From raw market data to usable intelligence.
Strata is a pipeline, not a dashboard. Feeds from four data domains are normalised onto one clock and one symbology, scored by five independent modules, and resolved into a single composite that means the same thing whether it describes an equity, a token or a pool.
How often every covered market is recomputed.
Wall-clock time of the most recent computation across all markets.
Same inputs and weights reproduce the same score.
Every computation keeps its full factor breakdown.
Each stage is independent and observable. A failure in one venue degrades coverage for that venue only; it never silently changes a score.
Four domains, one ingestion contract. Every record is stamped at the venue clock before it enters the pipeline.
Reference prices, corporate actions, venue calendars and halt state.
Consolidated tape, auction prints and extended-hours activity.
Spot and perpetual order books across every covered exchange.
Pool state, settlement events and wallet-level activity per block.
Five modules read the same normalised inputs and never read each other. Their weights are published and fixed for the life of a scoring version.
The composite is a weighted sum — nothing more exotic. That is deliberate: a score you cannot reconstruct by hand is a score you cannot trust.
| Module | Reading | Weight | Contribution |
|---|---|---|---|
| Momentum | 93.8 | 38% | 35.16 |
| Volume | 13.8 | 28% | 3.87 |
| Trend | 9.0 | 25% | 2.24 |
| Volatility | 6.4 |
Scores are compared against their own history. What has moved further than this asset usually moves, held long enough to mean it, and can be explained by the components underneath becomes an intelligence event.
An event is one condition, carried forward while it holds and closed when it stops. Fifteen passes over the same condition produce one event seen fifteen times — never fifteen events.
Rankings, arena rounds and signals are three views onto one computation — not three separate products.
Ordered standings across every class, sortable by any single factor.
The same scores resolved into competitive rounds with eliminations.
Threshold crossings emitted the moment a factor breaks its baseline.
5 stages · 5 modules · every figure on this page is read from the engine
Market, stock, crypto and onchain feeds are ingested continuously and stamped at the venue clock.
Heterogeneous feeds are resolved onto one symbology, one clock and one unit system before any maths runs.
Five independent modules score every market on the same normalised inputs. Modules never read each other.
Module outputs are weighted into a single composite that is comparable across stocks, crypto and onchain markets.
Scores resolve into ordered views — rankings, arena rounds and the signal feed.
Rate-of-change across three windows, each normalised by the market's own realised volatility so a 2% move in a quiet market outranks a 2% move in a violent one.
Traded notional measured against a 30-day rolling median, then down-weighted when volume concentrates on a single venue.
Breadth of participation rather than size of participation. Dispersion across many accounts scores higher than the same notional from few.
Depth within 50bps of mid, spread stability over the window, and modelled slippage on a standardised clip size.
Cross-sectional standing against the asset's own class, which is what makes a crypto 90 and an equity 90 mean the same thing.
| 9% |
| 0.60 |
| Strata Score | 41.9 | ||
Scores are cross-normalised within each asset class before the composite is formed, so an 88 on an onchain pool and an 88 on a large-cap equity describe the same relative strength.