The 0.53 and 1.12 from the two regime script can be checked without any simulation. With the hedge ratio set to 0.85 times gold vol over silver vol, the hedged book's vol as a share of gold's is the square root of 1 plus 0.85 squared minus 1.7 times rho.
At rho 0.85 that gives the square root of 0.2775, about 0.53, which matches. At rho 0.30 it gives the square root of 1.2125, about 1.10, so the 1.12 is that number plus sampling noise from roughly 400 event days.
The more useful output of the formula is the break even. Setting the ratio to 1 gives rho equal to 0.7225 over 1.7, about 0.425, so any minute window where correlation falls below that makes the hedge noisier than holding gold alone.
That turns the open question about where 0.85 and 0.30 come from into a simpler test. Measure rolling one minute correlation around releases and count how often it drops under 0.43, rather than trying to pin down the exact event value.
It also assumes the hedge ratio stays fixed. Refitting it inside the event window changes the numbers, so the backtest should say which it uses.
Simulation Monte Carlo on corelation gold/silver price
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm