I am a Ph.D. candidate in Economics at Southern Illinois University Carbondale (SIUC). I study how what we see—color, composition, texture—and where we are in the market cycle shape prices in digital markets. Empirically, I like models that are interpretable first and scalable second: hedonic mixed-effects, Bayesian dynamics, and carefully engineered machine learning and deep learning features.
[Download my CV] · Email: samiha.bintetariq@siu.edu
Pixels to Prices: Visual Traits, Market Cycles, and the Economics of NFT Valuation
I assemble 94,039 trades from 26 major Ethereum NFT PFP collections and extract 196 visual descriptors (palette structure, saturation at the focal region, curvature, texture; plus compact CNN/Gram - deep learning components) from 24,148 unique images. I show that transparent, human-readable traits carry robust premia, while “black-box” deep embeddings add little once explicit traits enter while brand premium plays a vital role. Trait premia are state-dependent—they expand in hot markets and compress or flip in cool ones—so I estimate a cycle-aware Bayesian dynamic hedonic to track those shifts.
[Job Market Paper: Pixels to Prices / arXiv]
My interest in this research project grew out of two habits: studying economics by day and sketching and painting at night. I post art anonymously online, and a scammer tried to rope me into an NFT scheme—ironically that was my first introduction to NFTs and the markets around them. I honestly didn’t think art and economics could come together—until my supervisor suggested an interesting idea: pick colors from images, extract features, and build an economic model. And that showed me how art and economics could meet!
Clarity over complexity. I keep things understandable and effective, adding complexity only when it adds value and serves a clear purpose.
Research interests
Microeconomics; Macroeconomics; Monetary Economics; Financial Economics; Applied Econometrics; Behavioral Economics.
Visit my Research page to explore more of my work!
My approach in teaching is to build intuition first, then layer in algebra and, finally, code—so students can intuitively explain a result before they estimate it. Courses assisted include Labor Problems (ECON 310), Money & Banking (ECON 315), and Microeconomic Theory II (ECON 540B).