AI Market State
As of Jul 10, 2026 ET Method 1.1.5
DEGRADEDDefinition
Risk Appetite Score
A 0–100 causal composite of State, Transition, Breadth and Fragility Health. Higher values indicate stronger, more broadly supported and healthier market structure relative to the same group’s own causal history.
- Weights: 30% State, 25% Transition, 25% Breadth and 20% Fragility Health.
- Each underlying feature is first converted to a causal 156-week rolling percentile, using at least 52 weekly observations and only information available at that date.
- Related features are averaged inside subgroups before the four dimensions are combined, reducing implicit double-counting.
- Regime is selected separately from the four-dimensional profile, so it is not a simple score bucket.
The score is not a return percentage, forecast probability, price target or trading recommendation. Cross-group comparisons should be read with care because each group is standardized against its own history.
How to read
Semiconductor: score and regime
37.2 is the current 0–100 Risk Appetite Score. Higher means the underlying market-state features are stronger and healthier relative to Semiconductor’s own causal history; lower means weaker, narrower or more fragile structure.
- Score = 30% State + 25% Transition + 25% Breadth + 20% Fragility Health after subgroup aggregation and causal percentile normalization.
- Late/Crowded: the four-dimensional profile is closest to the Late/Crowded prototype, which represents stronger State combined with weakening Transition, narrower Breadth and higher structural fragility.
- 63 names: approved formal members with an active security series and at least 63 available sessions at the cutoff. Longer-window features can have a smaller valid sample, tracked through coverage.
- 16% regime confidence: the distance separation between the nearest and second-nearest prototypes, calculated as (second distance − nearest distance) / second distance. It is not a forecast probability.
- If confidence is below 15%, the label adds Mixed because the nearest prototype is not sufficiently distinct from the runner-up.
The composite score and regime describe current market structure. They do not estimate expected return, downside probability or investment conviction.
How to read
AI Infrastructure: score and regime
37.0 is the current 0–100 Risk Appetite Score. Higher means the underlying market-state features are stronger and healthier relative to AI Infrastructure’s own causal history; lower means weaker, narrower or more fragile structure.
- Score = 30% State + 25% Transition + 25% Breadth + 20% Fragility Health after subgroup aggregation and causal percentile normalization.
- Risk Contraction / Mixed: the four-dimensional profile is closest to the Risk Contraction prototype, which represents low State, low Transition, narrow Breadth and weak Fragility Health.
- 33 names: approved formal members with an active security series and at least 63 available sessions at the cutoff. Longer-window features can have a smaller valid sample, tracked through coverage.
- 11% regime confidence: the distance separation between the nearest and second-nearest prototypes, calculated as (second distance − nearest distance) / second distance. It is not a forecast probability.
- If confidence is below 15%, the label adds Mixed because the nearest prototype is not sufficiently distinct from the runner-up.
The composite score and regime describe current market structure. They do not estimate expected return, downside probability or investment conviction.
How to read
Cloud & AI Platform: score and regime
36.9 is the current 0–100 Risk Appetite Score. Higher means the underlying market-state features are stronger and healthier relative to Cloud & AI Platform’s own causal history; lower means weaker, narrower or more fragile structure.
- Score = 30% State + 25% Transition + 25% Breadth + 20% Fragility Health after subgroup aggregation and causal percentile normalization.
- Risk Contraction: the four-dimensional profile is closest to the Risk Contraction prototype, which represents low State, low Transition, narrow Breadth and weak Fragility Health.
- 11 names: approved formal members with an active security series and at least 63 available sessions at the cutoff. Longer-window features can have a smaller valid sample, tracked through coverage.
- 29% regime confidence: the distance separation between the nearest and second-nearest prototypes, calculated as (second distance − nearest distance) / second distance. It is not a forecast probability.
- If confidence is below 15%, the label adds Mixed because the nearest prototype is not sufficiently distinct from the runner-up.
The composite score and regime describe current market structure. They do not estimate expected return, downside probability or investment conviction.
How to read
AI Software: score and regime
47.3 is the current 0–100 Risk Appetite Score. Higher means the underlying market-state features are stronger and healthier relative to AI Software’s own causal history; lower means weaker, narrower or more fragile structure.
- Score = 30% State + 25% Transition + 25% Breadth + 20% Fragility Health after subgroup aggregation and causal percentile normalization.
- Early Recovery: the four-dimensional profile is closest to the Early Recovery prototype, which represents still-moderate State with strong improvement, expanding Breadth and healthier structure.
- 17 names: approved formal members with an active security series and at least 63 available sessions at the cutoff. Longer-window features can have a smaller valid sample, tracked through coverage.
- 29% regime confidence: the distance separation between the nearest and second-nearest prototypes, calculated as (second distance − nearest distance) / second distance. It is not a forecast probability.
- If confidence is below 15%, the label adds Mixed because the nearest prototype is not sufficiently distinct from the runner-up.
The composite score and regime describe current market structure. They do not estimate expected return, downside probability or investment conviction.
Historical risk appetite
Five-year current-universe retrospective
How to read
Historical Risk Appetite
Each line shows the causal weekly Risk Appetite Score for one Section or Basket. The x-axis is the ET weekly session and the y-axis runs from 0 to 100.
- Rising values indicate improving market structure; falling values indicate deterioration.
- Hover over the chart to compare values recorded at the same weekly session.
The five-year view is a current-universe retrospective and retains disclosed survivorship and taxonomy look-ahead limitations.
Participation
Latest section structure
How to read
Participation
For each Section, Participation is the number of eligible constituents with a positive split-adjusted 21-session price return divided by the number with a valid 21-session return.
- The constituent return is P(t) / P(t−21 sessions) − 1; the calculation uses the active security series and ET exchange sessions.
- Eligible constituents must be approved for the formal score and have at least 63 observations. Longer-window features can have a smaller valid sample, which is tracked separately as coverage.
- A high value means gains are broadly shared across the Section.
- This is a raw current breadth diagnostic. The Breadth dimension also includes relative participation, equal-weight versus value-weight diffusion and Small-minus-Large evidence after causal normalization.
It measures the fraction of positive names, not the magnitude or durability of gains. Returns are price returns rather than total returns.
Share of constituents with a positive 21-session price return.
Macro risk
Context overlay, outside the score
How to read
Macro Risk Conditions
Latest observations available by the publication cutoff for volatility, rates, credit and financial conditions.
- The series follow their natural release calendars, so observation dates can differ.
- Macro variables provide context and do not modify the industry Risk Appetite Score in the current method.