The dual trigger system is brillant. Most risk managment frameworks miss the erosion failure mode entirely, focusing only on catastrophic shocks. Your Student-t likelihood approach for handling fat tails is critical, a Gaussian assumption would trigger false positives on every vol expansion. One question though: have you tested this on mean-reverting strategies where drawdowns are a feature not a bug? The erosion trigger might fire prematurely in those cases.
One of the most scientific articles on substack!
The dual trigger system is brillant. Most risk managment frameworks miss the erosion failure mode entirely, focusing only on catastrophic shocks. Your Student-t likelihood approach for handling fat tails is critical, a Gaussian assumption would trigger false positives on every vol expansion. One question though: have you tested this on mean-reverting strategies where drawdowns are a feature not a bug? The erosion trigger might fire prematurely in those cases.
Hi Mate! For the moment I tested only in a certain models based on factors. But if the PnL from the MR systems is smooth, the result should be similar