Z Systems, CBI, New Versions, Reinvesting, & Withdrawing

Hi jcl.

How should the Z Systems (and other strategies) CBI be handled when new Zorro versions are implemented; and when funds are reinvested and/or withdrawn? Specifically:

a) New version: Say I start a Z system with Capital at 5000 based on Test result CR and whatever safety margin I want. A new Zorro version comes out, and to maintain the same CR and safety margin I need to set Capital to 7500. Do I continue to use the old .dbl file? Or do I copy the new .dbl file from the Test on the new Zorro version to the Production Data directory?

b) Reinvesting: Say I start a Z system with Capital at 5000 based on Test result CR and whatever safety margin I want. The Z system does wonderfully, makes a lot of $$$, and I want to reinvest by increasing Capital to 6000. Do I continue using the same .dbl file? Or do I run a new Test to reflect the reinvestment and copy the new .dbl to the Production Data directory?

c) Withdrawing: Say I start a Z system with Capital at 5000 based on Test result CR and whatever safety margin I want. The Z system does wonderfully, makes a lot of $$$, and I want to withdraw (of course per the square root rule ). Do I continue using the same .dbl file? Or do I run a new Test and copy the new .dbl file to the Production Data directory to account for the withdrawal?

Thanks.

As long as the .dbl file was created with the same initial Capital slider setting, you need not recreate it when you change the Capital later at runtime. This is compensated.

Thanks jcl.

I can see that this makes sense for the c) question - as long as I don't withdraw more than the square root rule specifies then I'm still within the Test parameters and the CBI should be good using the original .dbl file.

Re. b), I can also see how that makes sense for reinvestment, for basically the same reason. But does the compensation also work if I add totally new Capital and change the slider for that reason? Or withdraw more than the square root rule allows?

Re. a), I think it hinges on what "the same initial Capital slider setting" means. Different Zorro versions have produced very different results for the same Capital (previously Margin) value. If version A Tests to a Capital value of 5000 for a CR of 4000, and version B Tests to a Capital value of 10000 to get the same CR of 4000, when I upgrade from A to B, which .dbl file do I use? The original one from version A? Or the new one from version B? It seems to me since I'll be changing the Capital slider from 5000 to 10000 to maintain the same CR, I should also replace the .dbl file. Is this correct?

Lastly, a d) question: Since the CBI is based on equity, how does a safety margin figure in? For example, if Test gives a Capital setting of 5000 for a 4000 CR, but I want a 2x safety and fund the account with 8000 doesn't this distort CBI using a .dbl file from the 5000 Test?

Thanks.

The CBI measures drawdown probability. Funding or withdrawing from your accounts have no effect on the CBI. The .dbl file must be from a backtest with the same initial capital and asset list as in live trading.

Thanks jcl. I believe the exact definition of "same initial capital" needs to be clarified when upgrading from one Zorro version to the next. Using a specific example:

On Version A, Testing Capital=10000 gives CR=4000. I fund the account accordingly and start Trading with the Capital slider set to 10000. I use the .dbl from this Test for the CBI.

Version B appears. I find via Testing Capital=10000 gives a CR much greater than 4000 - in fact way more than I funded the account with. I need to set Capital=5000 to get the same CR=4000.

When I upgrade to Version B, I therefore reduce the Capital slider from 10000 to 5000.

In this example, what is the "same initial capital"? 10000 or 5000? Do I continue using the .dbl file from the Version A Test with Capital=10000? Or do I change to the .dbl file from the Version B Test with Capital=5000?

It seems to me since Version B has required reducing the Capital slider to 5000 as a consequence of the version change rather than reinvesting, withdrawing, etc. that I need to use the .dbl file from the Version B Test. Is this correct?

Thanks.

Hi jcl. Please confirm my understanding is correct and I should use the Version B Test .dbl file.

Thanks.

Yes, when you change to version B, then use the .dbl file from version B.

I have some uncertainties regarding CBI, too.

1. The Z strategies (at least Z12) always sets the Capital slider to 2000 at the very beginning when you start trading. Immediately I set it to the desired value then. But the question is: Doesn't that mean that we *always* need to have backtested with Capital slider at 2000 beforehand to get a correct .dbl file and thus meaningful CBI calculations?

2. If we increase the Capital slider later as I understood this is compensated. But if we stop trading at some point and resume is there a need for a new test run with the then applicable Capital slider? Or does the CBI calculation remember that the old .dbl file was created with the old Capital setting and compensate that even across trading sessions?

3. Is there a way to see which initial Capital setting was used? Or more precicely: Which Capital setting the CBI calculation takes into consideration for the available .dbl file.

Thanks.

jcl, do you have some insight about these topics you could share, please? Also some clarifications in the manual would be helpful I think.
Thanks.

You must start with the same slider position in the backtest and in live trading. In live trading it remembers your last slider position from the previous session, so run the backtest with the same position. When you stop and resume, you must indeed run a new backtest with the current slider position. I think the initial slider position is not displayed.

Ah, I see! Thanks, this makes it clear and also renders my question 1 irrelevant, indeed.
In my opinion it would make sense to add the clarification to the manual, e.g. to
http://manual.zorro-trader.com/ddscale.htm
http://manual.zorro-trader.com/zsystems.htm
The pages currently state "run a backtest with the Capital slider at the same position as in live trading at session start" and "Before starting live, run a backtest with the Capital slider at the same position as in the trading session".
So, if the reader knows that "stop and resume" means a new "trading session", then it is clear implicitly. But explicitly explaining the need of a new backtest with the current slider position after stop and resume would make that more unambiguous.

Yes, I'll make that more clear in the manual.

Just to confirm then, I'm trading live on a VPS that cannot really run backtests. So the procedure then is to run a backtest on my desktop PC whenever I want to change the slider, copy the resulting DBL file up to the VPS and then adjust the slider on the VPS?

I would say yes.