Harry Crane

Harry Crane is Professor of Statistics and Affiliated Faculty in the Graduate Program in Philosophy at Rutgers University. He is co-founder of Researchers.One, a decentralized research platform for peer review, scholarly publication, community building and research dissemination. Crane is currently Fellow at the London Mathematical Institute, and has previously held positions as Chancellor's Excellence Scholar at Rutgers, Co-Director of the Graduate Program in Statistics at Rutgers, Visiting Scholar in Mathematics at UC Berkeley, Research Associate at the RAND Corporation, and Research Fellow at the Foreign Policy Research Institute. Harry completed a PhD in Statistics from the University of Chicago and BA in Mathematics, Economics, and Actuarial Science from the University of Pennsylvania.

: With Ryan Martin, I co-founded researchers.one, a decentralized research platform for peer review, scholarly publication, community building and research dissemination. Researchers.One facilitates publication and peer review, journal publication and conference hosting. Since inception, Researchers.One has pubilshed hundreds of articles, onboarded a number of academic journals across multiple disciplines, hosted conferences attended by researchers and scholars around the world, and provided an outlet for research to be disseminated and discussed openly. Visit researchers.one for more information.





Book on Probabilistic Foundations of Statistical Network Analysis discusses foundational aspects of modern statistical network analysis.

Research Interests

Current research interests: Developing methods and designing mechanisms for effective decision making. The primary question guiding this research is understanding mechanisms by which we gain insight into uncertain outcomes. Two predominant approaches are (i) using statistical techniques to project future outcomes based on past information and (ii) designing markets in which trading activity about a future event (such as an election outcome or key economic indicator) provides insight about the future event. Over time this research program has led to practical insights about probability, uncertainty, risk, and complexity.

This research program divides into three complementary components: (I) prediction markets, betting exchanges, and sports analytics, (II) data analysis and the limitations for statistical methods in complex settings, and (III) the role of intuition and common sense in effective decision making. More details on each of these are given below.

Prediction markets, betting exchanges, and sports analytics. Markets and exchanges are the natural setting for this research in a few respects. I focus specifically on event contract markets (prediction markets and betting exchanges) because they are markets for well-defined future events, making them relatively straightforward to study for properties such as efficiency and accuracy.

I. Optimal design of markets. Well-designed markets provide the liquidity and other means necessary to suit the needs of traders as well as provide a reliable price signal to market observers. Fee structures and other distortive frictions can disrupt markets from serving their stated purpose. I collaborate with exchange operators on market structuring and new product design.

II. Market efficiency and price discovery. Well-designed prediction markets are useful for aggregating information and representing future uncertainty by a meaningful quantity in the present. My research on political prediction markets (especially PredictIt) studies the accuracy of PredictIt electoral markets compared to more widely publicized polls and data analytics outlets (e.g., FiveThirtyEight).

III. Adverse selection, market making, and complex market dynamics. The fundamental paradox of trading is that every potentially profitable trading opportunity can also be interpreted as evidence that you're wrong. On the one hand, it's impossible to trade profitably without finding what appear to be "profitable trades". On the other hand, the tendency for markets to be efficient means that when you find a "profitable trade" it indicates that there is something missing from your original opinion. Understanding how to incorporate this component into a profitable trading strategy is an area of ongoing fascination and research.

In addition to political prediction markets and exchanges for other events, I also consult and study a number of problems from advantage gambling, risk analysis, and sports analytics. Some of this work has been in collaboration with Analytics.Bet, where I have developed course material on how to use statistical and data science techniques for profitable sports betting.

Data analysis in complex domains. Quantitative methods are often a core component of many trading strategies, business decisions, medical research, etc. Over time, these approaches have proven themselves useful in a wide range of applications. But they are not well-suited to all problems, especially those involving complex processes (such as those with high levels of dependence and feedback loops). This research studies the limitations of traditional statistical methods for modeling complex data (e.g., network data) and develops new approaches in those situations. This is the topic of my first book on Probabilistic Foundations of Statistical Network Analysis. Invariance principles are a common theme in this work, mostly in the concept of exchangeability and its variants (partial exchangeability, relative exchangeability, edge exchangeability, relational exchangeability), and their limitations for modeling complex data structures.

Probability, Intuition, and Common Sense. As statistical techniques and analytics become more widespread in decision-making throughout business, government, policy-making, medicine, law, science, and sports, the role of intuition and common sense is often diminished. While there are some clear (quantifiable) benefits to analytical approaches in these fields, this trend is accompanied by diminished critical thinking ability, especially in situations involving severe uncertainty and systemic or even existential risk. This research investigates the practical consequences of over-dependence on data and anayltics, and also studies the foundational role of intuitive reasoning in rigorous, logical decision-making. This is the topic of my forthcoming book on Probability, Intuition, and Common Sense.