Alessandro Baretta

Non-Parametric Conditional Density Estimation: I developed this unique ML algorithm to estimate the conditional probability distribution of a variable, typically the loss or the payout of an insurance contract or a financial derivative. To my knowledge this is the first, and possibly still the only, Machine Learning Algorithm that directly supports efficient risk underwriting.

Market-wide Covariance/Correlation for Intraday Portfolio Hedging: Hedging a financial portfolio requires offsetting some of the risk by entering into anticorrelated trades. In turn this requires accurate knowledge of the volatility and correlation structure of the market. Directly estimating the volatilities and the correlation matrix of the broad market is generally considered to be an intractable problem; however, I have built a 35 million parameter machine learning model that accurately measures the correlation and volatility of the entire US equities market.