The article treats quantitative finance sociologically. It is argued that although mathematical modeling dramatically changed the nature of modern finance, it did not eliminate sociality from financial markets. However, the traditional sociological approach to markets, with its focus on personal social ties and networks, should be transformed as well. Anonymous financial models have not replaced social cues; instead, those models are used socially. This means that investment bank traders use formal mathematical models to predict the decisions of competing traders. Moreover, traders use these models as a reflexive tool. This reflexive modeling creates distributed cognition or dissonance, which helps traders to avoid errors and financial losses. However, this mechanism has an implicit, but very dangerous, drawback: by creating cognitive interdependence, it may lead to massive errors and huge losses in an entire market. While the dissonance effect prevents individual errors, the resonance effect gives way to a collective (market) disaster. Two relevant case studies are considered. Both cases refer to anticipated mergers, but while one of the mergers was correctly predicted, the other one was not. Thus, the first case illustrates the bright side of financial models, while the second shows their dark side. The article contributes to the discussion of the new forms of sociality primarily associated with financial models. The study is based on three years of field research in a major investment bank focusing on traders’ everyday practices.