Ahh okay. So if I make target_abs = 0.5 then at median strength I’d be at half size and then could cap at -1,1 so that I’m only fully exposed at very strong trends (no margin in this example)
Yeh, you can allocate anyway you prefer. The -1/1 or -2/2 or even -20/20 (Rob Carver's), is irrelevant as what that represents in terms of sizing is then calculated separately. To me, 1/-1 means a normal position, and -2/2 is the maximum position. How I allocate between those two is up to my risk management models.
Makes perfect sense. The way I’ve been testing dynamic position Sizing is in a similar direction, but I didn’t have anything related to median scaling. I’m excited to test this out. Thanks again.
Makes perfect sense. The way I’ve been testing dynamic position Sizing is in a similar direction, but I didn’t have anything related to median scaling. I’m excited to test this out. Thanks again.
Makes perfect sense. The way I’ve been testing dynamic position Sizing is in a similar direction, but I didn’t have anything related to median scaling. I’m excited to test this out. Thanks again.
So, the target_abs is the target size you want the median of the transformed signal to reach.
In this case, I've chosen target_abs = 1, meaning that when the signal is at its median strength, I want to allocate 100% of the intended position size.
The signal is then capped between -2/+2, which means more aggressive signals (beyond the typical) will have stronger allocations (up to 200% or -200% of intended size), but the cap is to prevent risk exposure in the tails to go wild.
Nice article! One question I have, where does TARGET_ABS come from? What does that represent?
Ahh okay. So if I make target_abs = 0.5 then at median strength I’d be at half size and then could cap at -1,1 so that I’m only fully exposed at very strong trends (no margin in this example)
Yeh, you can allocate anyway you prefer. The -1/1 or -2/2 or even -20/20 (Rob Carver's), is irrelevant as what that represents in terms of sizing is then calculated separately. To me, 1/-1 means a normal position, and -2/2 is the maximum position. How I allocate between those two is up to my risk management models.
Makes perfect sense. The way I’ve been testing dynamic position Sizing is in a similar direction, but I didn’t have anything related to median scaling. I’m excited to test this out. Thanks again.
Makes perfect sense. The way I’ve been testing dynamic position Sizing is in a similar direction, but I didn’t have anything related to median scaling. I’m excited to test this out. Thanks again.
Makes perfect sense. The way I’ve been testing dynamic position Sizing is in a similar direction, but I didn’t have anything related to median scaling. I’m excited to test this out. Thanks again.
Hey man, thanks!
So, the target_abs is the target size you want the median of the transformed signal to reach.
In this case, I've chosen target_abs = 1, meaning that when the signal is at its median strength, I want to allocate 100% of the intended position size.
The signal is then capped between -2/+2, which means more aggressive signals (beyond the typical) will have stronger allocations (up to 200% or -200% of intended size), but the cap is to prevent risk exposure in the tails to go wild.