Core Concepts¶

Domain namespaces¶

Version 1.2.0 groups the public surface by research problem rather than by implementation detail. The primary namespaces are:

data, hypotheses, alpha, factors, risk, microstructure, backtesting, portfolio, stats, ml, options, rates, stochastic, mc, vol, research and trading.

Canonical verbs¶

Where appropriate, domain namespaces expose a small vocabulary:

load / validate
discover
run
optimize
price
analyze / calibrate
regress
fit

These wrappers standardize names and result contracts while preserving the lower-level 1.x functions.

Result contracts¶

Structured 1.2 results favor:

summary for compact metrics;
to_frame() for tabular payloads when available;
to_dict() for serializable output when available;
fingerprints or metadata when the underlying engine supports them.

Chronology is part of the model¶

A valid quantitative workflow must make information timing explicit. ASRQuant therefore treats lagging, walk-forward splits, execution delay and data provenance as research mechanics rather than presentation details.

Research evidence is not a trading claim¶

Statistical significance, novelty, robust backtesting and implementation correctness answer different questions. ASRQuant keeps those layers separate rather than collapsing them into a single performance score.