I love fantasy football, so I've spent a lot of time building models to predict player outcomes. The pipeline produces a full weekly scoring distribution for every player: tree models estimate the mean and variance, and a flexible distribution family captures the shape, with parameters that adapt to the player's projected scoring level. Those distributions feed a Monte Carlo simulation that correlates same-game players and layers in persistent season-long mean shifts to model injuries, role changes, and hot streaks. The output is calibrated outcome ranges, finish probabilities, and value-over-replacement draft rankings.

Explore the 2026 rankings board of projections for the upcoming season. This will hopefully be moved to a more permanent location in the future, but for now you can find it here. A brief reminder that these projections are NOT my opinions, they are purely the output of a statistical model with no human adjustment.

Here you can find implementations of the algorithms we developed in the paper Improved methods for finding imaginary quadratic fields with high n-rank. For a given prime p, these can be used for very rapid generation of discriminants of quadratic number fields with p-rank at least 2.

Here you can find the code used to complete the proof of Theorem 1.1 in the paper Square-free smooth polynomials in residue classes and generators of irreducible polynomials. The algorithm takes as input a prime power q and a positive integer n. Then for each irreducible polynomial f of degree n over the finite field of order q, the code searches for a square-free representative for each residue class modulo f with no factors of degree equal to or exceeding the degree of f.