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.
5,018 documents
The document defines a Bregman divergence from a differentiable convex function and asks how to find the point minimizing its expected divergence from a distribution. The answer writes the expected divergence as an objective in the candidate point, separates…
This explanation distinguishes sample uniqueness from sample weights in financial machine learning. A model’s sample weights tell it how strongly to emphasize observations during fitting. Uniqueness measures how much an observation overlaps with others, and…
The document discusses why quantitative research interviews often use elementary probability puzzles involving cards, dice, coins, and conditioning instead of direct financial modeling questions. One response criticizes this practice, arguing that research…
The document surveys several ways discrete mathematics and graph theory appear in quantitative finance. Network analysis can model links among banks and trace how credit stress spreads, examine industry concentration and stock returns, and study how…
The document asks how to generate draws from a multivariate copula conditional on known values of some coordinates. It considers Gaussian or Student copulas with empirical marginal distributions and seeks a numerical procedure when the margins do not have…
The document describes an introductory stock-market assignment: devise a score from information about a stock, then compare that score with returns over several future horizons. The student has calculated a correlation coefficient between score values and…
The document asks how to predict near-term best bid or ask price direction from high-frequency limit order book data. One answer recommends reconstructing the actual book from a venue’s full order log when available, rather than relying on a simulated book.…
The document asks whether normalized prices, returns, or nominal price changes are preferable inputs and targets for a deep neural network forecasting daily stocks or indices. The response does not recommend one representation or provide a literature review.…
The document considers whether a Hurst exponent different from one half justifies fractionally differencing stock prices before forecasting. It cautions that an estimated exponent is a statistic subject to sampling uncertainty: even data generated by a…
The document explains why C++ skills can matter for quantitative researchers at high-frequency trading firms even when Python is more convenient for exploration, data analysis, and machine learning. It distinguishes research modeling from work that…
This note addresses whether nonsignificant AR terms in a fitted ARFIMA model can be treated as redundant across candidate ARFIMA-GARCH combinations. It distinguishes sequential fitting, where an ARFIMA model is estimated first and its residuals are then…
The document surveys ways random forests can be used in quantitative finance. A forest combines decision trees built from resampled observations and subsets of features; it can classify discrete outcomes or predict numeric targets. Suggested applications…
The document explains that the differential Sharpe ratio used in online trading models is a derivative with respect to the decay rate of exponential moving estimates of return moments. It defines a running estimate of the first moment and a running estimate…
The document addresses how to interpret weekly and monthly stock return predictions from regression trees. The training targets are recorded on Fridays: weekly returns for each Friday and monthly returns on the last Friday of each month. The question is…
The document considers whether stock prices should be z-score standardized or log transformed before calculating correlations or clustering distances. It explains that correlation is already invariant to linear shifts and positive rescaling, so z-scoring…
The document surveys ways to search historical price data for segments resembling a recent market path, then use what followed those matches to inform a forecast. Suggested approaches include nearest neighbors, clustering, seasonal pattern analysis, and…
The document considers how to evaluate a trading rule that enters an asset only when a metric exceeds a threshold. It compares the selected observations’ returns with the asset’s overall returns and illustrates the idea with correlated simulated data and a…
The document describes a proposed short-term stock strategy built around predicting whether a purchase can reach a small profit target within a fixed holding window. Its historical-data program labels potential entries according to whether a market purchase…
The discussion corrects a misunderstanding about statistical evidence for positive average returns. Rejecting a null hypothesis that mean daily return is nonpositive does not imply every day or investment period must have a positive return; it supports a…
The document asks whether a Random Forest can be applied directly to a non-stationary financial series or whether the data should first be differenced. The reply warns that fitting the model to price levels can lead it to learn those levels and overfit, and…
The document outlines how to investigate whether macroeconomic indicators such as growth, inflation, and monetary policy help forecast silver prices. It recommends first considering the economic relationships, timing, and functional form linking those…
This question concerns whether borrower loans in a Lending Club dataset can be segmented into prime and subprime categories using the available grade and sub-grade fields. The dataset excerpt includes many borrower and loan characteristics, alongside the…
The document asks how to adapt a noisy high-frequency price-direction labeling method to trading at a more ordinary frequency. The cited method uses the stock’s bid-ask midpoint and compares averages over nearby time windows, assigning upward, downward, or…
The document discusses methods for estimating the early exercise boundary of an American put under geometric Brownian motion. For continuous or proportional discrete dividends, a binomial tree can approximate the boundary; fixed discrete dividends may be…