#statistics

The number someone repeated In 1966, five a question that sounds like science fiction: if...

We froze the rules, opened a sealed year of data exactly once, and the result came out below zero.

A Form 144 is the notice an officer, director or large holder files when they plan to sell stock....

A significance checker answers one narrow question: if the two versions of a page were actually...

I built a lab to compare the models running in my agent fleet on per-turn behavior like tool-error rate and re-edits. The numbers were the easy part. The work was everything that would have made a naive comparison lie — pooling two roles that are not comparable, averaging across policy regimes that are different treatments, folding in 871 threads I could not attribute to any model, and reading a bootstrap interval of [0, 0] as precision when it was a tie collapse. This is the list of things I refused to count, and why.

Hamiltonian Monte Carlo: How Momentum Turns Random Walks into Efficient Sampling Most...

Score Matching: Why Learning the Gradient of a Log-Density Beats Learning the Density...

What a strict, honest backtest looks like, and the one small thing that did help.

Measuring whether two models are the same: why held-out data must be truly unseen and why paired windows beat averages. Case: an NPU vs B200 pretraini

If you compare prop firm passing services by headline fee, you will pick wrong about half the time....

A grid trading bot backtest is only useful if it models how grid orders actually fill, charges every...

Bayesian inference is simple to state and hard to compute. Given data, we want the posterior...