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.
742 documents
This notebook explains why a generic target-weight risk overlay is unsuitable for the described S&P 500 short-straddle engine. The engine assigns fixed capital shares to weekly cohorts and normalizes weights within each cohort to sum to one. Scaling those…
Risk-Premium PCA modifies the covariance matrix by adding a weighted outer product of the training mean-return vector. Its leading eigenvectors define static factor portfolios: ordinary PCA is recovered when the weight is zero, while positive weights tilt…
This notebook measures how a selected US firm-characteristics strategy’s validation performance changes as assumed commissions and slippage vary. It re-runs the same configuration across a declared cost grid, splitting each total charge evenly between…
This notebook describes how to produce predictions for an untouched holdout period after choosing a configuration on validation data. It resolves the selected configuration, rebuilds its training specification for the holdout, and derives the training window…
This notebook describes how to produce holdout predictions for a selected CME futures model configuration. The configuration is chosen using validation results, then refitted on data ending before the holdout window opens. A label buffer separates training…
This notebook describes how to produce holdout predictions for a selected NASDAQ-100 microstructure configuration after model and strategy choices have been made on validation folds. It reconstructs the selected training specification, derives a holdout fold…
This guide introduces the public CFTC Commitment of Traders reports as a source of weekly futures positioning data. It distinguishes the Traders in Financial Futures report, which categorizes participants such as dealers, asset managers, and leveraged money,…
This Python module describes methods for quantifying uncertainty in strategy backtests using daily out-of-sample returns. It combines stationary block bootstrap intervals for return-series statistics, a Sharpe standard error adjusted for autocorrelation and…
This notebook synthesizes registered validation results for five latent-factor estimators across a set of case studies. It first maps which estimators have results for each panel, then identifies the leading estimator by mean daily Spearman rank information…
This notebook demonstrates single-factor evaluation using real ETF data and a price-derived momentum example. It measures cross-sectional Spearman information coefficients against forward returns, compares horizons, and explains the distinction between…
This case study screens candidate financial and model-based features against forward returns for a NASDAQ-100 microstructure strategy. It calculates cross-sectional rank information coefficients through time, adjusts inference for dependence and multiple…
This notebook selects a CME futures case study from registered validation backtests spanning signal rules, allocation, and risk overlays. It compares candidates across both return horizons and chooses the configuration with the highest validation Sharpe.…
This notebook explains how to construct forward intraday return labels for NASDAQ-100 stocks using minute bars and NBBO quotes. It defines returns from the next bar after a signal to a specified future endpoint, uses quote midpoints to reduce bid-ask bounce,…
This case study compares gradient boosting models that predict returns to expiry for short at-the-money straddles. It varies tree capacity and three fitting objectives: squared error, absolute error, and Huber loss. Walk-forward validation uses two periods,…
The document presents an operational workflow for keeping daily market data current: load an initial history, fetch a small overlapping delta, check for missing dates, and monitor freshness and validation issues. The overlap helps capture vendor revisions to…
The document describes how a CME futures study generates predictions for a holdout period after selecting a model configuration on validation data. The selected configuration and checkpoint are recovered from the validation results, then the model is…
This notebook fits a patched transformer to predict a primary label in an S&P 500 equity and options study. It divides each stock’s historical input window into contiguous patches and treats each patch as a token for attention. Attention can compare distant…
This configuration describes a weekly at-the-money straddle-selling study on S&P 500 constituents. Positions are opened on Friday and held toward expiry, with daily stock hedges triggered by changes in option delta. The portfolio spreads capital across…
This notebook explains an ETF stochastic discount factor model that fits a single series of weights to price returns across funds. Unlike models that separately estimate fund exposures and factor returns, it trains two networks against pricing errors: one…
This notebook describes how to create holdout predictions for a previously selected US equities model configuration. Since model choices and strategy settings were made using validation results, those same results cannot independently establish the quality…
This notebook presents GReaT, a method for generating synthetic financial tables with a language model. It serializes each row as text, fine-tunes a GPT-based model on ETF features, and parses generated text back into mixed numerical and categorical columns.…
This notebook explains how to construct model-based features for a US equities panel without letting future observations influence historical values. Its central method is an estimation schedule: models use a burn-in period, fit only on data preceding the…