Current research agenda
Three Current Hypotheses
Qadam is currently testing three foundational hypotheses. They are falsifiable research questions, not proven claims:
The Short Version
Qadam is a local-first macro intelligence and governed paper-trading system. An unattended Python operator coordinates source ingestion, pattern research, strategy formation, Akber review, shadow observation, portfolio governance, guarded paper execution, lifecycle polling, learning and public visibility.
Qadam is not a public financial-advice product, a signal channel, or a live-capital trading bot. The dashboard is an explanatory, public-safe, public read-only projection of Qadam's operating state.
Use the dashboard to answer five questions:
- What is Qadam observing, and which evidence is usable?
- Which relationships and strategies are being tested?
- What did Akber's filter and Qadam's decision governance conclude?
- Did a conclusion become a guarded Alpaca Paper order or position?
- What can Qadam legitimately learn, and may that lesson change future behaviour?
A blocked, held, or empty state often means Qadam's controls are working.
Whitepaper versus User Guide
Public Reading And Protected Member Access
Public read-only access
Anyone can inspect the sanitized paper portfolio, evidence flow, research, decision rationale, paper-order mirror, learning loop, team, and system health without signing in.
Protected member features
An authenticated member account is required for this guide and any enabled governance features. Signing in does not create trade, broker, shell, deployment, or capital authority.
The local operator is a separate role responsible for repository commands and configured services. Public visibility and membership do not imply local operator access.
The Hedge Fund Team Inside The Laptop
The Core Authority Ladder
These states are related, but they are never interchangeable:
Every dashboard page is read-only. It cannot promote research, approve risk, submit paper orders, write to a broker, alter code or policy, award proof credit, or enable live capital.
Stable Operating Model Versus Current State
The stable model is evidence-gated, paper-only, real-calendar, idempotent, and lineage-backed. It does not backfill days, simulate elapsed time, or force trades. The strict validated-strategy lane is separate from a bounded discovery lane that can gather small real paper outcomes without claiming an edge.
An append-only temporal evidence graph gives Qadam persistent experiment memory. Its rebuildable local index connects observations, patterns, tests, rejections, strategy versions, decisions and outcomes, while preserving the difference between observed, inferred, tested and governed evidence.
The frozen risk ladder allows up to US$500 for a first discovery experiment, up to US$2,000 after independent repeat confirmation, and an absolute US$5,000 ceiling only for a validated paper setup. One score or one winning trade cannot advance a tier.
At this edition, graph-assisted discovery is implemented and its five-real-market-day trial counts
only actual eligible sessions. Operator health, PaperOps state, Router handoffs and validated-edge
state are mutable: read observation_ready, ready_idle and
provisional_soak from the timestamped dashboard rather than treating this guide as live
telemetry.
Current counts, providers, source freshness, service state, certifications, research maturity, candidates, hardware evidence, and paper-route readiness can change. Read those facts from the current dashboard - especially System Overview, Decision Room, Order Monitor, and page-specific status cards - not from this guide.
First Ten-Minute Tour
- Start with Portfolio for the current paper-fund picture.
- Open Trading History for the complete broker-mirrored chronology.
- Read Data Sources and Trading Universe to understand usable evidence and affected markets.
- Read Pattern Recognition before Quantum Edge.
- Read Trading Strategies for hypothesis and validation state.
- Read Decision Room from evidence to consequence to decision.
- Open Order Monitor only for the current or recent lifecycle after a governed decision.
- Read Results & Lessons before Tests & Improvements.
- Open Qadam Team for qualitative operating-role context.
- Use System Overview to diagnose infrastructure and operating evidence.
Portfolio
What does the Alpaca Paper fund currently hold, and what is the present financial result?
Decision Room
What reached Akber, what emerged from the filter, and what did final governance conclude?
System Overview
Is the operating infrastructure trustworthy, and which root issue needs attention?
The Current 13 Dashboard Routes
These are the canonical destinations. Historical implementation labels are not the normal user journey.
system/team — Who performs each hedge-fund role, and where does its authority stop?fund/portfolio — What does the paper fund currently hold and report?fund/timeline — What broker-mirrored paper events occurred, in chronological order?observe/sources — Which sources are connected, fresh, trusted, degraded, or missing?observe/universe — Which markets and liquid paper instruments are watched?patterns/findings — Which relationships are live, under testing, validated, disproved, or faded?patterns/nonlinear — Did specialist analysis add anything beyond the strongest fair classical comparison?decide/strategies — How did a pattern become a hypothesis, and has it earned edge status?decide/decision — What entered Akber, what emerged, and what is the final governed position?trade/orders — Did a decision become a paper order or position, and what happened next?learn/outcomes — What happened, what is attributable, and what can Qadam learn?learn/improvements — Has a lesson earned a tested, versioned change?system/overview — Is Qadam's infrastructure healthy, and what root issue needs attention?Deep links use /dashboard/?module=<module>&view=<view>.
The Canonical Ten-Stage Lifecycle
This is a loop, not a single progress bar. Independent records can occupy different stages simultaneously.
How To Read Decision Room
Decision Room is the investment-committee governance page for Stages 6 and 7. It may contain no ideas, one idea, or several ideas; it is not a forced daily-trade view.
The Akber explainer and historical review archive are collapsed until selected. The final verdict is still read-only: it cannot allocate capital, approve risk, create an order, or write to Alpaca.
How To Read Quantum Edge
Quantum Edge is an independent proof room for selected nonlinear or quantum-assisted comparisons. It does not claim that every pattern needs quantum computation. Python prepares point-in-time evidence, the strongest fair classical method establishes the benchmark, and specialist methods are compared on the same frozen evidence.
The current conclusion comes from the canonical Quantum Edge projection rather than this guide. Its three principal rows are collapsed by default and should be opened in this order:
The canonical projection keeps provider readiness, local simulation, reproducibility, hardware execution, untouched market comparison, strategy contribution, paper impact, and freshness as separate facts. Provider readiness does not prove hardware execution. A successful simulation does not prove predictive or economic advantage. “Classical preferred” is a valid scientific outcome.
How To Read System Overview
The top answer separates infrastructure health, expected operating restrictions, and deduplicated root incidents. Its six diagnostic disclosures are collapsed by default:
Start with the verdict and root incidents. Open only the disclosure needed for the next diagnostic question.
Results & Lessons Versus Tests & Improvements
A result is not automatically a lesson, and a sensible lesson is not automatically an improvement. Reference-only broker history remains context rather than Qadam proof. Only a separately approved, timestamped, monitored, reversible version may re-enter Observe.
Status And Paper-Lifecycle Language
observation_readyThe unattended service and its projections are healthy enough to continue collecting evidence; it does not mean a trade exists.ready_idleThe guarded route passed its checks, but no accepted Router handoff existed for that pass.provisional_soakThe reliability implementation is complete while required multi-session real-time observations continue.active_discovery_trial_runningThe evidence-graph implementation is certified and is counting only eligible real market days. It does not imply an edge or advance the 30-day paper growth trial.implementation_certified_evidence_maturingGraph, memory, strategy and safety contracts passed while empirical outcomes remain too limited to claim conversion or profit.How Qadam Could Make Money In Paper Trading
The economic hypothesis is that Qadam can find information or structural relationships before market prices fully reflect them, prove that they repeat after realistic costs, and express them through a governed liquid paper instrument. The ten-stage lifecycle is the operating flow.
This is an intended mechanism, not a promise of profit. Any edge must be proven through attributable, out-of-sample, cost-aware paper evidence.
Validated strategy
The relationship has survived Qadam's full historical, holdout, cost, robustness and forward evidence standard before current tradeability review.
Discovery experiment
A complete current hypothesis may collect a small paper outcome before full validation. It still requires a trigger, direction, confirmation, expectancy, invalidation, liquidity, decision-time shadow evidence and lineage.
How Qadam Finds And Tests Edge
An edge is not one headline, one chart, one model opinion, a private prior, a Telegram message, a green label, or a high research score. A Pattern Recognition score ranks evidence for investigation; it is not automatically a calibrated probability, expected return, or chance of profit.
- Record a point-in-time observation and only the information available then.
- Map it to a watched price or probability without future leakage.
- Measure the outcome only after the relevant horizon elapses.
- Test repetition after realistic costs and false-discovery controls.
- Challenge it with walk-forward and untouched holdout evidence.
- Use Quantum Edge only where nonlinear structure could plausibly help.
- Observe the frozen idea forward over real time without an order.
- Promote, hold, reject, or mark it faded based on the evidence.
Akber's 6-Stage Decision-Making Filter
Akber does not originate an idea, prove a historical edge, approve risk, or execute an order. It asks whether an evidence-classified setup is practical in current market conditions. Validated strategies and bounded discovery setups use declared evidence profiles, but neither may omit its required trigger, confirmation, expectancy, invalidation, liquidity or lineage.
An Akber pass creates research eligibility only. Research eligibility is not approval. It does not create a candidate by itself, risk or execution approval, a PaperOps handoff, an order, a broker write, proof credit, or live-capital authority.
Paper Evaluation Rules
- Preserve every declared real-calendar evaluation window.
- Do not backfill calendar days, simulate elapsed time, or force trades.
- Keep mirror-only historical broker records separate from proof of Qadam's decision quality.
- Record the no-trade rationale when no setup qualifies.
- Keep drawdown, concentration, duplication, spread, liquidity, freshness, and idempotency controls active.
- Require complete lineage and a postmortem before a closed outcome can enter the paper proof ledger.
- Keep live capital outside the paper-only route.
Daily Reading Routine
- Start with Portfolio and Trading History.
- Read Data Sources and Trading Universe.
- Review Pattern Recognition, then Quantum Edge only when specialist comparison is relevant.
- Review Trading Strategies for hypothesis and validation state.
- Read Decision Room in evidence → consequence → decision order.
- Use Order Monitor for current or recent paper lifecycle state.
- Use Results & Lessons for attribution, then Tests & Improvements for change governance.
- Open System Overview when a page is stale, contradictory, unavailable, or unexpectedly empty.
- Add a precise governance comment only through an enabled protected member feature; a comment is not approval.
- Accept a no-trade state when no setup qualifies.
Telegram: Two Separate Read-Only Rails
Outbound explanation
Public-safe notifications and learning briefs explain what Qadam noticed, what changed, and what remains uncertain when the configured automation is due and allowed.
Inbound research intake
Configured member messages may supply articles, event context, ideas, or questions as provenance-linked, read-only research intake.
Neither rail can create commands, candidates, approvals, orders, broker writes, provider jobs, deployments, proof credit, strategy mutations, or live-capital authority. An inbound message cannot grant itself evidentiary status or bypass Akber, Router, risk, or PaperOps.
What Visitors, Members, And Operators Can Do
Public visitors can
Read the sanitized dashboard and public documentation, inspect public-safe evidence and paper states, and use explanations to understand why Qadam waited or acted.
Members can
Use protected member documentation and enabled governance features, challenge the record, and submit useful context through the configured inbound Telegram rail.
Local operators can
Run repository commands, inspect configured services, execute governed automation passes, and manage local secrets outside Git and public artifacts.
No dashboard, guide, comment, or Telegram user can place or approve a trade through the interface, bypass governance, mutate strategy or code, run shell commands through the web, write to broker-live endpoints, award unsupported proof, or enable live capital.
Data Source Rules
- One source is rarely enough; freshness, provenance, and point-in-time safety matter.
- Private worldview priors are context, not evidence.
- Supplemental providers remain context until registry and trust rules say otherwise.
- Broker receipts and portfolio state are execution evidence, not independent market corroboration.
- A connected provider does not prove complete history or current monitoring.
- Secrets, identifiers, raw payloads, and local paths must not enter public artifacts.
Troubleshooting
/dashboard/ without assuming login is required; ask the operator to inspect generation and deployment evidence./login/.ready_idle is a healthy no-handoff result, not an execution failure.Operator-Only Procedures
These commands are for the local repository operator, not dashboard users. They do not authorize backfill, forced trades, live capital, broker-live endpoints, Telegram commands, or promotion around failed gates.
scripts/bootstrap_runtime.shscripts/start_postgres_timescale_ingestion.sh.venv/bin/python scripts/check_cockpit_status.pyscripts/status_qadam_operator_launch_agent.sh and .venv/bin/python scripts/check_qadam_operator_service.py.venv/bin/python scripts/check_qadam_permanent_operator_reliability.py.venv/bin/python scripts/run_paperops_autonomous_pass.py.venv/bin/python scripts/check_postgres_timescale_replay.py --require-full-source-coverage.venv/bin/python scripts/run_daily_learning_automation.py --live.venv/bin/python scripts/check_qctrl_fire_opal_ibm_quantum.py --probe-devicesThe PaperOps summary is data/runtime/paperops_autonomous_pass_summary.json. Run each
automation once against the actual calendar and allow current evidence and authority gates to
determine the result. The installed operator uses macOS launchd and
caffeinate, with single-instance leases, resource locks, bounded retries, circuits,
repair requests and disk-pressure limits. It cannot edit code, install software, change secrets or
expand authority autonomously. Device discovery is not a hardware experiment or a market-edge claim.
Red Flags
- A stale or hard-coded count is presented as current runtime truth.
- A source claims health without freshness or monitoring evidence.
- A research score is presented as guaranteed profit.
- A pattern is called a validated edge without the required evidence.
- An Akber pass, candidate, Router review, or PaperOps handoff is presented as an order.
- Reference-only broker history is used to claim Qadam performance.
- A supported lesson changes Qadam without testing, approval, versioning, monitoring, and rollback.
- Telegram creates or accepts a command.
- Any interface claims shell, deployment, broker-write, proof-credit, or live-capital authority.
- A secret, token, member identifier, raw payload, credential, or local path appears publicly.
A Successful Reading Of Qadam
A new reader should be able to distinguish observation from evidence, pattern from edge, hypothesis from decision, handoff from order, outcome from lesson, and lesson from approved improvement. They should also know which dashboard page contains the current answer and whether the infrastructure is healthy enough to trust it.
Qadam should feel understandable before it feels powerful.