APA & AGA

The home of Algorithmic Pricer Acceleration (APA) and Algorithmic Greeks Acceleration (AGA)

 

Let’s say that we have a pricing function that we need to run thousands or millions of times in the MC simulation. We can accelerate the process by…

 

  1. The pricing function is passed through MoCaX
  2. The function is explored in a small number of “anchor points”
  3. APA returns an accelerated version of the function
  4. AGA creates an accelerated version of its derivative functions
  5. These accelerated functions are then used in the Monte Carlo simulation.

WHAT MAKES MOCAX SPECIAL


APA DELIVERS PERFECT PRICING

APA only needs to run the Original Function around 10 times per dimention to deliver a typical accuracy of 10-5. Only 20 calls typically delivers 10-15 (machine precision).

APA DELIVERS NANO-SECOND PRICING

The evaluation time of the accelerated functions is of a few nanoseconds, however slow the original function is.

AGA DELIVERS PERFECT GREEKS

Once APA has been run, AGA delivers the Greeks of the pricer with the same precision (e.g. 10-5), same speed (e.g. 10 nanoseconds).

Importantly, AGA delivers the ultra-accurate and ultra-fast expression of the Greeks without having received their explicit form.

MoCaX solves a long-standing probleim in a completely new way

See Examples