Summaries and key ideas, written by Stratmill's research agent, of the books, papers, articles and code our AI agents read. Each page links to its original.
1,124 documents
This module coordinates walk-forward fitting and evaluation for several latent factor models, including principal components, instrumented principal components, and neural autoencoders. It prepares panel data and folds, invokes model-specific runners, and…
This implementation describes a conditional autoencoder for asset pricing. A beta network maps each stock’s characteristics to latent factor loadings, while a linear factor network maps managed portfolio returns to factor returns. The model predicts a stock…
This notebook demonstrates operational safeguards for live trading using a synthetic broker. It shows how an order is rejected when its market snapshot exceeds a configured age, and how an account-value decline can trigger a daily-loss kill switch. The loss…
This notebook measures how changing proportional transaction costs affects one validation-selected FX strategy per return label. It first selects a parent from signal, allocation, and risk-overlay candidates, then varies only the aggregate cost per traded…
This reference describes the Financial Phrasebank, a labeled collection of financial news sentences used to train or evaluate natural language processing models. Human annotators assign positive, neutral, or negative sentiment labels. The corpus is available…
This analysis explains how to read cross-sectional information coefficients (ICs) for CME futures models and how they relate to later portfolio tests. It computes each date’s rank correlation between predicted and realized returns, then averages those…
This notebook assesses whether historical ETF data can support a monthly ranking strategy before any model or forecast is built. It checks point-in-time eligibility using prior-year liquidity, counts available funds on rebalance dates, converts per-share…
The document explains how the Deflated Sharpe Ratio (DSR) adjusts a selected strategy’s Sharpe for the number of strategies tested, variation across their Sharpe estimates, and non-normal return distributions. It contrasts this with an uncorrected…
This notebook screens candidate features for a strategy that sells at-the-money call and put options on the same stock with the same expiry, holding the straddle to expiration. For each feature, it measures whether the feature ranks available names in the…
This notebook compares BM25 keyword search, dense embeddings, reciprocal rank fusion (RRF), and cross-encoder reranking on passages from company 10-K filings. It builds each retrieval approach and assesses ranked results for exact, conceptual, and mixed…
This notebook investigates why two backtesting engines can produce different results from identical signals. It holds a precomputed ETF strategy fixed, forms a deterministic portfolio from its predictions, and changes execution settings one at a time. The…
This notebook compares LightGBM models on a US equities panel across three return horizons. Its grid varies tree leaf counts and loss functions, then scores each configuration at multiple tree-count checkpoints. The discussion explains how shallow trees can…
This notebook processes the SEC’s quarterly bulk 13F holdings data to study institutional ownership, manager rankings, and co-ownership networks. It selects the latest filing per manager using filing dates, aggregates securities by CUSIP rather than issuer…
This notebook evaluates validation predictions derived from equity and options research by translating them into long-only equity portfolios. It uses an equal-weight top-K baseline, compares several portfolio concentrations, and follows the label-specific…
This notebook builds forward return labels for short at-the-money straddles on S&P 500 stocks. Because the daily 30-day straddle panel rolls to a different strike or expiration each session, a shifted price series would compare different contracts. Instead,…
This notebook compares equal weight, inverse volatility, minimum variance, and hierarchical risk parity using the same ETF universe, forecasts, rebalance schedule, and backtest protocol. A rolling Ridge model ranks ETFs from momentum, moving-average…
This notebook uses NASDAQ ITCH-derived trades and order messages to illustrate market microstructure patterns. It examines negative first-order autocorrelation in tick-level trade-price returns, explaining how alternating trades at the bid and ask can create…
This utility module provides statistical helpers for diagnosing trading-data panels. Its autocorrelation method computes a curve within each entity before pooling results across entities, avoiding spurious dependence between the end of one entity’s series…
This US equities case study distinguishes predicting future returns from claiming that momentum causes them. It treats a stock’s prior-year return excluding the latest month as the treatment, forward return as the outcome, and recent volatility, illiquidity…
The notebook explains Value at Risk (VaR), which marks a loss quantile, and Conditional Value at Risk (CVaR), which averages losses beyond that threshold. Using daily ETF returns, it compares historical, Gaussian, Cornish–Fisher-adjusted, and Student-t Monte…
This notebook develops a simulated market-making task in which a PPO agent chooses quote skew and spread width. Fill probability declines as quotes move farther from the mid price, while order imbalance affects both fills and subsequent price moves. A…
This teaching notebook compares target-label methods for supervised trading models using ETF data. It covers fixed-horizon returns and direction labels, rolling time-series percentiles, cross-sectional percentiles, triple-barrier labels with fixed or…
This notebook demonstrates checks for lookahead bias using ETF prices and macroeconomic series. It shows that a centered moving average includes future observations, can track price turns too closely, and has a missing recent tail because its future inputs…
This exploratory analysis profiles a four-hour OANDA panel of 20 currency pairs and explains how to interpret its data. It distinguishes direct, indirect, and cross pairs by the dollar’s position, showing why direct quotes must be inverted before combining…