1 TB, Mandarin
2 Modeling software: R,matlab,python
3 c++,c#
4 Niche Platforms: Mq,oq
5 self-developed platform of the company
What are the pros and cons, if it is 5, compared to 1-4 what advantages, thank you!
Reply content:
Deal with the contract for the month back to test the matter, pass the market on almost all platforms. And this matter is extremely important in the commodity. @Cheney have been very detailed, I would add MQ and OQ, both of which I have used.
First say OQ, formerly developed by the Russians, used in physical experiments. Later acquired by the Rice, converted into a quantitative trading software quantdeveloper, mainly for small to medium-sized institutional customers. Later, Quantdeveloper was the prototype of the SmartQuant Suite, where Openquant 3X was the flagship product, combining Quantbase (data solution), Quanttrader (trading solution), Quantrouter (Multi-account, sub-account, cross-market trading solutions).
------------------------------------------Why I was splitting floss------------------------------------------------------
Below I am just the Internet porter:
Openquant Product Technical Features
Running platform: Windows. NET
Main programming and scripting languages: C #
Openquant and Visual Studio can be used for policy editing
Visual Studio can be used for debugging (attaching to a process via VS)
Quantbase can process 1 million I/O operations per second (depending on hardware)
Quantbase supports multiple data sources
Quantrouter supports multiple inputs and outputs
Leverage the fix and API interfaces to support historical data, real-time quotes and trading services from over 20 service providers.
------------------------------, why am I splitting floss--------------------------------------------------
Looks very good is not, oh, this leek to tell you, the data is not complete, a lot of characteristics need to write their own code to experiment, we are tossing for six months, just calculate. Damn, at this time, Quantbox suddenly announced that OQ3 no longer updates the plugin. Forced to upgrade to openquant2014, the giant pits have to be re-filled again. Openquant2014 brings a lot of advanced strategy development, management concept. I'm not going to say one by one more here.
-----------------------------------Damn it, I'm splitting the line--------------------------------------------------------
Magicquant, known as Tsinghua University professor led the development of a quantitative trading platform (you know). Oh, I have quantdevoloper source code you believe it? 100,000 lines of code are omitted here.
No matter how he came, the key is not good to use it? To tell the truth, it is much higher than Mandarin and tb,b. Fast trading (said to be a little faster than OQ), authentic programming languages, libraries, easy to expand custom features. Compared to OQ's shortcomings, is the stability of almost, not OQ so powerful suite combination.
------------------------------------------------I swear, I'm the last split line-------------------------------------
Well, what about the Backtesting platform? Hey mom, it's a run off. The backtesting of the two softwares is certainly oq2014 to be much stronger. You can write your own set-up to get back to the test engine, with oq2014 2 million times per second event processing engine back and forth to put the minimum granularity of data, as far as possible to real restore the transaction occurred at the time, to avoid falling into the logical trap.
But, this leek tells you, but, do you think it is necessary to engage in such a complex? You get a MACD on the punch, get a double moving average, get a bollinger band, get a KDJ Gold fork Silver fork, get an ideal forum copied over the strange shapes to do backtesting, you also go to write a matchmaking device, you toss it.
The leek suggested, subtraction can fix, with Mandarin TB. Calculus, linear algebra, probability statistics, game theory, neural networks, machine learning what else can you use besides Matlab and Python? Entangled, Zhouyi, I Ching, Yi Jiao Jing, Diamond Sutra, Big day Buddha Sutra, you are almost immortal to do the trade do.
After the modeling analysis has the potential, writes the OQ code, uses the tick data to replay again. Consistent, with small funds firm, real disk data and OQ playback and MATLAB modeling analysis results to do tracking comparison, and then gradually overweight.
In addition, to mention a very useful software multicharts, powerful, young and old, professional non-professional take-all.
Portal: MultiCharts Chinese version, I'm not Amway. No need to indicate the source, can be reproduced at will. 1.TB is now a more professional programmatic use, writing language is similar to C, the function is more powerful; Mandarin Finance is its own development of the wheat language, compared to TB writing more simple, suitable for beginners to get started.
2. Modeling, three languages are similar. Matlab in the numerical calculation does not say, it is born for this, but also face the software is too bloated, the calculation efficiency is relatively low problem. R doesn't use much, but I think it's better to learn python than to learn R. So it's more recommended to use Python, because Python is extensible and compatible, and more readable, easier to communicate with engineering lions, and I understand that quant departments such as Goldman Sachs and Morgan are mostly using Python to model.
3.c++,c# surpasses the above three languages in processing speed, but lacks in modeling and numerical calculations. It is not recommended to use this language for people who are not computer majors or have a solid programming foundation. Of course, if the real disk play high-frequency, or the trading time is very high, it is recommended to use this language, or directly using the FPGA implementation, but this also involves the location of your server storage, server hardware conditions, speed and the rate of the trader itself and many other factors. Some irrelevant, in this unknown talk!
4. Niche platform OQ, MQ, have not been used, do not evaluate.
5. For their own development platform, the individual feel to look at the size of the company, as well as personnel configuration. The most important is the purpose of use. If, as you say, just for backtesting, the individual feels that the cost of development and subsequent maintenance costs are too high. Small and medium-sized companies and non-financial companies are not recommended to build themselves. As recommended upstairs, you can use the guest
, Ricequant-beta
, Gifted mine
, Joinquant Poly-wide Quantitative trading platform
, Quantopian
and other quantitative platforms.
I.Highly recommended in Quantopian
Study above, after all, foreign quantitative transactions for decades, many of which are worth reference, although not very suitable for the Chinese stock market, but also inspire some inspiration. Of course, in plain English, not everyone likes it. and trading rules and domestic differences, learning model can be, back to test it.
II.Joinquant Poly-width quantization trading platform
, Quantopian localization of one of the products, the use of Quantopian design concept, the basic can be freely switched with the Quantopian, but the backtesting speed compared to Quantopian has a huge increase (who knows), the community active, support multiple backtesting at the same time, There are reminders of simulated trading. Joinquant most impress me is the response time of customer service, 12 points ask questions still someone reply, make me dare not to ask questions at night, afraid I a problem affect others sleep-! Personally think Joinquant and Quantopian together, can be regarded as a combination of two swords, model-all bar.
III.Gifted mine
, Quantopian localization products of the second, backed by the data, the data is complete, but many are paid, this is a pain, but also can understand, after all, the connection is to survive. The backtesting speed is slower than joinquant, the interface is fresher, and it makes up for the lack of backtesting. Complete data but call API many and miscellaneous, a lot of redundancy, the basic use what to check what, a lot of attributes exist in multiple APIs, use is also the benevolent see bar.
VI.Ricequant-beta
, Quantopian localization products of the third, with not much, only a little feeling, they should change pm.
V.Condition Guest
Should be the only one of the four in the use of the R language platform, specifically useless, because feel r to engage in academic, self modeling can also. Once the combination of engineering, it is not applicable, no engineer will give you to learn r, unless you are Daniel! [formerly] A cloud-based computing platform for cutting-edge R language Technology: the Customer technology R SDK trial
Objective
Customer is a Zhang Dan teacher 's entrepreneurial project, Zhang Dan Teacher is a senior R language users, with "R's Geek Ideal" series of books and series of blog posts , on the road of R language Engineering, there are many worthy of our study, It can be said that the Zhang Dan teacher represents the forefront of R language Technology. This time, let's take a look at the latest works by Zhang Dan, highlighting the experience of using the cloud-based quantization platform.
Trial
Open, the customer's console, so that my eyes are the automation of statistical reports +rstudio-server computing environment, referring to the previous article based on Rstudio Webinars Statistical report web and Engineering practice summary , the perfect combination of scientific computing environment + Rmarkdown output, which means that even if I turn off the browser, the calculation site will remain intact.
Rstudio Environment
Refer to the previous RStudio server environment quick Configuration Tutorial in the RStudio environment, we can even introduce git versioning, shell scripting, shiny apps, Project engineering, and more. I suspect that the Rstudio on the platform may also be a docker-generated virtual environment.
In the shell we can even see the network configuration
~$ ifconfigeth0 Link encap:Ethernet HWaddr 02:42:ac:11:2f:77 inet addr:172.17.34.8 Bcast:0.0.0.0 Mask:255.255.0.0 inet6 addr: fe80::42:acff:fe11:2f77/64 Scope:Link UP BROADCAST RUNNING MTU:9001 Metric:1 RX packets:60686 errors:0 dropped:0 overruns:0 frame:0 TX packets:33607 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:24220177 (24.2 MB) TX bytes:52749143 (52.7 MB)lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:65536 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)
Write yourself, open source platform structure jumbled, logic can not bear to look straight, inefficient, poor scalability.
5 Advantages: The benefit of the backtesting platform is to separate the strategy part and the backtesting run part, and the strategy part only needs to call the historical data with the API function to place the order, so the most important thing is that the platform provides the API function must be suitable for any strategy, so its platform can easily add the API. (for example, the China market will join the ups and downs API, as well as the extraction industry, component stocks and other APIs. If someone else's shitty platform, what? Do you want to add a new API? So you're going to have to see the source code for one weeks. In addition, the backtesting speed is fast. Self-written platform is sure to be able to quantify everything to quantify, and others write may be lazy on the lazy, the speed is generally 10 times times more than the gap. (For example, order this function, the conscience of the platform is sure to quantify AH) recommend a non-programming person can also use a Stock strategy research tool-Quantitative master, this is a small work of our team, perhaps the target crowd may not be like these professionals upstairs, because the great people are programming themselves to implement a lot of modules. Our small tool is to help users who do not program, users simply follow the standard rules of trading system, the rules of buying and selling, position management and other rules assembly, a key backtracking, the app has implemented a number of commonly used technical indicators backtesting, the next will also achieve multi-standard strategy creation and backtesting, You can experience the experience with interest.
Quantitative master iOS Download: https:// Itunes.apple.com/cn/app /gu-piao-tou-zi-liang-hua-da/id1102702990?l=en&mt=8
Qt,wealth-lab, advantages, not all kinds of import, easy to get started, the shortcomings are the same, no import, poor extensibility. NET Platform
Make Basic observations with R first.
Then C # writes directly
Interface Module Building Well, it's only necessary to change the logic after the first time.
The advantage is not to learn the language of other platforms, and very free, self-written debugging rest assured. There are problems to know where. Life is too short, I use poly-width I just curious QQ also has a back test platform?
