Risk Management
Capital protection is a design constraint in TRADEOS.tech. The public docs explain the risk philosophy and safety boundaries without disclosing exact thresholds, formulas, parameter names, local configuration paths, or calibration logic.
Risk Philosophy
TradeOS treats every market observation as something to validate, not something to trust automatically. A signal must pass independent review layers before it can affect paper validation or live execution.
The system is designed around four principles:
| Principle | Public meaning |
|---|---|
| Fail closed | If a required safety check cannot complete, the system should prefer inaction. |
| Separate duties | Signal generation, risk validation, execution, and publishing are separate responsibilities. |
| Bound exposure | Position, portfolio, venue, and drawdown risk are constrained by deployment policy. |
| Measure outcomes | Signals and decisions are recorded so later reviews can evaluate what worked and what failed. |
Risk Layers
Signal Review
Signals are screened for quality, freshness, market context, and conflicting evidence. Low-quality or stale observations should not advance.
Portfolio Validation
Before any execution path is considered, TradeOS evaluates whether the proposed action fits the current portfolio, exposure, drawdown, and venue constraints. Exact sizing formulas and limits are deployment policy.
Position Management
Open positions remain monitored after entry. The system can respond when market evidence changes, risk rises, or the original thesis no longer appears supportable. Exact exit logic, thresholds, local state paths, and timing are not published.
Systemic Safety
System-level controls can pause, constrain, or halt activity when infrastructure health, data freshness, or operator policy requires it.
Profile Behavior
Risk profiles are qualitative operating modes, not public parameter tables:
- Conservative prioritizes capital preservation and requires stronger evidence.
- Balanced aims for moderate activity with controlled exposure.
- Aggressive accepts broader risk and should only be used by operators who understand the consequences.
Specific profile values are configured per deployment and are not part of the external documentation.
Public-Intelligence Safety
Public thesis intelligence follows its own safety boundary:
- research output is not investment advice;
- thesis drafts must stay inside source-backed evidence;
- token references should include chain and contract identity where available;
- unsupported claims should be blocked;
- publication should remain pausable.
Risk management is not only about trades. It is also about preventing unsupported public claims, symbol confusion, and narrative drift.