I wanna to design a method to monitor the whole system's risk/exposure,when the risk/exposure is too big,the method could limit the heat of the system by reducing the position size.
but I don't know the exact direction,could anyone here shed me the light,please?
thanks in advance.
A method to monitor and adjust system's risk?
Trading Recipes lets you do this with one statement. Supposing you want to reduce each position by 10% if the total risk across all positions ever becomes more than 30%, you would add this line to your system:
I'm sure it's equally simple to do in other software packages.
Code: Select all
IF (TOTALRISK > 0.30) THEN MULTIPLIER = 0.90
-
- Roundtable Knight
- Posts: 207
- Joined: Mon Feb 23, 2004 9:13 am
- Location: Vermont, USA
- Contact:
My estimate of risk is as follows:
Calculate the difference between the net liquidation value of your account right now versus the account value if all your stops were hit. The difference in these 2 numbers is how much risk you are currently taking.
Note that this assumes you will be able to get out of positions close to your stops. It also assumes you do actually have a stop on each and every position (real or mental). If you have a short position with no stop, then your risk cannot be calculated since there is no theoretical limit to how high a price can go, and therefore your risk is unbounded. It also makes no allowance for open profits versus unrealized losses - it is simply how much your account can go down from here based on your stops.
Any risk calculation is always an estimate (unless you are long options and have no stop - then your risk is equal to the current option premium) due to slippage on exit or other situations that get you out at a worse-than-anticipated price.
Hope this helps
Paul
Calculate the difference between the net liquidation value of your account right now versus the account value if all your stops were hit. The difference in these 2 numbers is how much risk you are currently taking.
Note that this assumes you will be able to get out of positions close to your stops. It also assumes you do actually have a stop on each and every position (real or mental). If you have a short position with no stop, then your risk cannot be calculated since there is no theoretical limit to how high a price can go, and therefore your risk is unbounded. It also makes no allowance for open profits versus unrealized losses - it is simply how much your account can go down from here based on your stops.
Any risk calculation is always an estimate (unless you are long options and have no stop - then your risk is equal to the current option premium) due to slippage on exit or other situations that get you out at a worse-than-anticipated price.
Hope this helps
Paul
-
- Roundtable Knight
- Posts: 1842
- Joined: Tue Apr 15, 2003 11:02 am
- Contact:
Paul's definition matches TradingBlox internal notion of risk.
TradingBlox will graph risk over the duration of the test. You can access the risk at the system, group or correlated market level if you are using the TradingBlox Builder. You can easily do the same sorts of things as you can with TradingRecipes as well, lighten positions by 10%, etc..
TradingBlox also lets you do something two-pass testing tools like Trading Recipes or Mechanica can't, adjust stops to keep risk levels capped at a certain percentage of your account.
Group or sector based risk limiting is one way of reducing drawdowns that seems to work pretty consistently for many long-term systems.
- Forum Mgmnt
TradingBlox will graph risk over the duration of the test. You can access the risk at the system, group or correlated market level if you are using the TradingBlox Builder. You can easily do the same sorts of things as you can with TradingRecipes as well, lighten positions by 10%, etc..
TradingBlox also lets you do something two-pass testing tools like Trading Recipes or Mechanica can't, adjust stops to keep risk levels capped at a certain percentage of your account.
Group or sector based risk limiting is one way of reducing drawdowns that seems to work pretty consistently for many long-term systems.
- Forum Mgmnt
Last edited by Forum Mgmnt on Wed Apr 19, 2006 9:12 am, edited 1 time in total.