Correlation Analysis Among Asset Classes
Modern robo-advisors utilize Mean-Variance Optimization (MVO) to build portfolios. However, the integrity of the model depends on the correlation assumptions. Our research tracks how these platforms adjust for "correlation break" events—moments where traditionally diverse assets begin to move in lockstep.
The transition from raw data to a risk score is not a straight line. It involves weighting variables such as age and income against subjective answers regarding market volatility. A failure to calibrate these weights leads to "profile drift," where the investor's actual risk capacity is ignored in favor of their stated (often emotional) risk tolerance.
We focus on the intersection of data integrity and automated decision-making. By auditing the "Logic Gates" of these platforms, we determine whether the algorithm prioritizes long-term institutional stability or short-term retail retention.
