Methodology — How Atlas Works in Detail
For educational and informational purposes only. Not investment advice. Specific parameters, weights, and formulas are proprietary and not disclosed. MAY — POTENTIAL — EDUCATIONAL.
This page describes the structure and philosophy of the Atlas framework in detail. It is intended for institutional prospects, professional allocators, and serious investors who want to understand how Atlas works before subscribing. We explain the approach and architecture; exact proprietary parameters that would allow duplication are intentionally not disclosed.
Section A: Data Sources and Coverage
Economic Data
The Atlas environment engine uses a curated set of publicly available economic series sourced from FRED (Federal Reserve Economic Data) and comparable public providers. These series span several broad domains, such as:
- Activity and growth (for example, employment, production, spending, and related indicators)
- Inflation dynamics (for example, price indices and market‑based inflation expectations)
- Credit and financial conditions (for example, spreads and curves)
- Sentiment (for example, household or business surveys)
All underlying economic series are publicly available at no cost. No proprietary or licensed economic data feeds are required for the framework.
Market Data
Atlas monitors 407 symbols across U.S. equities, sector ETFs, commodities, and major indices using daily price and volume data from public exchanges. The universe is designed to cover liquid, widely followed markets rather than thin or obscure instruments.
Volatility and Curve Data
Atlas also references standard volatility and curve measures such as:
- Equity volatility indices
- Treasury volatility indices
- Common yield‑curve spreads (for example, differences between key Treasury maturities)
These are sourced from public or widely used market data providers.
Section B: Environment Classification Engine
Philosophy
The environment framework is built on a straightforward idea: changes in growth, inflation, and related conditions matter for how markets behave. The aim is not to invent a new macro theory, but to codify a rules‑based way of describing the backdrop so it can be tracked consistently over time.
The conceptual approach — looking at how growth and inflation behave, together with other indicators — is standard in institutional macro work. What is specific to Atlas is the exact set of series, transformations, weights, thresholds, and presentation choices.
Rate‑of‑Change Focus
Raw levels of many economic series drift over time and depend on structural context. To make readings more comparable across eras, Atlas emphasizes changes rather than levels. Series are transformed into rate‑of‑change or similar derived measures over rolling windows. Exact window lengths and combinations are proprietary.
Environment Composites
For groups of related series (for example, activity‑related or inflation‑related), Atlas constructs internal composites. Each composite summarizes how that segment of the data is behaving under the framework — for example, whether recent changes are broadly improving, deteriorating, or mixed.
Composites are calculated using weighted combinations and smoothing techniques chosen to balance responsiveness with noise reduction. The specific construction details are proprietary.
Environment Labels
Based on how these composites and other inputs behave together, Atlas assigns an internal environment label. These labels are used inside the framework to organize research and communication. Examples include terms such as “Expansion,” “Acceleration,” “Stagflation,” or “Contraction,” which reflect how the model currently interprets the balance of growth and inflation‑related dynamics.
These labels are:
- Internal to the model (other practitioners may use similar words differently)
- Descriptive, not predictive
- Used to anchor discussion, not to promise performance
Any comments about typical asset behavior in a given label are educational observations drawn from history, not forecasts.
Confirmation and Confidence
Atlas calculates internal confirmation and confidence measures that indicate how many underlying indicators currently agree with the assigned environment label and how persistent similar configurations have been historically.
Higher internal confirmation generally means more indicators are aligned; lower confirmation means more dispersion. Confidence bands summarize how closely the current configuration resembles past instances under the framework. Exact thresholds and formulas are proprietary.
Directional Pressure
In addition to point‑in‑time labels, Atlas tracks whether the composites and other inputs are building pressure toward a different environment classification. This highlights when the underlying math is starting to move in ways that, historically, have been associated with eventual shifts in backdrop under the framework. Exact calculations are proprietary.
Section C: Symbol‑Level Condition Engine
Philosophy
The symbol‑level engine is designed to treat each symbol the same way, under a fixed ruleset, and to prioritize quality over quantity of conditions. It focuses on logging potential long and potential short conditions when multiple independent views of a symbol align, rather than trying to generate a constant stream of signals.
Independent Layers
Each symbol is evaluated across several internal layers that look at different aspects of its behavior. Examples include:
- Price structure over an intermediate horizon
- Rate of change and momentum characteristics
- Risk or volatility behavior and noise profile
- Participation and conviction measures
- How the symbol’s behavior relates to the currently published environment under the framework
Each layer has its own internal logic; layers do not tune themselves based on each other. A full condition requires agreement across the model’s predefined criteria.
Rolling Evaluation and Normalization
All symbol evaluations are performed on rolling windows based on daily data. Window lengths are chosen to reflect short‑to‑intermediate behavior without reacting to every tick. Outputs are normalized internally so that instruments with different volatility profiles can be evaluated on a comparable basis. Exact window lengths and normalization techniques are proprietary.
Condition Logging
When the framework detects that a symbol’s readings meet the internal criteria for a potential long or potential short condition, Atlas logs that as a time‑stamped observation. This log is what subscribers see as “conditions” or “alerts” inside Atlas.
Key points:
- Conditions are model classifications of the data, not trade instructions.
- The engine applies the same criteria to all symbols; it does not adjust based on news or opinion.
- If conditions are not met, nothing is logged; the framework does not force output.
Section D: Quality Controls and Model Risk
Data Handling
Automated checks are used to flag missing data, obvious outliers, stale feeds, and revision anomalies across economic and market inputs. When underlying data is revised by source providers (for example, economic agencies), Atlas updates historical readings to reflect the latest available information rather than freezing earlier, less accurate values.
Regime Coverage and Review
The framework is reviewed across a range of historical periods that include different market backdrops and stress episodes. These reviews are used to assess whether the logic behaves sensibly under varied conditions, not to guarantee any particular outcome.
Known Limitations
- Lag and revision: Many economic series are lagged and subject to future revision. Readings reflect the best available data at the time, not omniscient real‑time information.
- Model risk: No model can fully capture all drivers of markets, including policy decisions, geopolitical events, structural changes, and behavioral dynamics.
- Changing relationships: Historical relationships between economic conditions and asset behavior can change over time. What held in one era may not hold in another.
- Overfitting risk: Any systematic approach faces the risk of fitting to past noise. The framework is designed for robustness and simplicity over optimization, but overfitting risk cannot be eliminated.
What Atlas Does Not Claim
Atlas does not claim to predict future returns, guarantee outcomes, or identify “right” trades. It does not publish forward return estimates, model portfolio performance, or promises about hit rates.
Each output is a mathematical observation about how the framework is classifying data now, and how similar configurations have behaved historically under that same framework. What happens next is uncertain.
Section E: Governance, Conflicts, and IP
Framework Governance
Updates to the framework are made by the Given Analytics internal team and, when implemented, apply to all subscribers at the same time. Updates are made periodically and with an emphasis on robustness and clarity, not in reaction to individual conditions resolving well or poorly.
Conflicts of Interest
Given Analytics does not manage client capital, does not trade on behalf of subscribers, and does not receive compensation from issuers, ETF sponsors, or other market participants based on coverage or condition direction. Revenue is generated from subscription fees for access to the research platform.
Intellectual Property
The methodology described on this page reflects the structure and philosophy of the Atlas framework. Specific parameters, weights, thresholds, lookback windows, and detailed formulas are proprietary to Given Analytics, LLC and are not disclosed. This description is provided for transparency and educational purposes only.
For educational and informational purposes only. Not investment advice. Given Analytics is not a registered investment adviser. MAY — POTENTIAL — EDUCATIONAL. See our Educational Disclaimer for full disclosures.