Hi ButterMilk, what Trading Platform are you referring to that can do any of those things? Cheers Janus
Just a status update everyone. I'm nearly finished getting all the automated regression tests working again. Basically over the past several weeks, I added in better optimizations but the automated tests weren't up-to-date with the new architecture so I just brought them up-to-date. It was good too. The automated test found several things that I inadvertently broke. One of which was the session bars. Did I mention that TickZOOM has session bars? They're similar to day bars in that the occur every 24 hours. But you can set the start time of day and duration. The day bars operate based on your local time 24 hour clock and get used by the trading statistics to calculate daily win rate% and other daily stats. You can use the for trading too. Sessions can offer more flexibility. Why is this important? It's because there's several housekeeping things I need to do to clean up the code before releasing it. So it was necessary to make sure the system was fully running correctly and tested as thoroughly possible. That way, when I make these other clean up modifications, I can quickly run the tests to make sure nothing got broken in the process. So things are moving forward. Sincerely, Wayne
Like I mentioned, good luck with the project. I just want to publicly clear that I don't use, endorse, support or in anyway involved with TickZoom. As traders, as we all knew, we all deal with limitations. Limitations caused by market conditions Limitations caused by the cost of trading. Limitations caused by brokers. Our testing environment is one of them. It's part of our job to break that limitation. Obviously, that's why a lot of us learned how to program and all the IT stuff. That said, there's really no benefit for me to be part of a group that may cause a bunch of limitations towards what I develop and use. See... tools like risk management fitness or algo. snippets are fine. It comes to down to how people use it. But when it comes to the process of development, the "How-Tos", that's a different story. A lot of my trade / system management stuff are already coded. If I were to use TickZoom, I'll have to ask the devs to implement the engine for that. Which will end up as a teaching session for a bunch of newbs. on how to develop models. The most I do towards anyone (in ET or real life, in terms of trading) is to let them know that they're doing something wrong or missing out something. It's for that individual to find out what that is. At least, that's one of the reasons. But I'll say this again.... Good luck with the project. Wishing you guys all the best.
That's fine. Just an FYI for you and all the rest that the interface of TickZOOM for strategies and indicators which are both considered Formulas is very minimal in resolves alot of these concerns that TSGG has. It's because if you have working models on some other platform, you can just add layer for TickZOOM. In other words, you needn't rewrite them or redo them. TickZOOM uses a PONO technology. Do you know what that is? It means Plain Old .Net Object. Specifically, unline NinjaTrader for example, TickZOOM doesn't require you to "extend" some class in the framework to get all the functionality. That learn you FREE to you use whatever inheritance hierarchy you choose and makes it far easier to convert an existing model to TickZOOM. I'm committed to simplifying the process. I'll even be happy to make a NinjaTrader like layer for you already if you wish so that you can use them directly on TickZOOM...as an example. You can just let me know the interface your models need by some documentation or other wise, and I can map them to TickZOOM so your code runs virtually as is. No development cost needed to convert. Make sense? Wayne
Iâve developed a tick based trading system several years ago that Iâve integrated into NinjaTrader to utilize NT's many administrative and testing features: It utilizes a multi variable, multi strategy tick based G A walk forward optimizer I could use it to back testâ¦however I am not a fan of back testing â¦walking forward is much more realistic..
Excellent point. I much prefer walk forward optimization also. What time periods do you prefer to test and the out-of-sample period? TickZOOM can pretty easily add that feature and I agree it's necessary. That's a better way to avoid curve fitting. How did you get this to work in NinjaTrader? I tried using tick data myself and it sucked up all the memory and kind of locked up after only a month or two--I forget exactly--of tick data. Or do you only walk forward test for a month or so of tick data? If so, ouch. That's not much testing. Can lead to erroneous results. Sincerely, Wayne
I use GA to optimize the OHLC values based on the tick data stream for range, tick, time and volume bars. I often work forward using (tick data) 6 to 12 months ....manipulation of the OHLC bars size, shape and type is the MOST important variable that you can optimize. I do not use NT's optimization system . I upload the results of my GA optimization and allow NT do generate the final reports and automate the execution. If you go to the NT forum you can access discussions of GA.
Good news. After legal advice, the final decision is for TickZOOM to be open source under GPLv3. Thanks for input and comments. I'm starting a new thread next with the news since this one has become cluttered. I'll ask a moderator to close this one. Wayne