Market data
Canonical records
UTC timestamps
Source provenance
INSIDE SYRELYX / ARCHITECTURE OVERVIEW
Curiosity, meet
discipline.
A research system is only as useful as the boundaries around it. Here’s the thinking behind our data, evidence and simulation layers.
RESEARCH + PAPER ONLY01 / FROM OBSERVATION TO SIMULATION
A traceable path.
Not a black box.
A modular, Python-based architecture connects reproducible research to a bounded simulation environment. This is a conceptual overview, not a live dashboard or deployment-status report.
Research
Point-in-time features
Hypothesis evaluation
Evidence review
Risk gates
Portfolio state
Explicit limits
Execution-time checks
Simulation
Backtest / paper fills
Ledger records
Comparable analytics
02 / THE RESEARCH LAYERS
Built to ask
better questions.
01Canonical data & provenance
+
Public market-data and CSV adapters feed consistent records with UTC timestamps, precision-aware values and source context. Backtests use stored datasets rather than silently fetching missing history.
CONSISTENCY / TRACEABILITY / POINT-IN-TIME CONTEXT02Deterministic backtesting
+
Event-driven tests make signal timing, simulated fills, fees and slippage explicit. Benchmark comparisons and repeatable experiment records help make results explainable rather than merely impressive.
REPRODUCIBILITY / EXPLICIT COSTS / COMPARABLE RESULTS03Evidence-governed intelligence
+
Research hypotheses, candidate evaluation and specialist-agent analysis belong to a separate research control plane. Source attribution, critical review and human decisions remain part of the process. External source content is treated as untrusted.
HYPOTHESES / CRITICS / HUMAN REVIEW04Bounded paper simulation
+
Paper runs use public market inputs and simulated funds. Portfolio exposure, drawdown and daily-loss controls constrain simulated activity. Protective halts do not become something an agent can simply override.
SIMULATED FUNDS / RISK CONTROLS / EXPLICIT ACTIVATION03 / CONTROL BY DESIGN
Two gates.
One clear boundary.
Risk is checked before a simulated order is accepted and again against the current state before simulated execution. Intelligence informs the process; it does not own the permission.
Before acceptance.
Evaluate the intended action against portfolio and risk constraints.
Before simulated execution.
Re-check the latest state and require explicit simulation authorization.
No live brokerage credentials. No live-money orders.
04 / WHAT THIS MEANS
Clear about
the limits.
Can I use SYRELYX to trade live funds?+
No. The project scope described here is research, backtesting and paper simulation. It is not a live-money execution service.
Is the platform publicly available?+
SYRELYX is in development. This site introduces the project, not a public product release or an invitation to fund a trading account.
Does the website show real performance?+
No. The website’s architecture illustration is conceptual. We do not present simulated figures as real investment returns or claim a profitable trading edge.
Where can I follow the project?+
Use the official channel directory. YouTube @SYRELYXHQ is the main channel. TikTok is currently unavailable with an appeal pending; LinkedIn and the Bluesky link are still being finalised.