Hello,
It seems that the ATR parameter in "Percent Volatility" position sizer is calculated based on the Backtest Data Scale instead of always being based on the Daily scale.
For example, running the same code on the same data but on different scales (10-minutes, 30-minutes, 60-minutes, etc.) generates same positions but with different sizes which of course are all incorrect since only using daily ATR is meaningful for position sizing.
It seems that the ATR parameter in "Percent Volatility" position sizer is calculated based on the Backtest Data Scale instead of always being based on the Daily scale.
For example, running the same code on the same data but on different scales (10-minutes, 30-minutes, 60-minutes, etc.) generates same positions but with different sizes which of course are all incorrect since only using daily ATR is meaningful for position sizing.
Rename
QUOTE:
It seems that the ATR parameter in "Percent Volatility" position sizer is calculated based on the Backtest Data Scale instead of always being based on the Daily scale.
That's right. All position sizers in general operate on the backtest data scale. There's nothing wrong with this. It has never been a requirement to calculate the ATR on a compressed scale perhaps because there was no good reason for that?
QUOTE:
only using daily ATR is meaningful for position sizing.
This is a moot point. Some trading strategy may prefer the intraday ATR and another approach could involve Weekly ATR. Isn't it?
Your Response
Post
Edit Post
Login is required