MCFAM Seminars - Towards a Better Vega, Part I: A Primer on Upsilon
Speaker: Brian Huang, UMN
Abstract: Vega is one of the most familiar sensitivity measures in derivatives, but its interpretation is typically tied to implied volatility and therefore comes with important limitations. In particular, implied volatility is a derived parameter rather than an intrinsic object of the pricing problem, making it unclear whether Vega provides the most natural notion of sensitivity for option prices.
This talk introduces Upsilon, a sensitivity measure defined through Wasserstein on probability distributions, and explains the mathematical framework that underlies it, including why metricization matters for sensitivity analysis, why Wasserstein is a suitable metric in this setting, and why Upsilon may provide a more natural foundation than Vega for measuring uncertainty in option pricing.