Content
57%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is strong on concrete, engine-specific guidance — interface format, config schema, examples, and pitfalls are all actionable — but it pads token budget with standard finance textbook material, never sequences the run workflow explicitly, and references an example engine file that is not in the bundle. Trimming known-concept sections and fixing the dangling reference are the highest-leverage improvements.
Suggestions
Replace the Black-Scholes formula block and the Greeks definitions table with one line each (e.g., "European BS pricing with historical-vol-as-IV approximation; portfolio Greeks output daily") and keep only the engine-specific deviations from textbook assumptions.
Add a numbered workflow section (write code/signal_engine.py → create config.json with engine "options" → run the backtest → review artifacts/equity.csv, greeks.csv, trades.csv) including the command to invoke the runner.
Ship the promised example_signal_engine.py in the bundle (or remove/redirect the reference) and add one minimal runnable generate() implementation for a single covered-call signal.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The engine-specific sections (interface, config, artifacts, pitfalls) are tight, but "BS Model Principles" spells out the Black-Scholes formula with S/K/T/r/sigma definitions and "Greeks Meaning and Usage" defines Delta/Gamma/Theta/Vega — textbook finance Claude already knows, so the body could be noticeably trimmed. More than minor padding (keeps it below 4), but most sections do earn their tokens (keeps it above 2). | 3 / 5 |
Actionability | Provides the exact SignalEngine instruction dict format, a concrete iron condor signal example, a complete config.json with field-by-field explanation, the artifact file listing, and the options_pricing tool call format — mostly executable guidance. Minor gaps: no runnable generate() implementation, the referenced example_signal_engine.py is absent from the bundle, and no command for actually invoking the backtest runner. | 4 / 5 |
Workflow Clarity | The workflow (write code/signal_engine.py → set config.json with engine "options" → run → inspect artifacts/) is only implicit in section order, never sequenced explicitly, and the doc never shows how to launch the runner — matching 'sequence present but checkpoints missing or implicit'. Each step's details are well defined, so it stays above the many-gaps level of 2. | 3 / 5 |
Progressive Disclosure | The single file is cleanly sectioned and navigable, but it points to "example_signal_engine.py next to this file" which does not exist in the bundle (a dangling reference), and concept-primer material (BS model, Greeks table) that could live in a reference file is inlined. Good structure is undermined by the broken external pointer, placing it between the 3 and 4 anchors but with the reference failure pulling it to 3. | 3 / 5 |
Total | 13 / 20 Passed |