By the way, you mention data quality of historical testing. That is critical but it's even more critical when trading live. TickZOOM has a very advanced but efficient tick filter algorithm. I learned about it from some Phd research papers on cleaning time series data in real time. It's tuned for Forex. Some of the parameters might need different settings for equities. We can figure that out later. That tick filter runs both during historical testing and during real time data. So it cleans the data on the fly. It's pretty easy to clean historical data even visually. But doing it on the fly automatically in real time is more challenging and even more important. Wayne
"""Based on what I learned, people always talk about Linux being free, but he points out that most people who have Linux buy a distribution in a box. And many of those were "Official" Red Hat Linux. Plus Linus, the creator, gets plenty of funding.""" This is old information and mis-information and also what RedHat wants people to believe but is not correct! In reality very few users pay for the RedHat distribution, if they pay at all what they pay for is service from RedHat or others. In the case of RedHat, most people that want RedHat but dont want the service just use something called CentOS that is freely available and is built from the latest RedHat source code. RedHat only sells the Binary packaged distribution but few people buy it, they just use CentOS instead if they are only looking for the distro but no service. Redhat has seriously damaged their general business by promoting mis-information about how open source works but I don't think they care because they are now serving the commercial sector only (infrastructure for cell towers, process servers, some web servers, PBX servers, etc). They are now a commercial only service company. Ubuntu has instead widely distributed all of their versions for free (Source, Binary, Desktop, Server, etc) with no restrictions. Look what this has done to their market share. They (Canonical the parent company) sell service if you want it. And many people do want it so they are doing quite well now.
I just had to repeat the following because this is really the best way to look at it,,, this is how open source works at its best. Everyone has total free access to the code, everyone feels like they have "ownership" and is motivated to contribute. Many of the motivated people actually will contribute and a big community will develop. The leaders and other top community members can sell services and make money. The user base can pay to have features added if they want to. They can pay to get what they want and when they want it. OR they can write it themselves, OR they can just wait until it just happens while happily using the software. Anyway I just had to repeat: 12-18-08 01:53 AM Quote from greaterreturn: Exactly! It sounds like you understand open source. Or do you? By sharing with friends, you're performing "viral marketing"--the most powerful marketing because it works better than anything else and costs the company nothing. In fact, you paid for the privilege. Besides, your friends are not "freeloaders". So, your friends might create addons for all of us or they might decide to signup for service themselves and therefore contribute to future enhancements. Look, I apologize for using the term "freeloaders". It was wrong and I'm sorry. It will never happen again. The reality is that people are to be admired who have friends who care about them enough to share such great software. Sincerely, Wayne I was mainly responding to the term 'freeloaders'. But the GPL does keep you honest. If your project is not providing value, your project may fork like Redhat & CentOS.
Thanks for raising the possibility that there's misinformation about how open source works. BTW, you discuss only one model. There are a variety of different open source models that work. Linux is just one of the earlier ones. The industry has evolved since then and some open source has happened much, much faster. Many times those really fast projects are cases of "standing on the shoulders of giants". For example, without the ZedGraph project as the drawing engine for the GUI, TickZOOM would be nowhere. But when you take a drawing/charting tool from here, logging component from there, math component from other there and add your own custom peice, then... BAM!!! You have a powerful project virtually over night. Slooooowly by tedious coding is only one, very boring, way to build an open source project. Fast, is much more interesting. I hope you agree. The name of the game today in software development is "integration"--not coding. Coding is outdated--very "old hat". Sincerely, Wayne
Hey, I just realized that the big issue in the last several posts as to "where" TickZOOM relates to what different people need. If someone wants to historically test strategies on Forex tick data and they're okay with doing that in C# or VB using .NET, then they're ready to get started with TickZOOM right out of the box. With very minimal trouble they can do that with Java or C++ too. Plus, if they happen to have MB Trading as their broker they'll immediately be able to deploy and trade their strategy without code change as a black box, live. And if they want to trade equities or Forex, then WAIT! I just remembered that I have testing equities. It works fine with equities. I remembered because I use the equity test server of MB Trading over the weekend because the Forex server is down on weekends. So it already handles the DOM, ticks, bid/ask etc on Equities just fine. Okay... But if someone wants to use FIX for example, then they'll perceive TickZOOM as still a seedling that needs to grow--and understandably so. In short, TickZOOM isn't for everybody. but hopefully it will work for more people soon as we add other broker interfaces like FIX, etc. Sincerely, Wayne
You may want to take a look at Tradelink. They are open source and C# and do all the things you are planning on building and it is usable now. See how they do what they are doing and maybe you could merge tickzoom with them and save yourself a lot of work.
Did you read the issues list? There's only 2. 1 is a problem with tick speed. Someone was trying to use the software with 20 symbols at that same time in real time and couldn't do it. That's not even back testing ticks. TickZOOM runs in production collection 12 symbols with tick data and DEPTH of MARKET which is 10 to 100X as much data and uses almost zero CPU and very little memory. That user couldn't get 20 stocks with only tick data to work. And he was pointing out how much data it was, etc. It has all the classic problems with handling tick data. So what about all the GUI stuff. All the project do that stuff. Anybody can write a GUI. Wayne
Wenzi, you know what? Maybe you if you were more specific. Perhaps you had something else in mind. My guess from what I read which was technical in the issue list is that the whole infrastructue of TradeLink is wrong for Tick Data. Tick is a VERY different animal than bar data due to the sheer enormity of the volume. We're talking 1000 X more data than even the smallest bars in many cases. Especially because when I say "tick" I mean every change in the Depth of Market which happens many more times than tick changes. TickZOOM handles that. Wayne
Tradestation's Radarscreen can update 2000 issues with ticks [trades]. With 1000 or less instruments it can have history and do TA on all the issues, cpu doesn't have to run that high either. I never noticed much lag while it is doing that. Tradestation 2000 had a tick server called Global Server that could update more than that from DTN Satellite...