JBT release 4.05

[nonlinear5]

Project home: http://code.google.com/p/jbooktrader/
Latest release download: http://jbooktrader.googlecode.com/files/JBookTrader-4.05.zip
Release Notes: http://code.google.com/p/jbooktrader/wiki/ReleaseNotes
Market depth historical data (purchased from CME):
http://jbooktrader.googlecode.com/files/ES-CME-Jan8toApr17.zip
Market depth historical data (recorded from IB):
http://jbooktrader.googlecode.com/files/ES-IB-May30ToJuly23.zip

-- Historical data format changed. Recorded depth balance is now
"adjusted mean balance", as proposed by Dyno. Reference:
http://groups.google.com/group/jbooktrader/browse_thread/thread/9df17cd7245b0225#

-- Adjusted mean balance is used in all modes, including back testing,
forward testing, optimization, and trading.

-- Both CME and IB dataset have been re-generated to reflect the new
way of calculating adjusted mean balance. CME dataset is password-
protected. The password is the same as the one you used before. IB
dataset can be accessed by anyone.

-- Added new sample strategies

-- Please note that all strategies (both sample strategies AND the
custom strategies which you may have) will need to be re-optimized! I
only optimized Balancer and VelocityRider for this release, and only
using the IB dataset.

[yueming]

Just so you know, there is a compilation error in BalanceRSI in the
new version, the declared package name is wrong, nothing serious
though.

[nonlinear5]

> Just so you know, there is a compilation error in BalanceRSI in the
> new version, the declared package name is wrong, nothing serious
> though.

Yes, thanks. I use IntelliJIdea as my IDE, and sometimes it's too
liberal with the code. This will be fixed in the next release.

[Shane]

Is there any way to develop/test JBT on the same computer you are running JBT?  Once 9am rolls around, I can't test my changes, because I am forward testing strategies.

[Eugene Kononov]

Is there any way to develop/test JBT on the same computer you are running JBT?  Once 9am rolls around, I can't test my changes, because I am forward testing strategies.

You can remove the "JBT already running" code check and run two JBT instances from two different directories on the same machine, but I highly discourage that. For one thing, when JBT is running in the forward test or trade mode, it's interacting very heavily and frequently with TWS. If you are running some other CPU- or memory-intensive process on the same machine, this interaction may be delayed. Another thing is that JBT makes use reflection to dynamically load the classes, so another instance of JBT may interfere.

The bottom line is, JBT is best left alone when running in the "trade" or "forward test" mode.

Note that forward testing always produces *exactly* same results as backtesting using the recorded data for the same time period. Since the IB data is updated on the project "download" page, you can substitute forward test with the back test using that data.