Accessible ML tools & brute tests

Riskfolio‑Lib’s official examples list Hierarchical Risk Parity, CVXPY as the optimization backbone, and a MOSEK-backed “real applications” notebook that runs on 612 assets and 4,943 observations. (riskfolio‑lib.readthedocs.io ) PyQuant News has published an HRP walkthrough and newsletter post that demonstrates building ML‑style portfolio workflows with Riskfolio‑Lib and accompanying notebooks for practical implementation. (pyquantnews.com ) Pattern Computer announced PatternDE — the Pattern Discovery Engine — on Jan. 2, 2025, marketing it as an online platform to generate and evaluate large numbers of hypotheses over high‑dimensional tabular data. (patterncomputer.com ) Company materials for PatternDE and Pattern Computer’s product pages frame the engine as an “end‑to‑end” hypothesis generator with explainability, and cite industry pilots including a collaboration on battery diagnostics with Dynexus. (pcidiscovery.com patterncomputer.com ) Methodological literature cautions that automated brute‑force searches across parameter and rule spaces inflate false‑discovery risk and produce backtest overfitting unless controls for multiple testing and genuine out‑of‑sample validation are applied. (academic.oup.com ) Statistical properties of Sharpe‑ratio estimates matter: formal work on the sampling distribution of the Sharpe shows small‑sample bias and time‑series dependence can make ex‑post Sharpe numbers misleading without robust inference. (JSTOR / Stanford ) Riskfolio‑Lib’s GitHub and notebooks include integrations with backtesting tools (e.g., vectorbt) and concrete tutorial code for moving from optimization to performance metrics, while pattern‑discovery vendors explicitly emphasize the need for clean feature engineering and domain priors when scaling hypothesis tests.