Here is an interesting link to open source code. Perhaps it can be useful for new release: http://code.google.com/p/efficient-java-matrix-library/
It has Kalman filter, but optimized for speed and I suspect much faster than the one we have. That may be useful if latency is an issue. It also has Principal Component Analysis (PCA). This is a very interesting tool. Lets say we have many time-series of stock prices or indicators. Many, if not all, time-series may be somewhat or even strongly correlated and have redundant information. Using them all increases the risk of data mining but discarding most seems like a waste of good information. That is where PCA comes handy. It can take all those correlated time-series and transform them into a much smaller number of uncorrelated time series, while retaining nearly all of the contained information. PCA is often used to reduce the number of inputs to pattern recognition and neural net algorithms and, thus, reduce the risk of data mining. In the past, I have used PCA to aggregare 500 stocks in the S&P500 index into 8 buckets that tracked S&P500 but had uncorrelated to each other returns. Perhaps there is some use for it in jBookTrader. ________________________________ From: nonlinear5 <[email protected]> To: JBookTrader <[email protected]> Sent: Fri, October 29, 2010 7:14:35 PM Subject: [JBookTrader] JBookTrader release 7.10 Latest release download: http://jbooktrader.googlecode.com/files/JBookTrader-7.10.zip Release Notes: http://code.google.com/p/jbooktrader/wiki/ReleaseNotes User Manual: http://docs.google.com/View?id=dfzgvqp4_10gb63b8hg Historical market data sample: http://jbooktrader.googlecode.com/files/ES-GLOBEX.zip Historical market data (very large set, password-protected): http://jbooktrader.googlecode.com/files/ES-June2009-Oct2010.7z Changes in this release: 1. Added 3 new indicators, including the Kalman filter (thanks, Astor). 2. Added the "settlement time" for better trading/backtesting/ optimization. The settlement time is the time interval during which the indicators are allowed to "settle", i.e. reach some stable values. This is relevant to, for example, exponential moving averages. The default is set to 1 hour. This means that JBT would not open any new positions unless there is a continuous market data stream for at least one hour. 3. Both "brute force" and "divide-and-conquer" optimizers run about 1.3 to 2.0 times faster. 4. Improved the MovingWindow class: added an access method to retrieve any of the most recently added N elements (thanks, Konstantin). 5. Fixed a minor problem with a unicode character in the output of the NTP clock (thanks, Flo and Skunktrader). As I indicated in the other thread, I am trading live, and JBT feels solid and stable, so I don't anticipate any more JBT releases in the immediate future. That is, of course, unless there are any problems, which I would address promptly. -- You received this message because you are subscribed to the Google Groups "JBookTrader" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/jbooktrader?hl=en. -- You received this message because you are subscribed to the Google Groups "JBookTrader" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/jbooktrader?hl=en.
