point-in-time-quant-validator

Point-in-Time Quant Validation SDK

A zero-dependency Python engineering toolkit designed for quantitative research teams, algorithmic trading engineers, and financial machine learning systems to eliminate look-ahead bias and audit temporal data causality.

Core Capabilities

• Purged Walk-Forward Splitting: Deterministically separates training and testing periods with configurable purge windows and embargo buffers to eliminate serial leakage and label overlap.
• Causal Interval Validation: Strictly enforces temporal ordering between observation time, publication/arrival time, decision time, and realized outcome intervals.
• Automated Leakage Audit: Scans historical time-series matrices and feature pipelines for look-ahead contamination, retrospective restatements, and arrival order inversions.
• Tamper-Evident Validation Receipts: Generates deterministic SHA-256 integrity receipts capturing partition indices and boundary timestamps for regulatory and investor auditability.

Quick Start

import datetime as dt
from pit_quant.splits import PurgedWalkForwardSplitter
from pit_quant.leakage_detector import LeakageDetector

timestamps = [dt.datetime(2026, 1, 1, tzinfo=dt.timezone.utc) + dt.timedelta(minutes=5*i) for i in range(100)]
splitter = PurgedWalkForwardSplitter(n_splits=3, purge_window_seconds=1800.0)
folds = splitter.split(timestamps)
for f in folds:
    print(f"Fold {f.fold_index}: Train={len(f.train_indices)} -> Gap={f.purge_gap_seconds}s -> Test={len(f.test_indices)}")

Production Standards

• Pure Python 3.10+ standard library. Zero third-party runtime dependencies.
• 100% reproducible test suite.