Risk Overlay

[greenn1313]

Hi all!

I'm backtesting Rob's system with risk overlay and without. And I receive 100% identical annual returns:
annual_return = system.accounts.optimised_portfolio().percent.annual

Is it normal?

Here is risk overlay I added to config.yaml:
risk_overlay:
max_risk_fraction_normal_risk: 1.4
max_risk_fraction_stdev_risk: 3.6
max_risk_limit_sum_abs_risk: 3.4
max_risk_leverage: 13.0

[robcarver17]

Here is how you should use the risk overlay (in the relevant blog post and also in the new book):

• take all the risk overlays out (or set them to infinity)
• run a backtest without them
• plot the varius things within it (from here where self is system.portfolio):

normal_risk = self.get_portfolio_risk_for_original_positions()
shocked_vol_risk = self.get_portfolio_risk_for_original_positions_with_shocked_vol()
sum_abs_risk = self.get_sum_annualised_risk_for_original_positions()
leverage = self.get_leverage_for_original_position()

• decide based on distribution, or personal taste, where you want the limits to fall (not forgetting to normalise normal_risk and shocked_vol_risk for your personal risk target)
• put the limits in

voila

[greenn1313]

Thank you. All looks good, but I have a problem:

leverage = system.portfolio.get_leverage_for_original_position()
leverage.max()
3376807.843448575

Is it something wrong with multiple/adj prices (I use standard csv files)?

[greenn1313]

[robcarver17]

Look at your position plot over time and see if you can work out which instrument is causing that spike

I'm afraid I haven't got time to teach you how to debug :-)

[greenn1313]

Ok, sorry :)

I tried different values for risk_overlay, turned it on and off and I always receive the same result. My guess is that risk overlay doesn't work with optimised_portfolio()