คลังความรู้

This document explains how TWAP and VWAP divide a large parent order into child orders over a trading session. TWAP allocates equal shares to each time interval; VWAP allocates shares in proportion to a forecast of interval volume. The notebook estimates…

This notebook presents a supervised autoencoder for predicting ETF forward returns. The network compresses a fund’s feature row into a low-dimensional representation and predicts returns directly from that representation. Unlike conditional autoencoders in…

This document describes how to derive a single holdout training interval from an existing walk-forward validation specification. It takes the declared holdout dates from the study configuration, uses the earliest training start across validation folds, and…

This notebook develops forward return labels for short at-the-money straddles on S&P 500 stocks. Since the daily panel’s nominal 30-day straddle represents a different contract each session, a shifted price series would mix instruments. The label instead…

This notebook explains how to test whether events such as trading signals or announcements predict returns beyond a market benchmark. It demonstrates the workflow using momentum breakouts in liquid ETFs: estimate a market model from pre-event returns,…

This feasibility analysis evaluates whether quote data can support a market-neutral, cross-sectional strategy that ranks NASDAQ-100 constituents at a fifteen-minute cadence and holds both ends of the ranking. It checks whether enough eligible symbols are…

This analysis compares feature-triage ledgers across nine trading case studies. It distinguishes Benjamini–Hochberg false-discovery-rate significance from a second route to a PROCEED decision, which combines a case-specific effect-size floor with fold sign…

This notebook fine-tunes three transformer checkpoints for three-class financial sentence sentiment and compares their accuracy, macro F1, confusion matrices, parameter counts, and training time. It uses a stratified train, validation, and test split of the…

This document records a structured bull and bear debate about whether the Federal Reserve will raise its target rate during 2026. Three agents assess the question, with arguments drawing on Fed projections, inflation, labor-market conditions, forecasts from…

This notebook explains how to extract insider transactions from raw SEC Form 4 XML for equity research. It treats each transaction block as the unit of parsing, preserving the association among transaction code, date, share count, price, and direction. XML…

This notebook builds principal-component factors from a panel of CME futures returns, not from engineered characteristics such as carry, momentum, or volatility. PCA finds directions that explain variation across products; the first may resemble a common…

This notebook compares stop-loss, trailing-stop, and time-exit overlays on fixed NASDAQ-100 microstructure backtests. The strategy makes decisions every fifteen minutes, while risk rules monitor prices every minute, allowing an overlay to exit a position…

This document introduces two latent-factor approaches for futures returns. Principal component analysis finds uncorrelated directions that explain the most return variation, while a neural stochastic discount factor seeks combinations related to the…

This notebook describes slow-moving contextual features for daily trading decisions: accounting ratios, macroeconomic conditions, and calendar encodings. It outlines value and quality measures such as earnings yield, profitability, accruals, leverage, and…

This notebook screens candidate features for a strategy that sells at-the-money straddles and holds them to expiration. For each trading day, it compares the cross-sectional ranking of available names by a feature with the ranking of their eventual straddle…

This document explains how to build daily forward-return labels for ETFs using adjusted closing prices at the observation date and a later trading session. It creates a primary monthly-horizon label and a shorter weekly variant, with labels computed on each…

This notebook compares split-conformal prediction, conformalized quantile regression, and adaptive conformal inference for forecasting ETF forward returns. In walk-forward folds, it trains on one subset, calibrates interval errors on another, and evaluates…

This notebook evaluates validation predictions as long-only equity portfolios, using equal-weight top-K selection and a decision schedule determined by each label's cadence. It checks the close-to-next-open event ordering with a seeded random-signal run,…

This document describes ways to turn dated entity records into arrays for latent-factor case studies. One method keeps only entities that meet a coverage threshold learned from an eligibility dataset, producing a consistent entity axis with missing…

This shared utility module provides a common gradient-boosting pipeline for quantitative research. It defines capacity presets and maps canonical parameter names into the corresponding settings for LightGBM, XGBoost, CatBoost, and scikit-learn histogram…

The notebook studies whether order-book activity predicts near-term returns for NVDA. It builds a distance-weighted measure of book pressure from market-by-order adds, cancels, and fills, then smooths the signed activity to represent liquidity changes near…

This notebook examines a keyword-based ESG headline pipeline and contrasts its structured classifier output with the requirements a retrieval-augmented generation assistant would need to meet. A first keyword list selects headlines, and a second assigns…

The notebook uses SHAP to explain FinBERT’s three-class financial sentiment probabilities at the token level. It pins a model checkpoint, aligns output labels explicitly, then examines which tokens raise or lower the predicted class probability for several…

This notebook inventories model-based feature columns across nine market case studies by reading Parquet schemas rather than loading full datasets. It excludes identifier fields, counts columns by market, and assigns feature families through tokens in their…