top of page

99 Problems, is Gamma still one

Second order derivatives - errors introduced under current techniques
 

As can be seen in the Cost Reduction tab, Traditional Gamma​ gives decidedly questionable results for both the delta bucketing and PNL predicts, even showing negative gamma PNL for benign curve moves.

Traditional gamma is approximated by assuming a parallel shift in the yield curve and then re-calculating the delta. In the attached paper we analyze the risk and PL breaks of standard second order simplifications, and conclude that the current methods used to estimate gamma (ie ladders and interpolation) can lead to serious bucketing errors (leading to the risk and PL breaks). 

In the analysis, PLATSON also quantify how much of the error is due to this simplifying yet erroneous parallel assumption.

 

PLATSON reduces the non-linear PL estimation errors by roughly 95% without requiring additional computing resources

bottom of page