An environmental intelligence platform that helps traders read prevailing market conditions—and the AI-assisted research system built to keep its interpretations honest.
Financial markets are complex environments. Volatility, credit, interest rates, equities, currencies, and commodities interact, reinforce one another, and sometimes contradict one another.
Market Weather was built for traders who need to read those relationships as a single environment rather than a wall of indicators. It brings together volatility, credit, rates, liquidity, and related market signals to produce a coherent picture of the prevailing market weather: what conditions exist now, how they are changing, and what they mean for a particular trading approach.
Its Forecast Center translates market data into environmental states, tracks how those states change, highlights developing transitions, and allows earlier conditions to be revisited.
It is not a trade-prediction system, a signal generator, or a buy/sell recommendation engine. Behind the dashboard, a collaborative research system investigates what changing conditions mean, challenges existing assumptions, and determines which findings deserve to become lasting institutional knowledge.
2026 · Environmental Intelligence & AI-Assisted Product Development · Working Prototype
The Forecast Center reading Clear Skies in an Expansion regime at a composite score of 6.0/8, with the data-freshness indicator visible.
Market Weather at a glance
9
market signals
8
environmental frameworks
6
market-participant archetypes
12
environmental change types
5
weather states
4
market regimes
Verified implementation inventory. These are system capabilities and classifications, not validated predictive models.
Andrew’s role
Product leadership, research direction, and the development of institutional intelligence.
I conceived and directed Market Weather, establishing the product vision, the user experience and its priorities, research questions, implementation requirements, evidence standards, and the governance boundaries the system had to respect.
I coordinated AI-assisted development across the Forecast Center, market-data processing, environmental interpretation, historical tracking, replay, and research systems.
I also directed the collaborative development of the Brain’s research behavior: challenging assumptions, evaluating competing interpretations, refining terminology, determining what evidence was sufficient, and deciding which changes deserved acceptance or further investigation.
That work required more than coordinating implementation tasks. It required maintaining coherence between a functioning product, an evolving conceptual model, and the records the system was authorized to change.
AI collaborators contributed substantially to implementation, research, and architecture. I do not claim to have personally written the code or originated every technical mechanism, and I do not claim sole architectural authorship.
My contribution was directing a complex product and learning system while keeping its purpose, evidence standards, user experience, and institutional knowledge aligned.
THE PRODUCT
The Forecast Center
Understand the environment—not just the latest market move.
The Forecast Center is a functioning interactive Streamlit dashboard designed to explain the prevailing market environment in accessible language.
Instead of presenting isolated indicators and expecting the user to assemble their meaning, Market Weather organizes information into a coherent environmental assessment.
The dashboard answers eight practical questions.
01
What environment are we in?
A composite regime assessment and a named weather state summarize prevailing conditions. The engine classifies the environment as Clear Skies, Crosswinds, Overcast, Turbulence, or Severe Storm, and distinguishes its underlying regime score from that human-readable classification.
02
What’s driving the environment?
Active Market Frameworks identify relevant market structures, and the Weather Report translates conditions into a readable explanation. The system also calculates how much each monitored condition contributes to the composite assessment, ranks those contributions, and names the strongest current influence. That attribution describes the current assessment; it does not attribute recorded transitions to individual instruments.
03
Is the environment changing?
Environmental Transitions identify developing patterns such as deterioration, stabilization, escalation, and fragility. They describe the direction and character of change rather than merely reporting a new score.
04
What changed since the last observation?
The What Changed layer identifies shifts in monitored conditions, frameworks, weather states, and other environmental features.
05
Which developments matter most?
Environmental Significance ranks detected changes by their contextual importance, helping distinguish consequential developments from routine movement.
06
Are warning conditions emerging?
Storm Alerts and severity overlays identify elevated environmental stress. Warning overlays are separate from the underlying weather state and regime assessment.
07
How does this environment affect different traders?
Environment Compatibility interprets conditions through six participant archetypes—High Beta, Commodity, Defensive, Bond / Rates, Small Cap, and Crypto—because the same conditions carry different consequences depending on exposure, risk tolerance, and time horizon.
08
What did the environment look like before?
Score History, persistent environmental records, and Replay Mode allow previous conditions to be examined rather than relying exclusively on the latest dashboard reading.
The Forecast Center is not a trade generator or price-prediction engine. It is a system for understanding market conditions, their drivers, and their evolution.
The intelligence model
State → Transition → Change → Significance
A static market score can describe current conditions. It cannot, by itself, explain how an environment is developing.
Market Weather separates four distinct forms of intelligence.
STATE → TRANSITION → CHANGE → SIGNIFICANCE
01 — STATE
What conditions exist?
Market signals, composite scoring, regime classifications, weather states, frameworks, and overlays describe the prevailing environment.
A regime score is not interchangeable with a weather state. An elevated warning overlay does not automatically redefine the underlying environment.
02 — TRANSITION
Where are conditions heading?
The transition engine evaluates how the environment is evolving.
It can identify developing deterioration, stabilization, escalation, or fragility even when the current overall state has not yet changed.
This matters because two environments with similar scores may be moving in very different directions.
03 — CHANGE
What is different from the previous observation?
The What Changed system compares environmental records to identify meaningful shifts.
A framework may activate or weaken. A weather state may change. A new warning condition may appear.
The system makes those developments explicit rather than forcing the user to reconstruct them manually.
04 — SIGNIFICANCE
Which changes deserve attention?
The significance layer evaluates the relative importance of detected developments.
Not every change is equally consequential. Market Weather attempts to direct attention toward the developments that most affect the environmental interpretation.
Together, these capabilities describe not only what the market looks like, but how it is changing and which changes deserve attention.
From market signals to environmental intelligence
A layered interpretation pipeline.
Market Weather tracks nine market signals built from public data: VIX and VVIX for volatility and hedging demand, VIX3M for term structure, HYG for credit, Treasury futures for rates, S&P and Nasdaq futures for equities, oil for commodity pressure, and the dollar index for currency. Eight of the nine carry weight in the composite assessment; the dollar is tracked for context at zero weight.
Those inputs move through a structured processing system:
MARKET DATA → SIGNALS → OVERLAYS → FRAMEWORKS → WEATHER STATES → TRANSITIONS → WHAT CHANGED → SIGNIFICANCE → HISTORY → FORECAST CENTER
Each layer has a different responsibility.
Signals measure conditions. Frameworks identify broader market structures. Regimes and weather states summarize the environment. Overlays communicate exceptional stress. Transitions and change detection describe evolution. Significance ranking prioritizes developments.
The output is not merely a generated paragraph. It is a set of structured environmental interpretations supported by underlying records.
The dashboard then translates those interpretations into information a human can navigate and understand.
The environment has a memory
Market Weather tracks conditions over time.
A dashboard that only displays the latest observation loses much of the information needed to understand change.
Market Weather maintains separate historical records for its principal environmental layers:
Signal history — underlying calculated conditions.
Score history — changes in composite environmental assessment.
Weather-state history — recorded environmental classifications.
Framework history — activation and evolution of market structures.
Overlay history — elevated warning conditions.
Transition history — developing environmental changes.
What Changed history — detected differences between observations.
Significance history — ranked environmental developments.
These records preserve distinctions between what was measured, how it was classified, and how the interpretation evolved.
The verified local implementation uses inspectable, append-oriented records. Seven of these records are populated in the verified snapshot; significance history is implemented with a defined schema, reader, and dashboard support, but is not populated in that snapshot. Their existence does not imply years of uninterrupted live operation.
The product was designed to preserve environmental history, not simply overwrite yesterday with today’s interpretation.
Replay Mode
Revisit an earlier market environment.
Replay Mode allows the Forecast Center to examine historical environmental snapshots.
Instead of inventing a separate conceptual interface for the past, the dashboard can substitute persisted historical data for current inputs and present the earlier environment through familiar components.
An operator can examine recorded weather states, frameworks, transitions, and changes in their historical context.
LIVE ENVIRONMENT → RECORDED SNAPSHOT → HISTORICAL INTERPRETATION
Replay is intended to reconstruct what the system recorded, not creatively regenerate history.
This is historical environmental inspection—not proof of forecasting accuracy, profitable trading strategies, or validated backtesting.
Designed to make complexity understandable
Technical intelligence needs a human interface.
Market Weather uses recognizable weather-language metaphors, environmental states, named market structures, explanatory reporting, and progressive disclosure.
The design separates the overall environmental assessment from the evidence that supports it.
It also distinguishes three questions:
What is happening in the market?
How might those conditions matter to different participants?
What, if anything, should an individual decide to do?
The first two are within Market Weather’s environmental-intelligence purpose. The third remains a human decision.
Product development included work on the Forecast Center’s information hierarchy, terminology, readability, data freshness, trust indicators, and the relationship between headline intelligence and supporting detail.
The dashboard answers the second question directly: a trader selects their own archetype, and the environment is read back for that exposure.
SEVERE STORM · TWO PARTICIPANTS
High Beta — “Liquidity is deteriorating rapidly. High Beta assets historically struggle during this environment. Capital preservation becomes increasingly important.”
Bond / Rates — “High-quality duration assets may benefit from defensive capital flows.”
Both lines are the product’s own interpretation copy. Forty-two entries cover six archetypes across seven named environments.
Some limits are worth stating plainly. The interpretation layer covers seven named environments, but the current engine emits five of them, and Overcast has no archetype entry yet. The Environment Panel’s driver readings are static mappings rather than live calculated attribution, and the Forecast Center’s hero transition line is drawn from a placeholder rather than recorded transition history. Some proposed dashboard surfaces remain scaffolding; the implemented product should not be confused with every feature envisioned in its design documents.
THE INTELLIGENCE BEHIND THE PRODUCT
How the Brain learned to change its mind
The research system did more than produce market commentary.
The Forecast Center could describe the environment.
The harder challenge was developing a Brain that could reconsider what an environment meant, examine competing explanations, refine its own concepts, and preserve the reasoning behind those changes.
Market Weather developed a collaborative research process involving AI analysis, Brain review, human-directed challenge, adjudicated reflection, and reconciliation with persistent institutional state.
The objective was not simply to produce a more persuasive daily report.
It was to make the next investigation more informed because the previous investigation happened.
A research process with consequences
New evidence could change the system’s understanding.
A market observation might initially support a tentative explanation.
The AI analyst would examine relevant signals, identify supporting and contradicting evidence, and propose observations, hypotheses, or research questions.
The coordinating Brain would compare those proposals with existing research objects, earlier interpretations, unresolved disagreements, and historical evidence.
The collaborative process then asked:
Does this strengthen or weaken an existing hypothesis?
Is the apparent new pattern actually a continuation of something already being tracked?
Are two different mechanisms being incorrectly treated as one?
Does the evidence justify a new concept, or only more precise terminology?
What evidence would confirm or invalidate the interpretation?
What should be retained, challenged, researched further, or left unresolved?
The answers could lead to revised hypotheses, terminology candidates, replay requirements, research assignments, or explicit decisions not to promote a conclusion.
The Brain was not merely accumulating information. It was developing a more disciplined way to interpret information.
The collaboration, reflection, and reconciliation process
Multiple perspectives. One accountable institutional record.
SIX-STAGE COLLABORATION CYCLE
ANALYZE → COMPARE → REFLECT → COLLABORATE → ADJUDICATE → RECONCILE
01 — ANALYZE
The AI analyst examines the market environment, develops interpretations, identifies contradictions, and proposes research questions.
02 — COMPARE
The Brain evaluates those interpretations against prior observations, hypotheses, research candidates, terminology, and unresolved evidence requirements.
03 — REFLECT
The collaborators identify competing explanations, assumptions that may no longer hold, gaps in evidence, and possible changes in conceptual framing.
04 — COLLABORATE
The interpretations are challenged and refined. A hypothesis may be narrowed, a mechanism split, a label questioned, or additional evidence requested.
05 — ADJUDICATE
Human-directed review determines what the evidence supports, what remains provisional, and which proposed changes should be accepted, rejected, or deferred.
06 — RECONCILE
Accepted decisions are translated into specific proposed changes to research objects, terminology records, replay requirements, assignments, and other institutional state.
Required artifacts, readiness conditions, and verification determine whether a proposed change may actually be committed.
The governing distinction: an improved interpretation does not automatically authorize a change to institutional memory.
A concrete example: when the Brain’s vocabulary evolved
A market recovery became a more precise research concept.
During a multi-session sequence in June 2026, the research system was examining conflicting market evidence.
Equities and semiconductor-related instruments showed signs of repair. But other signals—including crypto proxies, volatility and hedging conditions, and company-specific catalysts affecting apparent leadership—complicated the interpretation.
A simple description suggested a growth or semiconductor recovery attempt.
The collaborative reflection process produced a more precise framing.
EARLIER INTERPRETATION
“Growth/semi repair test”
REFINED INTERPRETATION
“Compression Path — Tentative Repair Test Phase”
This was not merely a writing improvement.
The revised framing preserved continuity with an earlier compression sequence. It treated the apparent repair as a developing phase of an existing environmental pattern rather than prematurely declaring a new, confirmed market condition.
It also preserved the contradictions.
What changed: The working interpretation became more precise, better connected to earlier observations, and more useful for subsequent research.
What remained unresolved: Whether the apparent repair was durable and whether the new terminology deserved formal promotion.
“Tentative Repair Test” remained candidate terminology requiring additional confirmation.
The Brain developed a better explanation without pretending the evidence justified greater certainty.
Changing understanding without corrupting institutional state
Meaning can evolve without overwriting the record.
One of Market Weather’s most consequential distinctions was between analytical interpretation and formal institutional state.
A collaborative reflection could produce a better understanding of the market while existing research objects retained their governed lifecycle statuses.
The June 25 reflection provides a concrete example.
It produced changes affecting observations, research candidates, replay branches, and terminology records. Seventeen of twenty-two proposed actions were executed and independently verified; five were deferred. The executed changes touched ten files, eight of them registries.
But every one of those changes was recorded at observational or candidate status. The adjudicated position on promotion remained deliberately restrained:
0
confirmed intelligence promotions
0
formal demotions
0
Research Object promotions
0
cluster-registry mutations
0
Intelligence Memory writes
Those results did not mean the research produced no learning.
The learning appeared in refined interpretations, clarified disagreements, new evidence requirements, and proposed future research.
What the process refused to do was treat intellectual progress as automatic permission to promote a conclusion to confirmed intelligence.
This was the deeper achievement: the system could change how it understood something while preserving the authority, provenance, and verification requirements governing what it officially knew.
What the collaboration produced
Better concepts. Better questions. Better research.
Persistent disagreement
Contradictory signals and unresolved assumptions could remain visible rather than being smoothed into a confident narrative.
New research assignments
Disputes could produce concrete evidence requirements, Brain Homework, and follow-up investigations.
Continuity across sessions
New observations could be compared with earlier research, control-variable hypotheses, and unresolved questions.
Consequential human direction
Andrew could challenge interpretations, require stronger evidence, accept a refined framing, reject premature promotion, or redirect research.
This was not simply multiple AI models editing one another’s reports.
It was an effort to make AI-assisted investigation cumulative, challengeable, and governed.
From conversation to institutional memory
Learning should survive the conversation that produced it.
Market Weather separated temporary reasoning from accepted institutional knowledge.
Its research lifecycle connected:
OBSERVATION → RESEARCH → FINGERPRINT → REPLAY → REPLAY REVIEW → INTELLIGENCE MEMORY
Observations became research questions.
Fingerprints preserved interpretations, supporting and contradicting evidence, and invalidation conditions.
Replay and review evaluated claims against historical conditions.
Qualified findings could then be retained with provenance, limitations, and links to earlier work.
Collaborative reflection and state reconciliation operated alongside that lifecycle, determining how emerging insights related to existing objects and whether proposed changes had earned formal acceptance.
A useful insight might remain a research candidate, a terminology proposal, a replay requirement, or an unresolved question rather than becoming trusted intelligence.
The goal was cumulative institutional learning without premature certainty.
IMPLEMENTATION EVIDENCE
Technical evidence
A functioning application supported by substantial research infrastructure.
Forecast Center
Interactive Streamlit dashboard with environmental interpretation, weather reporting, transitions, What Changed, significance, alerts, framework context, history, and Replay Mode.
Environmental intelligence
9 tracked signals (8 weighted in the composite), 8 frameworks, 6 participant archetypes, 12 change types, 5 weather states, and 4 regimes in the verified implementation inventory.
Historical persistence
Separate records for signals, scores, states, frameworks, overlays, transitions, changes, and significance.
Research lifecycle
Structured observation, research, fingerprint, replay, review, and qualified-memory processes.
Research artifacts
26 fingerprint objects and 11 completed replay reviews in the verified repository snapshot.
Product evaluation
A documented 14-panel dashboard audit, followed by implementation work that made dedicated Environmental Transitions and Environmental Significance panels available.
Repository scope
484 commits, 1,204 tracked files, 779 test functions, and 556 design documents on the verified research branch.
Collaborative learning
Documented reflection, terminology refinement, adjudication, and state-reconciliation controls, with explicit distinctions between proposed and committed changes.
These are measures of implementation and research activity, not evidence of validated predictive performance or commercial deployment.
What Market Weather demonstrated
A product that interprets environments. A research system that develops its understanding.
Market Weather demonstrated two complementary capabilities.
A working environmental-intelligence product
The Forecast Center organizes complex market information into states, transitions, changes, significance, historical context, and readable reporting.
It gives users a structured way to understand changing conditions without pretending to predict prices or prescribe trades.
A governed collaborative learning process
The Brain’s research system could compare new evidence with prior understanding, challenge assumptions, refine concepts, preserve disagreement, and distinguish analytical progress from authorized changes to institutional knowledge.
The combination is what makes Market Weather distinctive.
The dashboard makes the intelligence usable.
The research and reconciliation system makes the development of that intelligence inspectable and accountable.
Current maturity
Working local prototype · Research phase complete
Market Weather is not a commercially deployed product, validated forecasting service, or production trading system.
The verified research period was bounded. Some proposed dashboard surfaces remain scaffolding, and repository test counts should not be mistaken for proof of production readiness.
The collaborative research process was exercised and documented, but its long-term learning effectiveness has not been established through external evaluation.
The work nevertheless demonstrates substantial AI-assisted product development, environmental-intelligence architecture, user-experience direction, research operations, and governance of evolving institutional knowledge.
Explore Sentinel →
Explore Cairn / SIA →