Universe Coverage & Freshness
Schwab verification, broad-screen coverage, hydration freshness, and API delay
Eight Primary Agents
Two independent $100k replicas at each risk tier; AI cost is shown separately from trading performance
| Agent | Risk / Replica | Decisions | Actual NLV / Return | AI cost | Economic NLV / Return |
|---|
Risk & Replica Checks
Decision counts and outcomes expose within-tier A/B variability
| Group | Decisions | Trades | Wins / Losses | Evaluated P/L |
|---|
Account-Level Evaluation
Risk-adjusted outcomes, calibration, plan discipline, hedge value, trading friction, and abstention cost for each independent primary agent
| Agent | Drawdown / downside | Economic benchmark alpha | Calibration | Plan / rule / hedge | MFE / MAE | Turnover / execution cost | Plan counterfactuals | ITT opportunity accounting | Status |
|---|
Persistent $20k Mechanical Shadows
The same ex-ante primary intents, resized by trusted core code; no extra model calls and no claim of adaptive $20k reasoning
| Primary twin | Risk / Replica | Shadow NLV / Return | Realized P/L | Positions | One-unit rejects | Status |
|---|
Pre-Registered No-Token Benchmarks
Cash, SPY, monthly 60/40 SPY/IEF, and frozen trend/volatility policies under executable-cost assumptions
| Benchmark | Value / cash | Return | Eligible-session marks / executable cost | As of | Frozen definition / mark evidence | Status |
|---|
Data & Capability Request Backlog
Studies, instruments, and simulator capabilities requested by the agents
| Time | Agent | Type | Request | Decision use / fallback |
|---|
Decision Memoranda
0 reports