Abstract
The article proposes a theoretical and methodological approach to analyzing investment behavior in the cryptocurrency market, based on the ontological distinction between risk and uncertainty. This distinction, introduced by F. Knight and experimentally confirmed by D. Ellsberg, serves as the starting point of the study, since most classical decision-making models are oriented toward situations of risk rather than ontological uncertainty. It is substantiated that the cryptocurrency market operates under conditions of uncertainty. None of the three necessary preconditions for applying the probabilistic approach (sufficiently long history of observations, stationarity of processes, and the existence of mechanisms for convergence to equilibrium) are met in this market. Four types of uncertainty are identified – regulatory, technological, market, and informational – each with its own sources and mechanisms of influence on investment behavior. Standard risk assessment tools (VaR, ES, GARCH models) prove to be of limited effectiveness in this context. Since the classical prospect theory of D. Kahneman and A. Tversky presupposes the existence of (at least subjective) probabilities of outcomes, its direct application to the cryptocurrency market is limited. A modification of this theory is proposed by introducing an ambiguity aversion parameter a∈[0,1], which reflects the worst-case scenario strategy in accordance with the concept of maxmin expected utility. The modified probability weighting function makes it possible to model the behavior of cryptocurrency market participants under conditions where the classical premises of prospect theory are violated. A two-dimensional typology of investors is developed, based on two criteria: the ability to tolerate uncertainty and the time horizon. Four types of investors are identified: conservative day trader, memecoin trader, institutional HODLer, and DeFi farmer. The proposed classification further develops the authors' previously published typology, supplementing it with an indicator of uncertainty tolerance and accounting for the ability of agents to change their behavioral model in response to market conditions. For empirical verification of the developed theoretical approach, a multidimensional method for operationalizing uncertainty is proposed, combining the analysis of textual data (news, social media), on-chain metrics (active addresses, liquidation volumes), and exchange indicators (spreads, market depth). Three empirical strategies for testing the hypotheses are outlined: event study analysis, econometric modeling of transaction data, and a combination of surveys with the analysis of real crypto wallets.