Quant Calculator turns user-provided or machine-readable data into inspectable, deterministic results. It inspects schema first, normalizes dates and units, routes market history through adjusted fields when returns matter, and uses a bundled series helper for summary stats, correlation, regression, and drawdown instead of mental arithmetic.
更新于—
内置技能
Driven
内置技能
Every run defines the requested output, date window, frequency, units, and missing-value treatment before calculating. CSV, JSON, and XLSX inputs are previewed with bounded reads; Driven market payloads preserve source, warnings, currency, and field semantics instead of assuming a flat .data[] shape.
Small transforms use jq or xan; bespoke logic uses Node on absolute /workspace/... paths. Repeated financial-series work routes through the bundled series-math.js helper for return summaries, dated return correlations, regression, and drawdown — invalid inputs return structured errors rather than silent estimates.
Returns, volatility, drawdown, and correlation default to dividend-adjusted historical prices unless the user asks otherwise. Multiple price series align on date inner-joins; return correlations join only intervals with matching start and end dates — never row-position pairing.
Answers state method, sample coverage, exclusions, and limitations. Large or reusable outputs land in verified workspace artifacts with absolute paths; numeric claims in chat must match the structured helper readback, not recomputation from memory.