Compares three algorithms for streaming covariance estimation and gives practitioner guidance for online, block, and batch settings.

Greenhouse pre-analysis plan investigating acoustic stimuli as an agricultural production factor.

Cov(X) = (X'X − (1/n)ss') / (n−1). Avoids explicit centering, outperforms numpy.cov.

High-freq retail panel. AT treated items: +13.01 index pts within 12m, +19.42 over full post period.

Sequential two-stage triathlon model. Drafting as multiplicative shifter on quadratic effort costs.

Fast unbiased variance estimator. Introduces the “Bariance.”

Staggered COVID starts as natural experiment for causal drafting effects in open water swimming.

Group size and bandwagon effects in multi-stage competition. Leave-one-out instrument for peer effects.

Strategic .99 pricing and spatial clustering from scraped online bike market data.

One outlier can make a coefficient significant in small samples. OLS vs robust alternatives.