Live allocation signal / methodology v1.0
A transparent reading of how defensive or risk-seeking the benchmark's AI models are through their saved portfolios. It measures allocation choices, not subsequent returns.
Signals tied to the latest live risk-taking score, weekly versus monthly agreement, and model clustering.
The newest live portfolios have a deterministic risk-taking score of 89.5 out of 100.
Risk-taking score is allocation-based, not performance-based: higher means more weight in growth, momentum, cyclical, and higher-risk assets.
The newest weekly portfolios allocate +60.00% to growth and technology, while the newest monthly portfolios allocate +64.38%.
Horizon agreement compares the newest weekly and monthly live portfolios to see whether short- and longer-window model stances line up.
The closest live allocation pair is GPT-5.5 and GPT-5.6 Sol with +95.99% cosine similarity. The current allocation outlier is Grok 4.3.
Cosine similarity measures allocation overlap between model portfolios. A value near 1.00 means the weights are very similar.
As of July 10, 2026. The current pulse combines CB-2026-07-10-1W and CB-2026-07-10-1M. Weekly and monthly are calculated separately, then receive equal weight.
The latest Monthly and Weekly decisions are calculated separately and then receive equal weight in the Combined line.
Historical points are reconstructed only from portfolios and dates in the public record. Prices and subsequent returns are excluded.
Version 1.0 uses only public, locked portfolio allocations. Prices, realized returns, model rankings, and later information do not affect the signal.
The 1–5 risk score measures standalone allocation risk and powers the existing historical model profile. The -1 to +1 risk-on loading measures whether holding the asset is defensive or expresses risk seeking.
Portfolio Pulse = 50 + 50 × Σ(allocation weight × asset risk-on loading)A fully defensive portfolio approaches 0, a neutral mix approaches 50, and a fully risk-on portfolio approaches 100.
Each model has equal weight within the latest weekly test and within the latest monthly test. Portfolio size, provider, and historical performance do not change a model's vote.
Combined Pulse = 50% × Weekly Pulse + 50% × Monthly PulseThis prevents the larger number of open monthly portfolios from overwhelming the faster tactical signal.
Agreement is the population standard deviation of each model's combined latest score. Tight means below 5 points, Mixed means 5 to 12 points, and Divided means above 12 points. Change compares the current weekly/monthly pair with the immediately preceding pair.
As of July 10, 2026. Uses only the newest weekly and monthly decisions. This is the primary reading for current model sentiment.
Uses all 146 unresolved portfolios across 25 tests. It describes capital still exposed, including older decisions.
The historical model cards retain their existing 1–5 format. Both those cards and this live signal now read from the same versioned asset definitions.
Measures standalone allocation risk. Long-duration bonds can score as risky because they can be volatile, even though their risk-on loading is defensive.
Measures the direction of the allocation signal. Oil can be volatile but only moderately risk-on when models use it as an inflation hedge.
Separates growth, defensive equity, real assets, rates and credit, international equity, broad cyclicals, liquidity, and crypto.
Behavior profiles are deterministic summaries of official frozen portfolios. They update whenever the generated public read model updates, and they separate allocation style from investment performance.
Each portfolio is scored with the same -1 to +1 risk-on loading used by AI Risk Appetite, then converted to a 0-100 scale. A model profile averages that score across its official saved portfolios.
Average top holding and HHI measure whether a model spreads capital across many assets or expresses views through a small number of larger positions.
Round-to-round turnover equals one-half of the summed absolute allocation change across assets. Weekly and monthly turnover are calculated separately, then combined for the headline metric.
Portfolio overlap is cosine similarity between allocation vectors from models in the same official round. Higher values mean a model tends to look more like the rest of the roster.
Return, Portfolio Minus S&P 500, rank, and CapitalBench Score use resolved rounds only. Live portfolios can affect behavior labels but never finalized performance.
Models with fewer than 8 official saved portfolios are labeled Early sample. The current behavior dataset has 327 portfolios across 8 models.
All 70 assets are explicitly rated. A new universe asset must receive a complete definition before the public build passes.
| Asset | Regime | Risk 1–5 | Risk-on loading | Flags |
|---|---|---|---|---|
| Regional Banks KRE | Broad and cyclical equity | 5 | +0.80 | None |
| Aerospace and Defense ITA | Broad and cyclical equity | 4 | +0.45 | None |
| Consumer Discretionary Sector XLY | Broad and cyclical equity | 4 | +0.55 | None |
| Financials Sector XLF | Broad and cyclical equity | 4 | +0.45 | None |
| Real Estate Sector XLRE | Broad and cyclical equity | 4 | +0.45 | None |
| US Mid-Cap Stocks IJH | Broad and cyclical equity | 4 | +0.60 | None |
| US Small-Cap Stocks IWM | Broad and cyclical equity | 4 | +0.70 | None |
| US Small-Cap Value IWN | Broad and cyclical equity | 4 | +0.65 | None |
| Equal-Weight S&P 500 RSP | Broad and cyclical equity | 3 | +0.40 | None |
| Industrials Sector XLI | Broad and cyclical equity | 3 | +0.40 | None |
| S&P 500 SPY | Broad and cyclical equity | 3 | +0.35 | None |
| Total US Stock Market VTI | Broad and cyclical equity | 3 | +0.35 | None |
| US Large-Cap Value IWD | Broad and cyclical equity | 3 | +0.30 | None |
| Bitcoin ETF IBIT | Crypto | 5 | +1.00 | None |
| Ethereum ETF ETHA | Crypto | 5 | +1.00 | None |
| Consumer Staples Sector XLP | Defensive equity | 3 | -0.05 | Defensive |
| Healthcare Sector XLV | Defensive equity | 3 | 0.00 | None |
| US Dividend Equities SCHD | Defensive equity | 3 | +0.10 | Defensive |
| US Low Volatility Equities SPLV | Defensive equity | 3 | -0.05 | Defensive |
| Utilities Sector XLU | Defensive equity | 3 | -0.05 | Defensive |
| Emerging Market USD Bonds EMB | Rates and credit | 4 | +0.50 | Defensive |
| High Yield Corporate Bonds HYG | Rates and credit | 4 | +0.55 | Defensive |
| Long-Term US Treasury Bonds TLT | Rates and credit | 4 | -0.50 | Defensive |
| Agency Mortgage-Backed Bonds MBB | Rates and credit | 3 | -0.35 | Defensive |
| Intermediate-Term US Treasury Bonds IEF | Rates and credit | 3 | -0.55 | Defensive |
| International Aggregate Bonds BNDX | Rates and credit | 3 | -0.25 | Defensive |
| Investment Grade Corporate Bonds LQD | Rates and credit | 3 | -0.20 | Defensive |
| Treasury Inflation-Protected Securities TIP | Rates and credit | 3 | -0.35 | Defensive |
| US Aggregate Bond Market AGG | Rates and credit | 3 | -0.55 | Defensive |
| Municipal Bonds MUB | Rates and credit | 2 | -0.65 | Defensive |
| Autonomous Technology and Robotics ARKQ | Growth and technology | 5 | +0.90 | Technology |
| Biotechnology XBI | Growth and technology | 5 | +0.85 | Technology |
| Broad AI Technology AIQ | Growth and technology | 5 | +0.90 | Technology |
| Semiconductors SMH | Growth and technology | 5 | +0.95 | Technology |
| Solar Energy TAN | Growth and technology | 5 | +0.85 | None |
| Communication Services Sector XLC | Growth and technology | 4 | +0.60 | None |
| Cybersecurity CIBR | Growth and technology | 4 | +0.75 | Technology |
| Nasdaq 100 QQQ | Growth and technology | 4 | +0.75 | Technology |
| Software IGV | Growth and technology | 4 | +0.75 | Technology |
| Technology Sector XLK | Growth and technology | 4 | +0.75 | Technology |
| US Large-Cap Growth IWF | Growth and technology | 4 | +0.70 | Technology |
| US Momentum Equities MTUM | Growth and technology | 4 | +0.65 | None |
| Brazil Equities EWZ | International equity | 5 | +0.80 | None |
| China Equities MCHI | International equity | 5 | +0.75 | None |
| Emerging Markets VWO | International equity | 5 | +0.70 | None |
| India Equities INDA | International equity | 5 | +0.75 | None |
| South Africa Equities EZA | International equity | 5 | +0.75 | None |
| South Korea Equities EWY | International equity | 5 | +0.80 | None |
| Taiwan Equities EWT | International equity | 5 | +0.85 | None |
| Australia Equities EWA | International equity | 4 | +0.45 | None |
| Canada Equities EWC | International equity | 4 | +0.45 | None |
| Developed Markets ex-US VEA | International equity | 4 | +0.45 | None |
| Europe Equities VGK | International equity | 4 | +0.45 | None |
| Japan Equities EWJ | International equity | 4 | +0.45 | None |
| Mexico Equities EWW | International equity | 4 | +0.55 | None |
| United Kingdom Equities EWU | International equity | 4 | +0.40 | None |
| Japanese Yen FXY | Cash and defensive FX | 4 | -0.45 | None |
| Euro FXE | Cash and defensive FX | 3 | -0.10 | None |
| US Dollar UUP | Cash and defensive FX | 3 | -0.40 | None |
| Short-Term Treasury Bills BIL | Cash and defensive FX | 2 | -0.90 | Defensive |
| Cash / Do Not Invest CASH | Cash and defensive FX | 1 | -1.00 | Defensive |
| Crude Oil USO | Real assets and inflation | 5 | +0.30 | None |
| Metals and Mining XME | Real assets and inflation | 5 | +0.55 | None |
| Agriculture Commodities DBA | Real assets and inflation | 4 | +0.10 | None |
| Broad Commodities PDBC | Real assets and inflation | 4 | +0.20 | None |
| Copper CPER | Real assets and inflation | 4 | +0.50 | None |
| Energy Sector XLE | Real assets and inflation | 4 | +0.35 | None |
| Silver SLV | Real assets and inflation | 4 | +0.05 | None |
| Gold IAU | Real assets and inflation | 3 | -0.25 | Defensive |
| Materials Sector XLB | Real assets and inflation | 3 | +0.45 | None |
It is a descriptive aggregate of a small model panel operating under CapitalBench prompts. It is not a forecast of market direction, a volatility estimate, a measure of investor positioning, or investment advice. Asset loadings are methodology assumptions and are versioned when changed.