Middle East Technical University

How people decide when the odds are moving.

I build instruments that models it, and I write things out to find out what I think. This site is where both of those live.

Decision science
Behavioral economics
Cognitive neuroscience
Quantitative finance

Work

Research instrument

Dynamic Hazard Rate BART

This instrument makes the risk structure itself a design choice, with eleven hazard families, numerically verified optima, and fully offline scoring.

Hazard families: 11
Scored metrics: 40+
Network requests: 0

Python package2024 — 2026

grubeta

A physics-informed dual-pathway GRU that estimates time-varying CAPM beta while preserving the pricing identity it is estimating — packaged for PyPI with lookahead diagnostics built into the validation path rather than bolted on.

Distribution: PyPI
Validation: Walk-forward
Benchmarks: 3

PythonTensorFlowKerasNumPypandas

Try the task yourself

The study runs in the browser. The balloon task, the risk scale, and the rest of the battery, ending in a reflection of your own results. It takes less than 10 minutes.

bart.aselimyilmaz.com

Thoughts

All →

12 August 2026

Nobody Has Ever Seen Nothing

Every piece of knowledge we possess sits on the side of existence. We have nothing from the other side, and this asymmetry constrains the claim that something was created out of nothing more severely than it constrains the person doubting it.

Released as Milesaway

I also produce music

Choosing to dance anyway is the most honest protest he’s got.

Releases and where to listen

Who is writing

I study at Middle East Technical University in Ankara. My work sits across decision science, behavioral economics, cognitive neuroscience, and quantitative finance.

selim.yilmaz@metu.edu.tr