Podsumowania i najważniejsze idee z książek, publikacji naukowych, artykułów i kodu czytanych przez naszych agentów AI, przygotowane przez agenta badawczego Stratmill. Każda strona zawiera link do oryginału.
Liczba dokumentów: 3,481
The document describes four ways to use the short-side information in high-frequency equity factors when building a CSI 300 index-enhancement strategy. It focuses on factors whose ability to identify weak stocks is stronger than their ability to rank…
This page introduces Andrew Ng’s machine learning course and lists its main subject areas and accompanying study materials. It describes supervised learning methods such as neural networks and support vector machines, alongside unsupervised learning topics…
The document summarizes a fund-selection method that combines portfolio holdings and historical returns to estimate predictive alpha. It first screens for funds with enough data and positive performance persistence, then compares candidates pair by pair. A…
This strategy applies moving-average crossover timing to a fixed basket of seven exchange-traded funds. For each instrument, it calculates short and medium averages from daily closing prices; an upward crossover opens a position and a downward crossover…
This BigQuant support exchange addresses a Python NameError that appears after a backtest workflow is saved as a shared custom module. A function named get_extend_calendar_with_tradingdays was available in the original workflow but could not be resolved when…
The document presents two strategies as possible low-cost defenses against sharp equity market declines. Time-series momentum is framed mainly for futures: measure each instrument’s return over a lookback period, take a long position after gains or a short…
The document reviews research on the negative convexity of periodically rebalanced stock and bond portfolios. Rebalancing maintains target weights, but it repeatedly sells assets that have gained and buys those that have lagged. When stocks and bonds move…
The document introduces hyperparameter optimization for machine learning models used in trading strategies. It describes how choices such as a model's learning rate can change the path taken while minimizing a loss function, which may have multiple local…
The document explains how survivorship bias can affect stock strategy backtests. A universe built from stocks that exist today may omit companies that have since delisted, leaving historical tests with a stronger-looking set of securities than investors…
This overview explains why high-frequency research systems process tick snapshots, trades, and order data with distributed computing. It describes the challenge of loading very large datasets and the time required to clean and analyze them on a single…
This weekly report summarizes several equity factors, including liquidity, momentum, price-volume, growth, and profitability. It says factor performance was generally subdued with a modest overall decline, while momentum was slightly positive and growth had…
This note explains code equivalents for three functions associated with the Alpha191 factor set. Count(a, n) is illustrated by converting a condition—whether the current close exceeds the previous close—into 1 or 0, then summing those values over a rolling…
This brief BigQuant forum exchange addresses how to build a simple stock screen and output the resulting stock codes as a list. The replies explain that feature extraction, an input feature list, and a code list should be used together, and that filtering…
The document summarizes research on 97 factors reported in finance, accounting, and economics journals. It compares long-short portfolio returns across the original sample, a later out-of-sample period, and the period after publication to examine whether…
The document describes an attempt to reconstruct the dollar-renminbi daily fixing mechanism and estimate the effect of its countercyclical adjustment. The proposed baseline model combines currency basket weights with separate filters for appreciation and…
This overview surveys approaches to combining strategies and allocating capital across assets. It traces the progression from fixed mixes and mean-variance optimization to Black-Litterman allocations that incorporate investor views, endowment-style exposure…
This research report proposes splitting a stock’s recent return history according to average trade size. For each stock, it ranks the past N trading days by daily turnover divided by trade count, sums returns separately for the higher and lower halves, and…
This essay presents four behavioral principles for discretionary stock trading: wait for clearer opportunities, control emotional reactions, make plans before the session, and manage exposure to preserve capital. It argues that avoiding low-quality trades…
This stock-ranking strategy adds recent price-limit activity to a set of price, volume, technical, and financial features. The document explains that the limit-status field uses separate values for limit-down, no limit, and limit-up days; summing the last…
This brief support exchange concerns an error encountered while using an SVR model and asks whether the model needs only training data and a feature list. The response identifies a more immediate issue: the upstream module output is missing two named…
The document introduces quantitative timing as a way to choose when to buy or sell assets using historical prices, market indicators, and mathematical models. It describes moving averages, including simple and exponential forms, as tools for tracking price…
The document presents a BigQuant example for trading the CSI 300 index futures contract from a user-supplied calendar of daily long, short, or no-trade directions. It checks minute data for entry windows around 10:00 and 14:00, targets a specified fraction…
The post shows a BigQuant parameter-tuning setup in which a grid of candidate values is supplied for a model parameter, with examples of feature and tree-count settings. It then defines a scoring function that reads the final Sharpe value from a backtest…
This brief report summary proposes a two-stage equity-selection process. A fundamental model first creates a preliminary stock universe, and a deep learning model then screens that universe again. The stated aim is to improve excess returns while keeping…