Agentic Trading
Our Agentic Trading system is a rules-based options framework designed to combine machine-speed market monitoring with disciplined, pre-defined risk controls.
Rather than simply generating trade ideas, the system operates through a structured decision loop:
Observe → Decide → Record → Execute → Evaluate
How It Works
The system continuously evaluates a liquid universe of U.S. securities and their options markets.
It considers factors including:
- Market and sector context
- Liquidity and trading activity
- Options pricing and spreads
- Volatility and sensitivity metrics
- Upcoming earnings and other known events
- Portfolio concentration and correlated exposures
- Expected reward relative to defined downside
A trade is only considered when multiple independent conditions align. Trading frequency is a secondary objective; the system is explicitly permitted to do nothing when suitable opportunities are absent.
Pre-Trade Reasoning
Every position begins with a documented thesis.
Before an order can be submitted, the system records:
Why the opportunity may exist.
A mechanical screen alone is not considered an investment thesis.
What could make the thesis wrong.
Each trade has identifiable invalidation conditions.
Expected risk and asymmetric upside.
Positions are structured so losses remain bounded while successful trades may retain meaningful upside.
Confidence.
The system assigns an internal confidence estimate that can later be compared with actual results.
This creates an auditable record of what the system believed before knowing the outcome.
Systematic Execution
Execution follows its own rules.
Orders are generally entered using controlled limit-order logic rather than indiscriminate market execution.
The system also monitors:
- Portfolio capacity
- Available capital
- Position-level exposure
- Sector and thematic concentration
- Liquidity deterioration
- Time remaining in an option
- Changes to the original investment thesis
Execution parameters are deliberately separated from the investment decision itself.
Risk Architecture
Risk limits sit above the trading strategy.
The system maintains predefined constraints on:
- Individual position size
- Aggregate capital at risk
- Minimum liquidity reserves
- Correlated positions
- Sector exposure
- Drawdowns
- Consecutive losses
Certain conditions automatically suspend new trading.
Crucially, losses cannot cause the system to increase position sizes, loosen entry standards, or attempt to rapidly recover previous losses.
Risk controls have priority over opportunity generation.
Asymmetric Position Management
Agentic Trading is designed around asymmetric outcomes rather than fixed symmetric profit-and-loss targets.
Losses are generally cut according to predefined thresholds or thesis invalidation.
Profitable positions may be partially realized while retaining exposure to unusually large moves.
This allows the portfolio to limit the effect of unsuccessful trades without mechanically eliminating the upside of its strongest positions.
Calibration
The system also measures whether its own confidence estimates are reliable.
If trades assigned a certain confidence level consistently perform materially differently from that expectation, the discrepancy becomes measurable.
The goal is not merely to generate predictions.
It is to determine how much trust those predictions deserve.
Continuous Validation
Agentic Trading is built as an evolving operating system rather than a static trading algorithm.
Its rules, assumptions, execution behavior, and realized outcomes are continuously examined for:
- Statistical weaknesses
- Execution friction
- Calibration errors
- Hidden correlations
- Market-regime dependence
- Operational failures
When an unexpected result occurs, the system first asks whether the market produced an unusual outcome—or whether the process itself failed.
That distinction is central to the architecture.
Human Rules. Agentic Execution.
The investment framework defines the boundaries.
Within those boundaries, autonomous agents can monitor markets, evaluate opportunities, document decisions, manage positions, enforce risk controls, and review completed trades.