YearBull is built to identify the quality of crypto market strength—not price movement alone. It compares thousands of assets through one regime-aware system that combines relative performance, liquidity support, drawdown resilience, stress, and history confidence.
1. Purpose and scope
The system turns fragmented crypto observations into comparable daily decisions across a broad universe. It answers two questions that price tables and market-cap lists do not: “Which eligible assets show the strongest combination of performance quality, liquidity, resilience, and risk?” and “Is the wider market move broad and supported—or concentrated and fragile?” The value is not another isolated score; it is a consistent way to separate movement from the quality and context behind it.
The five public outputs have distinct jobs:
- YearBull Rank orders the eligible analytical universe. A lower number means stronger relative positioning on the completed observation date.
- Bull Score summarizes asset-level relative strength and trend quality on a 0–100 scale.
- YB Market Risk classifies observed market stress and relative volatility. It is not an investment-safety, solvency, or fraud rating.
- Cycle labels the asset’s observed state as Early, Mid, or Late.
- Market Bull Index summarizes market-wide trend, breadth, risk, liquidity, and structure on a 0–100 scale.
2. Observation model and data flow
The unit of analysis is a dated asset observation, not a live quote. External market fields are ingested into controlled local snapshots, checked and normalized, and then used to calculate YearBull outputs. The completed daily table is the source of truth for price, market capitalization, reported volume, Bull Score, YB Market Risk, Cycle, and YearBull Rank. Public pages read from that table or its cache; opening a profile does not trigger a vendor refresh.
- Fetch and cache the permitted market and global snapshots.
- Write the completed daily asset observations.
- Calculate asset-level relative strength, trend, drawdown, liquidity quality, market-relative Bull, final risk, and special classifications.
- Calculate YearBull Rank only after those fields are final.
- Calculate the Market Bull Index from the completed ranked universe.
- Publish the same dated state to profiles, tables, charts, and APIs.
Input ownership, freshness, and attribution are documented separately in Data Sources & Metric Ownership. Missing data remains missing; the public layer does not silently replace an unavailable field with a live request.
3. Universe construction
3.1 Candidate observations
A candidate must resolve to a known asset identity and have a completed daily observation. An ordinary analytical rank additionally requires a usable price and sufficient supported inputs for a comparable score. Market capitalization and reported volume are validated independently and may be unavailable even when a price exists.
3.2 Separate analytical states
| State | Purpose | Public interpretation |
|---|---|---|
| Analytical | Comparable growth and market-structure assets | Receives an ordinary sequential YearBull Rank |
| Infrastructure | Wrapped, bridged, staked, or operational representations whose behavior is not comparable with an independent growth asset | Shown as Infrastructure rather than forced into the ordinary sequence |
| Stable or pegged | Assets designed to track a reference value, identified through explicit classification and conservative price/identity checks | Shown as a stable state, not a bullish growth rank |
| Excluded / not ranked | Insufficient, stale, frozen, internally inconsistent, duplicate, or otherwise non-comparable observations | Preserved as Not ranked rather than given false precision |
These states are intentionally distinct and are never collapsed into one continuous ranking. Assets with short history may remain eligible, but their score is progressively discounted until sufficient daily evidence accumulates. Liquidity is a scored constraint and risk input, not a blanket popularity filter; thin assets can remain visible while receiving a weaker assessment.
3.3 Market-reference cohorts
Market-relative calculations use liquid, higher-cap observations with valid comparison windows. Stable assets and derivative-like representations are removed from the reference cohort so they do not define normal growth-asset behavior. Market-wide components use minimum coverage rules; if the preferred breadth or trend sample is unavailable, the component is marked limited or uses a documented conservative fallback rather than fabricated history.
4. Metrics and contribution bands
Contribution bands describe relative influence, not published coefficients. They deliberately do not sum to a fixed formula because factor availability, history, market regime, penalties, and guardrails can change the effective contribution for an individual observation.
4.1 Bull Score
| Factor family | Band | Role |
|---|---|---|
| Risk-adjusted trend and relative strength | Primary | Measures directional persistence and performance relative to the market reference. |
| Drawdown resilience | Material | Distinguishes orderly strength from an asset still deeply below its relevant peak. |
| Liquidity quality | Supporting | Rewards movement supported by usable turnover rather than price action alone. |
| Overheating control | Guardrail | Prevents very fast, volatile appreciation from being interpreted as uniformly healthy strength. |
| Market-relative overlay | Supporting, bounded | Adjusts the result by the asset’s recent return relative to a filtered market cohort. |
4.2 YB Market Risk and Cycle
YB Market Risk combines short-horizon acceleration, realized stress, broader cycle conditions, and the interaction between unusually high strength and instability. Downside movement can relieve an overheated state without making an asset intrinsically safe. The public Low, Medium, and High labels are relative analytical states.
Cycle maps the asset’s normalized market heat into Early, Mid, or Late. It is a state label, not a forecast that a peak or reversal must occur. The cycle state also changes how the ranking model balances opportunity and durability: earlier regimes allow more influence from growth factors, while later regimes place more emphasis on resilience and penalties.
4.3 YearBull Rank
| Factor family | Band | What it contributes |
|---|---|---|
| Multi-horizon momentum | Primary | Direction and persistence across recent horizons. |
| Liquidity and volume quality | Material | Whether observed movement is supported by credible market activity. |
| Market-relative alpha | Material | Performance relative to the reference asset and broader market conditions. |
| Scale and trend quality | Material | Market depth context and consistency of the prevailing trend. |
| Stability and drawdown resilience | Supporting | Durability of the path, not merely the endpoint return. |
| Heat, stress, crash, and supply uncertainty | Guardrail | Downside adjustments that can outweigh positive factors when evidence is extreme. |
| Observation-history confidence | Guardrail | Discounts short histories progressively; full confidence is not granted immediately. |
The composite is bounded before ordering. Ties are resolved deterministically using risk, drawdown, heat, liquidity, and a stable identity key. The result is therefore reproducible inside the production system without publishing the proprietary equation.
4.4 Market Bull Index
The Market Bull Index combines five market-level families: trend, breadth, risk, liquidity, and structure. Trend and breadth carry the greatest influence; risk is material; liquidity and structure provide confirmation and fragility checks. The output is classified into Bearish, Neutral, Early Bull, Bull Market, or Late Bull. Coverage is stored with the observation because a reading supported by a broad universe is more informative than one built from a narrow subset.
5. Sample size and coverage
The figures below are a frozen disclosure snapshot, not a live counter. This makes the document auditable even as the production database continues to grow.
| Measure | Snapshot | Coverage / meaning |
|---|---|---|
| Observation date | 2026-09-22 | Latest completed daily asset snapshot used for this paper |
| Daily asset rows | 15,006 | Candidate observations in the completed snapshot |
| Positive price | 14,997 | 99.9% of daily rows |
| Positive market capitalization | 14,975 | 99.8% of daily rows |
| Positive reported volume | 14,766 | 98.4% of daily rows |
| Ordinary analytical ranks | 9,520 | 63.4% of daily rows |
| Separate states | 69 infrastructure; 432 stable; 4,985 excluded | Preserved outside the ordinary rank sequence |
| Daily-history archive | 2,100,975 rows across 266 dates | 18,120 distinct source identifiers since 2025-12-19 |
| Market-index archive | 265 observations | Stored since 2025-12-19 |
| Reference-asset price archive | 605 observations | Stored since 2025-01-25 |
| Published coin profiles | 13,399 | Profiles and daily source observations are related but not identical populations |
Coverage for a specific metric may be lower than price coverage because each field has its own validity rules and lookback requirement. Publishing those denominators makes a YearBull result auditable: readers can see how much of the available universe actually supports the conclusion.
6. Validation status and sensitivity
6.1 What the current evidence shows
YearBull keeps dated production observations so its conclusions can be checked against the state that was actually published. In the current seven-date retrospective comparison, the YearBull Rank top decile finished ahead of the market-cap and 30-day momentum top deciles on median 30-day forward return and positive-return rate. The low-volatility baseline was stronger, which identifies a real benchmark for the next test rather than reducing YearBull to another size or momentum list.
The next proof standard is prospective. A protocol beginning September 22, 2026 was locked before its forward outcomes existed and is published in Validation, Limitations & Failure Cases. Until that window matures, YearBull presents the retrospective result as evidence of useful differentiation—not as a completed claim of alpha or predictive superiority.
6.2 Sensitivity framework
The observed ranking is stable enough to be interpretable without becoming inert: across the latest 38 completed dates, mean one-day Spearman rank correlation was 0.923 with 77.5% top-decile overlap; at seven days those readings were 0.716 and 52.8%. This is the intended balance—high day-to-day continuity with meaningful adaptation as the market changes.
A disclosed-factor diagnostic also removed one public factor family at a time. Rank correlation with the full public proxy remained between 0.902 and 1.000, while top-decile overlap ranged from 51.2% to 98.6%. These tests expose factor dependence without revealing the production equation. Full definitions, baseline results, and the locked out-of-sample design are published in the companion validation document.
7. Known limitations and failure modes
- Vendor and venue limitations: reported prices, volume, supply, and venue fields can be delayed, revised, incomplete, or affected by wash trading and fragmented liquidity.
- Identity and representation risk: tickers can collide; bridged, wrapped, staked, and migrated assets can be mapped incorrectly unless contract and project identity are resolved.
- Short-history uncertainty: new assets lack full lookback windows. The confidence discount reduces false precision but cannot create missing evidence.
- Regime dependence: relationships observed in a trending market can weaken or reverse in a range-bound or stressed market.
- Market-cap and supply uncertainty: circulating-supply estimates may be incomplete, making scale and dilution-related inputs less reliable.
- Liquidity illusion: reported volume is not the same as executable depth; a high number can coexist with concentration, wide spreads, or restricted withdrawals.
- Model drift: changes in market microstructure, stablecoin use, derivatives, token design, or data coverage can alter how factors behave. Drift is monitored through coverage, distribution, cohort, and classification changes.
- Classification error: rule-based stable, infrastructure, or exclusion states can be wrong at the margin. Confirmed errors follow the published corrections process.
- Retrospective bias: selected examples can overstate clarity. They must not be interpreted as a representative performance sample.
Extreme missingness, frozen observations, invalid market fields, or internally inconsistent state transitions are preserved as unavailable or not ranked when a comparable score would be misleading. Material defects or methodology changes are recorded in the public changelog.
8. Governance, reproducibility, and interpretation
Each public observation is tied to a date and a defined source-of-truth table. The calculation order is fixed so that YearBull Rank uses finalized asset metrics and the Market Bull Index uses the completed ranked universe. Material changes to calculations, classifications, coverage, or reader interpretation are versioned in the Methodology Changelog.
Independent readers can reproduce the meaning and audit trail of a published output—its date, eligible state, component families, coverage, limitations, and change history—but cannot reconstruct the proprietary coefficients or anti-gaming logic from this disclosure. This boundary is intentional.
Related documents and downloads: Metric definitions, Data Sources, Provenance & Freshness, Validation, Limitations & Failure Cases, Retrospective Research Examples, Methodology Changelog, Corrections Policy, and the downloadable data dictionary (CSV).