Research & method
Inspired by complex systems.
Accountable to measurement.
Our starting point is a risk question: what changes beneath an index when the options on its constituents reprice? That question shaped the analytical map we maintain: measurements of event risk, surface shape and market conditions, linked across names and through time.
The SOC foundation
A research lens.
A testable set of measurements.
Self-organized criticality (SOC) and related complex-systems research motivate questions about collective behaviour: how broadly stress is priced, whether names reprice together, and how downside skew responds as markets fall and rebound.
The theory inspires the questions. Definitions, construction checks and empirical comparisons support the readings. Each measurement has to earn its interpretation; a physics analogy alone cannot establish a market mechanism or a future outcome.
The analytical process
How we maintain
the analytical record.
Source and align
Use licensed market data, align observation times and conventions, and retain explicit coverage and missingness.
Construct and reconcile
Check new aggregates against an independently coded reconstruction from retained primitives. Reconcile values and masks, and identify lossy transformations.
Compare consistently
State the historical window, membership, weights and denominators. Distinguish historical reconstructions from records retained at the time.
Maintain and document
Update the record at the agreed cadence. Check coverage, document method changes and identify revisions so comparisons remain interpretable.
Two different historical questions
Comparable history.
The record as published.
A reconstructed history applies a stated method to past observations. It helps compare periods on a consistent basis. A publication record preserves what was actually reported at the time.
These answer different questions. Our records identify their method and vintage, and distinguish reconstructed or restated series from retained publications. An as-published comparison is available only where that record was retained.
One measurement from the research programme
Inspect a dated fragility reading.
Fragility summarises the measured asymmetry of downside-skew response across falls and rebounds. Its percentile places that observation against prior history. This retained example describes the displayed session; it is not a live reading or a probability of a future decline.
A measurement, its date and its history.
This is the discipline behind the wider service: inspect the quantity, its comparison set and its limits before interpreting the result.
Read the construction and limitations →Fragility percentile
90th
percentile, against its own history
HIGH
US large-cap equities · Historical market session · 21 Oct 2026
View dated readings
| Session | Percentile |
|---|---|
| 2026-10-21 | 90.08 |
| 2026-10-20 | 92.46 |
| 2026-10-19 | 93.39 |
| 2026-10-16 | 91.01 |
| 2026-10-15 | 50.79 |
| 2026-10-14 | 52.65 |
| 2026-10-13 | 34.13 |
| 2026-10-12 | 34.79 |
| 2026-10-09 | 34.39 |
| 2026-10-08 | 30.56 |
| 2026-10-07 | 28.04 |
| 2026-10-06 | 30.16 |
| 2026-10-05 | 30.16 |
| 2026-10-02 | 29.89 |
| 2026-10-01 | 28.84 |
| 2026-09-30 | 29.23 |
| 2026-09-29 | 30.16 |
| 2026-09-28 | 30.16 |
| 2026-09-25 | 20.11 |
| 2026-09-24 | 20.11 |
| 2026-09-23 | 19.44 |
| 2026-09-22 | 20.77 |
| 2026-09-21 | 20.37 |
| 2026-09-18 | 18.52 |
| 2026-09-17 | 20.24 |
| 2026-09-16 | 21.69 |
| 2026-09-15 | 15.61 |
| 2026-09-14 | 19.58 |
| 2026-09-11 | 18.12 |
| 2026-09-10 | 11.77 |
| 2026-09-09 | 11.51 |
| 2026-09-08 | 11.64 |
| 2026-09-04 | 26.19 |
| 2026-09-03 | 27.78 |
| 2026-09-02 | 26.46 |
| 2026-09-01 | 25.13 |
| 2026-08-31 | 18.78 |
| 2026-08-28 | 18.25 |
| 2026-08-27 | 19.18 |
| 2026-08-26 | 18.92 |
| 2026-08-25 | 18.92 |
| 2026-08-24 | 18.92 |
| 2026-08-21 | 17.20 |
| 2026-08-20 | 17.46 |
| 2026-08-19 | 14.95 |
| 2026-08-18 | 14.15 |
| 2026-08-17 | 10.05 |
| 2026-08-14 | 6.61 |
| 2026-08-13 | 9.26 |
| 2026-08-12 | 10.19 |
| 2026-08-11 | 10.05 |
| 2026-08-10 | 7.01 |
| 2026-08-07 | 6.61 |
| 2026-08-06 | 6.48 |
| 2026-08-05 | 6.35 |
| 2026-08-04 | 10.05 |
| 2026-08-03 | 13.49 |
| 2026-07-31 | 16.80 |
| 2026-07-30 | 16.40 |
| 2026-07-29 | 11.90 |
| 2026-07-28 | 40.74 |
| 2026-07-27 | 39.68 |
| 2026-07-24 | 39.68 |
Historical snapshot from Altus production · checked 22 Oct 2026 · history reflects the stored research vintage and may be restated.
Data and interpretation
Clear about what
the evidence supports.
Altus uses licensed market data with rights to publish derived analytics. We supply those measurements and comparisons, rather than raw exchange quotes. Altus is responsible for its own methods and interpretations.
Market-implied quantities reflect the pricing of risk. They are not automatically real-world probabilities. Event-variance estimates depend on a baseline; historical percentiles depend on a reference set; sparse observations limit what can be inferred.
Descriptive accuracy, predictive information and economic usefulness are separate tests. Our research records limitations and unsuccessful comparisons alongside its findings.
Examine the research →See it on a question you already ask.
We’ll inspect the relevant output together and discuss whether an evaluation would fit your research needs.