Hej! I'm a Swedish master's student of mathematical statistics at Lund University. My interests center on latent process-models and Monte Carlo methods, primarily as applied to financial statistics. But I get around. On this page I write about my recent projects with less-than-publishable rigor.
Prediction markets of the type "Will Kahmeini be ousted before 2025?" don't try to forecast the outcome of an event, but its timing. I present the first model of such prediction market prices, based on methods developed for defaultable corporate bonds.
Who will win the next election? Prediction markets turn such questions into tradable contracts, whose market prices reveal collective beliefs about uncertain events. I evaluate the state-of-the-art model of such markets, which models these contracts as derivatives on latent processes. Turns out it's not great!
Remember the game Mastermind, where you have to guess a secret color combination? I explain how we can use information theory to play it optimally. Play around with the interactive game!
Regularized regression techniques allows us to improve the crop yields models of the Swedish Department of Agriculture by up to 40 %! By modeling multiple regions simultaneously, we can get even better results. This was my bachelor's thesis, written in Swedish with a summary in English.