—
Broker verification required
YOUR RESEARCH. YOUR RULES.
A clear view of the market.
A firm grip on every decision.
Real-money trading is disabled by construction.
MISSION CONTROL
Sign in with your owner password and authenticator.
MISSION CONTROL
Research with purpose. Execute within your limits.
Broker verification required
Realized + unrealized, net of costs
Maximum planned risk $250 each
Persistent same-session lockout
PORTFOLIO
DETERMINISTIC POLICY
The daily loss threshold triggers liquidation. Gaps, slippage, or execution failures can cause final losses above $750.
MARKET COVERAGE
No confirmed feed coverage.
U.S.-listed equities of any market cap. Liquidity and opportunity evidence determine selection.
RESEARCH DIRECTOR
Strategies move through historical validation, independent testing, shadow observation and paper qualification.
OpenAI not connected. No research spend incurred by this build.
SCHWAB AUTHORIZATION
Day-5 early warning and Day-6 renewal reminders. Expiry must come from verified provider responses.
OAuth setup and live certification remain pending. No Schwab order submission route is included.
BROAD MARKET DISCOVERY
Candidate evidence, rejection reasons, tier membership and execution decisions. No fixed top-500 universe.
Saved targets apply on the next safe selection and confirmed subscription rotation. Open positions and pending orders keep full-feed protection. Lower counts can reduce CPU, memory, network and new data storage; existing archives are retained. Limits: 500 total, 100 full-feed, 400 lower-frequency. Higher capacity requires performance and provider review.
160 GB server planning budget. Defaults reserve 32 GB and cap the database at 60 GB; other services, system files and backups also need space. Low headroom blocks new entries and research, while exit handling remains available. New market observations use gzip level 6 compressed segments. Historical inline data is unchanged. Automatic expiry awaits a dependency index and reviewed cleanup; these retention targets do not currently remove data. Trade/audit evidence is preserved.
QUANT LAB
RESEARCH QUEUE
Questions enter a durable research queue. The separate research worker runs only permitted jobs; OpenAI requests remain blocked until credentials, model capabilities and cost bounds are verified.
REPRODUCIBLE RESEARCH
Historical fills are model assumptions. Unresolved exits, missing calendar coverage and insufficient samples cannot qualify a strategy.
Common signals isolate sizing effects. Holding periods are separate strategy variants. Results are diagnostic model assumptions; they cannot change owner settings or qualify trades. Registered adaptive methods are compared separately with the standard target, using original per-share R. Partial exits assume atomic protection resizing for research; broker execution is not qualified. Smaller risk does not create more valid signals.
Raw capture preserves evidence and retention holds. It does not verify point-in-time features, corporate actions or training eligibility.
Partial selected-tape OHLCV only. Calendar, corporate actions, full-market coverage and historical decision-time availability are not verified; these projections cannot qualify strategies.
Recent local journal commits, not publication order or verified credibility. Full provider articles may contain untrusted markup. Opening evidence does not authorize AI use or trades.
Requires separately verified schema/entitlement capability and a securely stored Massive key. A queued job does not establish a connection.
Offline packets preserve retained revisions for investigation. Credibility remains unknown; no paid research or trade instruction is authorized.
Owner/offline plans only. Exact richer model/schema/cost verification remains blocked; no paid AI request or trade approval.
Prepare an offline context from your question, then inspect its retained research records. Summaries are partial and untrusted. Full records may contain private research data; opening them does not authorize an AI request or qualify a model.
No local review context prepared.
MODEL RESEARCH
Model families are open-ended. A registered proposal does not mean a model has been trained, tested or put into operation.
The first trainer fits bounded native JSON tree ensembles. Other families remain proposals until their training and execution adapters are implemented and validated. Inputs must be registered numerical datasets; raw captures cannot train a qualified model. Fixed chronological training, validation and test periods are recorded. Results cannot activate strategies or change risk limits.
IMMUTABLE EVIDENCE
Ordered original observations and deterministic decisions. AI interpretations must be separately labeled.
EVENT INTELLIGENCE
Weekly discovery and daily targeted verification require OpenAI connection and a verified cost bound. Scheduled dates alone never trigger orders.
DAILY INTELLIGENCE
The internal ledger remains authoritative. Google publishing and AI review are separate pending integration gates.
SECURE API SETUP
Credentials are encrypted on the server. Stored keys are never returned to the browser.
A new MFA code is required for each sensitive action. Use your next code after login. Revoking the selected credential stops entries and requires provider revalidation.
HONEST CONNECTION STATUS
Massive stocks and Benzinga news: subscription reported by owner. Actual API access, SIP/NBBO coverage, stream limits and licensing remain unverified.
These probes do not place orders or certify the complete integration.
OWNER POLICY
Increases affect new positions. Tightening stops entries and triggers exposure review, including pending orders. The daily loss threshold stays separate; a smaller R does not create more valid signals or change stop distances. Daily lockout cannot be cleared. Whole-share or liquidity limits may use less than a full risk unit.
RELEASE READINESS
At least 10 full paper sessions are required before assessment. Time or profits alone cannot authorize real money.