Tag Archives: challenges

Two videos from challenges of flag, pennant, and wedges for crypto currency trading 

I have attached two videos from last nights ‘Meetup’ which are general topics of the future I have with crypto currency. I hope this becomes some kind of destiny. The first video talks about the challenges I had with those patterns called flags, pennants, and wedges. After last weeks ‘backtesting’ over 115 crypto currency pairs from Binance, I found I could NOT find any of these patterns with over 1000 hours in each pair. It can also confirmed these patterns are very rare but they definitely work and are very reliable when found.

Third Party Custom coding is the answer?

I am hoping to get these patterns custom coded by someone towards the fall. Other pressing areas I want to focus to move forward with instead of being bogged down by the coding of these patterns. All concerned will understand I am sure.

The second video is just to finish with some attendees since Youtube cut the stream prematurely so not much content is found in there.

Hoe this helps you out,

Bryan

P.S. Today is the last day I will be publicly posting the charts found in my Instagram from my Bloomberg News app. The Instagram account will be private after today which will be an added benefit for my paying member. Also, I will be posting more notes from the recent source of my crypto trading learning source in a few days as well for these same folks. You can get more info to be a member by going here at the online store.

http://quantlabs.net/academy/shopping-cart/

How to use trading patterns to prevent bad trading and perfect timing with less false positive signa

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Free psychology test provides information on the type of trade strength challenges a trader may have

Free psychology test provides information on the type of trade strength  challenges a trader may have

This came in from a local Meetup member:
Here is a free test below that will tell you and find out what kind of a trader of trader you are.  This test is a free psychology test provides information on the type of trade ,  the trader’s strength and what challenges the trader may have.

http://www.tharptradertest.com/default.aspx?question=1

Join my FREE newsletter to learn about tests like this

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Don’t let challenges of trading get in your way to sucessful profits

Hi there

Here are the latest additions in learning how to trade better:

1. When I thought I had accomplished my big ‘closing the loop’ on the platform, I came across a major technical challenge. It got me thinking so I made this video:

Here is my video you need passion and drive to succeed in quant trading and built HFT platforms

https://quantlabs.net/blog/2013/08/here-is-my-video-you-need-passion-and-drive-to-succeed-in-quant-trading-and-built-hft-platforms/

2. Risk management is important so someone I know did a quick video on it:

Is this best description of risk management? Protect your portfolio to prevent losses?

https://quantlabs.net/blog/2013/08/is-this-best-description-of-risk-management-protect-your-portfolio-to-prevent-losses/

3. I knew one day that this scam would come along for HFT. Be warned

First affiliate marketing internet marketing scam to profit from high frequency trading aka HFT

https://quantlabs.net/blog/2013/08/first-affiliate-marketing-internet-marketing-scam-to-profit-from-high-frequency-trading-aka-hft/

My listed technical challenges were:

Now remember I am closing off the QuantLabs.net Premium Membership to the public! I am in the midst of ‘closing off the loop’ where my custom platform will have:

1. Real time market tick data capture and analytics with threading

2. Visualization

They did not play not nice but I will figure it out!

As this is coming to a near completion, I will be posting more info on this. I may even create an expensive online course on how I did this. All in all, I cannot stress you need to join this Membership before the public will be closed off from it forever.

–> JOIN FOR IMMEDIATE ACCESS <–

Here are many benefits to the Membership

Thanks for reading

Bryan

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

More challenges of HFT with open source R with C or C++ vs Matlab for quant LIVE trading in production

Hi there

I thought this would be important to send out to all concerned.

List below is an email response I got a few days  ago (My responses are in italic with –>  );

This was from:

https://quantlabs.net/blog/2013/05/challenges-of-hft-with-open-source-r-with-c-or-c-vs-matlab-for-quant-live-trading-in-production-no-visual-debugger-in-rstudio/

Here it is:

….

 

This is great. I really want to use R to develop strategies. I don’t really want to use Matlab for the two reasons:

1) I live in Hong Kong and despite me sending a bunch of emails about getting set up they won’t get back to me. Their automated response when I fill out an online request is that a sales person in my region will contact me soon. I’m guessing they don’t have a sales person in Asia.

 

–> They should respond!

2) The cost of Matlab is significant relative to R. If you set up successful strategies then Matlab pays for itself but until you have built, tested, and implemented these strategies it puts a dent in your trading capital.

–> I agree but Matlab is more for 7 digit capital operations. My latest quote for my goodies exceeded $125K so I know the pain. I can tell you though you will easily make it back if you are hiring coders to hand code your trading ideas. Also, Matlab does not call in sick, steal your algos, or just quit mid way. Matlab is pretty easy work with. It just depends how serious you are as well as how much starting capital you have. If you are limited, yes R is the best way. Maybe even Python?

I just completed the free quants and econometrics course in R from University Washington on Coursera.org. It shows you how to code a trend forecasting AR model in R. You might want to put it down as a suggested intro to your courseware too. link here https://www.coursera.org/course/compfinance

–> I have heard about these but you can also learn through Apple Itunes U. It is all the open source courses through Ivey universities. Some killer stuff.

UW also run a course in R walking you through:

Rcpp interface from R to C++

Rinside interface from C++ to R

https://depts.washington.edu/compfin/amath-557-financial-software-development-and-integration-c

 

–> Yes I have tried this but it slows down when connecting the R seesion in your C/C++ environment. To connect an R session may take more nano/milliseconds than you want. This is why I went back to Matlab to look for the best code generation process to embed natively into your trading system. This is why I have looked at Simulink which is very cool with the rapid model development environment. I will be providing further online courses will be provided in this area as well.  You can code generate with C/C++ or even for FPGA deployment from the same Simulink visual model. Imagine that?  You will see a HUGE difference in productivity.

But the UW AMATH-557 is not online. There is bunch of stuff online when you google it so can figure things out for yourself, like this one http://dirk.eddelbuettel.com/code/rcpp.html . You might want to add something like this to your courseware.

–> Again, I like RCPP but see my point above. Those milliseconds delay could be an issue when you try to bridge R and C/C++ into the same session. I even looked at R with Java and .NET C#. I personally don’t think you would use this for a LIVE production environment but I may be wrong. As said, I am more comfortable doing it through a code generation process from Matlab. As for RCPP, I would encourage anyone to comment to set the record straight. I would love to investigate RCPP if there has been an significant engineering update from 2012. I am also still waiting for my visual debugger in Rstudio which would make my life much easier. My life would be pretty we ll a party if that happened.

I’ve spoken to a bunch of HFT desks and they always come back with the two questions:

1) Do you have a HFT strategy that’s got a good track record for over 3 months and is scalable?

 

–> This is a very standard response

2) What qualifications in programming automated strategies have you got?

–> Haha. Read the Max Dama online book on this. He basically says, real programmers/quants don’t care about the programming language.

quantlabs.net/blog/2013/05/who-knew-max-dama-would-be-put-out-a-book-on-automated-trading-use-matlab-or-excel-as-well/

ALso, it seems all institutions use Matlab more than R but again, it depends on your goals.

 

If your answer to the first question is yes then they don’t need to ask the second question. Your proposed training course should help people be able to answer that question with a yes which is the main goal.

If we are not there yet then being able to answer the second question strongly would be a secondary goal. The success of Coursera is it gives you certificates of completion. Being able to produce certificates of completed training for your courses would be very useful for those quant researcher roles.

–> Hey I could issue certificates to with quizzes at the end. I don’t think mine would be recognized by industry. But I thought about that with CQF as well. While I was in NYC, it appears I was dead wrong so maybe Courera may help.

I think what you are doing is great. Keep up the good work.

–> Thanks

….

Interested in all this stuff? Join my QuantLabs.net Premium Membership now to get access to this important section on the site. It iwll be included as part of the membership. Extra value with no price increase!

–> GET IMMEDIATE ACCESS TO THSE TOPICS NOW <–

There are many more benefits listed here as well.

Thanks for reading

Bryan

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Challenges of HFT with open source R with C or C++ vs Matlab for quant LIVE trading in production. No Visual Debugger in Rstudio?

Challenges of HFT with open source R with C or C++ vs Matlab for quant LIVE trading in production. No Visual Debugger in Rstudio?

Like this? Learn more through my FREE newsletter

I got this email asking for some points of view on R, Matlab, etc. This is in response to:

https://quantlabs.net/blog/2013/05/learn-the-most-popular-market-forecasting-technique-with-r-here/

My responses followed by:

 

Hi Bryan,

 

This is great. I really want to use R to develop strategies. I don’t really want to use Matlab for the two reasons:

1) I live in Hong Kong and despite me sending a bunch of emails about getting set up they won’t get back to me. Their automated response when I fill out an online request is that a sales person in my region will contact me soon. I’m guessing they don’t have a sales person in Asia.

 

–> They should respond!

2) The cost of Matlab is significant relative to R. If you set up successful strategies then Matlab pays for itself but until you have built, tested, and implemented these strategies it puts a dent in your trading capital.

–> I agree but Matlab is more for 7 digit capital operations. My latest quote for my goodies exceeded $125K so I know the pain. I can tell you though you will easily make it back if you are hiring coders to hand code your trading ideas. Also, Matlab does not call in sick, steal your algos, or just quit mid way. Matlab is pretty easy work with. It just depends how serious you are as well as how much starting capital you have. If you are limited, yes R is the best way. Maybe even Python?

I just completed the free quants and econometrics course in R from University Washington on Coursera.org. It shows you how to code a trend forecasting AR model in R. You might want to put it down as a suggested intro to your courseware too. link here https://www.coursera.org/course/compfinance

–> I have heard about these but you can also learn through Apple Itunes U. It is all the open source courses through Ivey universities. Some killer stuff.

UW also run a course in R walking you through:

Rcpp interface from R to C++

Rinside interface from C++ to R

https://depts.washington.edu/compfin/amath-557-financial-software-development-and-integration-c

 

–> Yes I have tried this but it slows down when connecting the R seesion in your C/C++ environment. To connect an R session may take more nano/milliseconds than you want. This is why I went back to Matlab to look for the best code generation process to embed natively into your trading system. This is why I have looked at Simulink which is very cool with the rapid model development environment. I will be providing further online courses will be provided in this area as well.  You can code generate with C/C++ or even for FPGA deployment from the same Simulink visual model. Imagine that?  You will see a HUGE difference in productivity.

But the UW AMATH-557 is not online. There is bunch of stuff online when you google it so can figure things out for yourself, like this one http://dirk.eddelbuettel.com/code/rcpp.html . You might want to add something like this to your courseware.

–> Again, I like RCPP but see my point above. Those milliseconds delay could be an issue when you try to bridge R and C/C++ into the same session. I even looked at R with Java and .NET C#. I personally don’t think you would use this for a LIVE production environment but I may be wrong. As said, I am more comfortable doing it through a code generation process from Matlab. As for RCPP, I would encourage anyone to comment to set the record straight. I would love to investigate RCPP if there has been an significant engineering update from 2012. I am also still waiting for my visual debugger in Rstudio which would make my life much easier. My life would be pretty we ll a party if that happened.

I’ve spoken to a bunch of HFT desks and they always come back with the two questions:

1) Do you have a HFT strategy that’s got a good track record for over 3 months and is scalable?

 

–> This is a very standard response

2) What qualifications in programming automated strategies have you got?

–> Haha. Read the Max Dama online book on this. He basically says, real programmers/quants don’t care about the programming language.

quantlabs.net/blog/2013/05/who-knew-max-dama-would-be-put-out-a-book-on-automated-trading-use-matlab-or-excel-as-well/

ALso, it seems all institutions use Matlab more than R but again, it depends on your goals.

 

If your answer to the first question is yes then they don’t need to ask the second question. Your proposed training course should help people be able to answer that question with a yes which is the main goal.

If we are not there yet then being able to answer the second question strongly would be a secondary goal. The success of Coursera is it gives you certificates of completion. Being able to produce certificates of completed training for your courses would be very useful for those quant researcher roles.

–> Hey I could issue certificates to with quizzes at the end. I don’t think mine would be recognized by industry. But I thought about that with CQF as well. While I was in NYC, it appears I was dead wrong so maybe Courera may help.

I think what you are doing is great. Keep up the good work.

–> Thanks

P.S. I missed your Feb 27 at 7PM Instructions on how to set up this trading platform. Did you record it?

–> Send out the exact link or topic as I will be able to give you an answer. It could be on my Youtube channel (youtube.com/quantlabs) or only for my Premium Members.

Like this? Learn more through my FREE newsletter

v

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Youtube video How to overcome Challenges of Matlab and R for your HFT C++ platform

Youtube video  How to overcome Challenges of Matlab and R for your HFT C++ platform

See how I accomplish by joining my free newsletter

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Challenges and rewards for high frequency, quant, and algorithmic trading in Hong Kong with the Experts

Challenges and rewards for high frequency, quant, and algorithmic trading in Hong Kong with the Experts

It has not gone unnoticed that algorithmic trading has taken a large portion of the market volumes in the U.S. and Europe, a development that is about to be mirrored in Asia. While quant trading, high frequency trading and algorithmic trading are in the end all quantitative, they are often talked about as the same thing. High frequency trading (HFT) is a subset of quantitative trading, and in fact many low frequency trading strategies are traded via algorithms, which distribute orders in slices. Even minimal execution timings aren’t just loved by the HFT ambassadors, also low frequency traders may be interested in reducing their execution to 50 milliseconds or even less.

The Hong Kong Roundtable script can be access here: http://www.opalesque.com/RT/RoundtableHongKong2012.html

 

 

==

Most high frequency trading consists of market making and spreading and will therefore make market movements less extreme while at the same time providing liquidity and bringing trading and execution costs down. These strategies are being supported by exchanges in both Tokyo and Singapore which have recently recognized the relevance and positive benefits.

 

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

The Opportunities and Challenges of Exascale Computing for Quant Development

The Opportunities and Challenges of Exascale Computing for Quant Development

Our recent blog focuses on the march to exascale including which technologies have shed some like to help us make that leap. http://www.appro.com/blog/the-opportunities-and-challenges-of-exascale-computing/

 

==

Combining low power ARM processors with GPUs is one way to approach this problem. In order to program these machines we can use a combination of MPI and OmpSs. I am infact in the process of building an ARM cluster and benchmarking various performance/watt attributes.

 

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Challenges that lead to FPGA technology deployment in the financial markets

Challenges that lead to FPGA technology deployment in the financial markets

 

bobsguide.com

Bobs Guide entry for…

—

I think the main challenge will be devising a cost-effective solution, since the target audience is a niche segment and the solution needs more frequent updates/tweaks.

—-

I agree with Sameer. However I would like to add that FPGA solutions are a yet unproven technology in the financial sector. I don’t see cost & risk averse countries (read India) agreeing to become the guinea pig. Whatever exchanges may say, developing countries aren’t seeing enough load to require a hardware intensive solution. The risk cost of an FPGA solution is simply too high for them to jump on the bandwagon. Also financial institutions don’t like to build systems they are not in the business. They like to acquire products. So until you find a risk management system or the like available off the shelf the technology will not be able to make inroads into this segment.

 

 

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

What are the main challenges when using Java for low latency work?

What are the main challenges when using Java for low latency work?
—-
one: it is non deterministic. What you see once in development is not likely to be seen again in production. A scattergram of response times is all over the map.
4 days ago
—-
Not sure about challenges part, but Real Time Java promises deterministic execution of code based on hard real time operating systems. That’s what I’ve read in a book.
—-

Yes, it’s true – there is real time Java, but it is soft real-time (OK for HFT) and has a little drawback – while minimizing dispersion of latency it does increase average of it.
—–
Real time Java is not regular Java. It is not, “write once run everywhere” any more.

Yes the OS has to participate in the solution by doing what it is told and not “helping” you when you do not expect it.

The various RT Javas were not consistent.
—-
The lack of “write-once, run-anywhere” with RT Java Spec isn’t a big deal, in my opinion.
If you’re investing the effort to write real-time Java code (which has some significant differences from non-real-time) you aren’t going to be switching JVMs very often, nor are you going to be running inside a browser (where WO-RA matters most).

The biggest challenge with writing low-latency Java is that you usually can’t use much, if any, of the standard libraries, particularly the containers. You end up having to rewrite these yourself to optimize them for your particular applications.

The unfortunate truth is that although Java is a fairly nice multi-threaded language, you can’t take advantage of it if you need consistently low-latency performance.

I’d say that if you’re being realistic about planning a new development project, you should assume that writing the code in Java will take about as long as writing in C++ would take (assuming your developers are good enough to write high-performance Java code). If they do a good job, the result should be about as fast as C++ code would be, plus or minus 10-20%, depending on the application and programmer skill.
—-
I understand it not being a big deal once you pick an environment to use. The point I was trying to make is that it turns out to be a different Java on each platform, so you have to really know the target well. Switching is not an easy option if you become disallusioned with your first choice.

Plus once you start programming with the quirks of RT Java you mighty as well be using C++ which is portable.
—–
To name just a few: the hotspot compiler takes a while to optimize your code. You don’t have good control over which thread runs on which cpu. Its 2-3 slower than comparable C.

Don’t use the real-time java. Its deterministic but slow.

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!