level in the base week, 10 Oct 2025; the weights are fixed and sum to zero, so a common move in every input leaves the index unchanged.
| Input | Weight | Group |
|---|---|---|
| Bitcoin (BTC) | +1/3 | growth |
| US equities (VTI) | +1/3 | growth |
| Semiconductors (SOXX) | +1/3 | growth |
| Gold (GLD) | −1/2 | defensive |
| US Treasuries (IEF) | −1/2 | defensive |
| Oil (USO) | 0 | zero weight, context only |
| Silver (SLV) | 0 | zero weight, context only |
Two scales
The risk-on score is a level: the mean log ratio of Bitcoin, equities and semiconductors to Treasuries, about 3.0 to 3.4 in the published weeks. The CCI is a weighted log change since the 10 Oct 2025 baseline, between -0.2347 and +0.2216 in the published weeks. Only changes in the risk-on score are read, never its level.
Timing rules
- One snapshot per week, cutoff Friday 23:59 UTC.
- All inputs are sampled at 21:00 UTC on that Friday; Bitcoin is sampled at the same hour as the ETF close.
- An observation counts as available one hour after it is observed. This is an assumed convention, not a verified provider timestamp; the history is labelled “historical availability assumed”.
- If any input is missing the week is incomplete and no index value is published. Nothing is interpolated.
- The four-week average needs four consecutive complete weeks.
Sources and observation timestamps (latest week)
Sources & observation timestamps
| Exposure | Source | Observed | Available |
|---|---|---|---|
| Bitcoin | CoinGecko:bitcoin | 2026-10-09T21:00:00+00:00 | 2026-10-09T22:00:00+00:00 |
| US equities | AlphaVantage:VTI | 2026-10-09T21:00:00+00:00 | 2026-10-09T22:00:00+00:00 |
| Semiconductors | AlphaVantage:SOXX | 2026-10-09T21:00:00+00:00 | 2026-10-09T22:00:00+00:00 |
| Oil exposure | AlphaVantage:USO | 2026-10-09T21:00:00+00:00 | 2026-10-09T22:00:00+00:00 |
| Gold exposure | AlphaVantage:GLD | 2026-10-09T21:00:00+00:00 | 2026-10-09T22:00:00+00:00 |
| Silver exposure | AlphaVantage:SLV | 2026-10-09T21:00:00+00:00 | 2026-10-09T22:00:00+00:00 |
| US Treasuries | AlphaVantage:IEF | 2026-10-09T21:00:00+00:00 | 2026-10-09T22:00:00+00:00 |
Signal rule
Δ4 = score(t) − score(t−4). σ26 is the sample standard deviation of the 26 weekly score levels from t−29 to t−4 inclusive. The risk-on signal fires when |Δ4| is strictly greater than σ26 with complete history; its direction is the sign of Δ4. A fired week is marked marginal when |Δ4| is less than 1.05 × σ26.
Evaluation samples
47 weekly samples were evaluated: an association test of the change in the smoothed CCI against the next-week change in Bitcoin, from 2024-01-01, compared with Bitcoin’s own past-week momentum on the same weeks. Recorded as: “Forecast association only; no executed trades or significance claim”. No performance guarantee.
Method notes
Address · 2026-10-11
Address moved to cci.calvyx.com/us/ on 2026-10-11. No numbers changed.
Method update · 2026-10-10
Bitcoin is now sampled at 21:00 UTC Friday, the same hour as the ETFs. Historical CCI values, group scores and signals were recomputed. Weights, the 2025-10-10 base date and signal rules are unchanged. Availability uses the assumed observation + 1 hour convention, not a verified provider publication timestamp. When an exact 21:00 observation is absent, the latest earlier observation that Friday is used; without one, the snapshot is incomplete.
Download previous v3 results (unaligned Bitcoin)
Every signal-status or direction change:
| Week | Model / score | Before | After |
|---|---|---|---|
| 2026-07-10 | risk-on score (ETF-price model) | not fired / down | not fired / up |
| 2026-09-25 | risk-on score (ETF-price model) | not fired / up | FIRED / up |
The JSON association statistics retain the archived v3 evaluation; this update recomputes the CCI and signal history.
Methodology note
The stock-value CCI models selected national-wealth assets as a composition of seven asset fractions summing to one; the ETF-price baseline measures price proxies, not national wealth. A signal compares the absolute four-week change in the risk-on or defense score against the sample standard deviation of 26 weekly score levels ending at t−4, and fires only when the change strictly exceeds that threshold with complete history. Bonds are the reference asset; oil is tracked as an external ratio only.
Disclaimer
This is a research tool, not financial advice. The stock-value model requires market caps, metal holdings, Treasury debt, and currency in circulation; where inputs are missing, the model returns insufficient data. Past signals do not guarantee future results.
Method and sources
Method specification · Data sources list
Method specification
- Stock-value composition
- Bitcoin market cap only (no stablecoins); total US equity market cap minus semiconductor market cap; semiconductor market cap; gold price × stated US gold holding stock; silver price × stated US silver holding stock; marketable Treasury debt; currency in circulation. All values are in USD; metal holdings use stated troy ounces and prices in USD per troy ounce. These seven positive parts are normalized together to sum to one. Missing inputs leave the model insufficient data.
- Coordinates and scores
- x_i = ln(f_i / f_bonds) for the six non-bond parts. risk-on score = (x_crypto + x_equity + x_semis) / 3; defense = (x_gold + x_silver + x_cash) / 3. Bonds have no weighted coordinate. Scores remain separate, with fixed equal weights.
- Signals
- delta4 = score_t − score_(t−4). vol = sample standard deviation (ddof=1) of score_(t−29) through score_(t−4), inclusive: exactly 26 consecutive weekly levels. Fire only when abs(delta4) > vol. Direction is the sign of delta4, separate from threshold clearance. Short, missing or nonfinite history returns no signal.
- ETF comparison
- The published ETF-price dashboard remains the baseline. Comparison group scores use BTC-USD, VTI and SOXX versus IEF for risk-on score; GLD, SLV and an explicitly supplied cash proxy versus IEF for defense. Prices use adjusted closes; VTI and SOXX overlap. No cash proxy is invented and ETF prices never substitute for stock values.
- Chart colors and validation
- Median colors are descriptive, not signal thresholds: each score is compared with its preceding 26 weekly levels. Both above median is mixed; missing history is unavailable. Available-year results are descriptive and in-sample; no fitted weights or active train/test split are claimed. Oil and external ratios never enter the scores or signals.
Data sources list
- CoinMarketCap
- World Bank
- SIFMA
- EIA
- LBMA
- World Gold Council
- Treasury
- Fed Diary of Consumer Payment Choice
- Glassnode
- SOX / Philadelphia Semiconductor Index
Reference links do not imply that data has been loaded. Each stock-value input needs its observation date, units, source, and ownership scope; metal holding stocks must be explicitly stated. Unavailable records remain insufficient data.