Category Archives: Quant Development

Quant Development

Neural network AI is simple. So Stop pretending you are a genius

Neural network AI is simple. So Stop pretending you are a genius

This was a wildly popular topic in my Facebook group.

Here are some quotes that may be relevant for you:

xNeural networks are dumb black box systems. If you torture them enough you can get great fit of test data, but you won’t get great results from randomly sourced tests. AI is really good at spurious correlations….

xqPeople want to ascribe magic powers to AI, but for the most part AI can’t do anything a human can’t. There are some exceptions, but only for transparent AI. Neural Networks aren’t transparent, and even in the transparent systems a human would be able to replicate the final result…

AI can be built to do great things. Neural networks have limitations.

Here is that popular article 

More questions from my previous post

Would appreciate your guidance

I wanted to reach out to you to ask you for some advice & guidance. I am not a programmer, but I am trying to create a solution that will allow me to work with live tick data (Forex mainly) (to create live dashboards, currency strength, momentum etc) and historical data to conduct further analysis.

I was thinking to feed this data into a database (I saw in your video that you are doing something similar). I wanted to get your advice as to the method of getting live tick data (something that does not cost a fortune would be ideal 🙂 ) ..

I spoke to a freelancer who said that Jfoex api is only a java library and it would not allow me to work with live tick data.

I’ve also had a look at a few tools to connect to dukascopy to export historical tick data, its just the ‘live tick-data’ part that I am unsure about.

Any advice would be much appreciated.

My answer was:

If you just want market data use IQFeed forex option for $30/month. Do understand as part of their terms of service, you cannot redistribute this outside of your own use. If you need it for trading, always use the broker’s data to keep in synch. Also, you will need a live Dukascopy account for the live tick data but again only for your purposes. Hope this helps

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!

 Linux Tutorial: Redis NOSQL and Purism hardware for privacy

Linux Tutorial: Redis NOSQL and Purism hardware for privacy

This was a one on one Linux tutorial for on my members. It seems during 1 hr in, my Apple Mac laptop decided to crash for the first time in 3 years. I even mentioned during the live tutorial on how Apple software has pretty well gone downhill in the last few months. This was just another example of a crash. It killed the first 1 hr video session. As you listen to this tutorial, I will only be able to help out people at this this level who become members.

Where this video is left from is helping this person build a Redis NOSQL database environment. This person could not install Redis Desktop Manager (RDM) due to some incorrectly install version of Redis. His chosen Linux distribution was Ubuntu Server but you must likely needed Ubuntu Desktop to allow RDM. So when you start watching this video, please keep this in mind.

Purism hardware for your privacy and security

This conversation also lead into find Purism the new computer hardware maker out of San Francisco USA. Purism has locked down hardware (e.g. laptop and tablet) with a new phone running a secure Linux called PureOS. I am not recommending this hardware maker quite yet but do understand I would prefer to use a company outside of the USA. One desired to ship hardware out of somewhere like Switzerland. This is all explained in my video.

I would further suggest others to view on various Ubuntu Linux install tutorials if you need certain resources like Redis.

Use in a Ubuntu Linux Server terminal session:

sudo apt-get install redis-server

Redis on Linux tutorial and Purism hardware for privacy

http://archive.ubuntu.com/ubuntu/dists/artful/main/installer-amd64/current/images/netboot/

https://askubuntu.com/questions/868848/how-to-install-redis-on-ubuntu-16-04

https://www.youtube.com/user/quantlabs/search?query=install+ubuntu+redis

http://linuxpitstop.com/install-redis-and-redis-desktop-manager-on-ubuntu-linux-16-04/

https://distrowatch.com/table.php?distribution=trueos

Front page

OpenBSD install with GUI video

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!

Popular indicators from TA-Lib Python wrapper

Different analytical indicators

Popular indicators were introduced from 2 recent videos. These were from the TA-Lib package Python wrapper.  I further tested to see which indicators could be found useful. Although, I would give you a different answer from my recent analysis video where I demonstrated popular indicators Williams R% and Stochastic, you would think everything looks clean. This is just another way of saying they do not lag when they change direction at fast timeframes like Tick or 1 Minute. I really emphasized the importance of this in the videos.

Different results vs Dukascopy JForex

I went ahead to start analyzing the same indicator with exact timeframes using TA-Lib within Python. As compared from within Dukascopy JForex, It seems I got different results so I tested a complete pile of them. This included:

1. Stochastics was just to jumpy to work with any form of trend.

2. Beta was more consistent which did not move around as much.

3. Williams R was again to volatile (in short periods) moving between in upper and lower ranges

4. Linear Regression and Chaikin A/D Oscillator was the same effect as Williams R

The only ones that seem to work were the popular indicators which include:

1. Normalized ATR

2. SMA

3. Bollinger Bands

4. Rate of Change Percentage

As a result, I will stick these 4 above to find out the magical combination. There are others to pick from within TA-LIB but I find you want to keep the combination simple. This also is important to really understand how properly interpret the unpredictable indicators I list above.

Next steps and considerations

I need to show how to  properly implement these but it depends when there data stored eg. database vs flat file. It looks the simplest is to use CSV with multiple instrument access is less utilized on the computer hardware front. You can obviously understand hoq critical these decisions are for when using an interpreted scripting language like Python.

There will be lots more to report as I continue along. Do note I find working with Python is much more product than Java within Dukascopy Jforex API.

Package info https://github.com/mrjbq7/ta-lib

Williams % and Stochastics most reliable Matlab technical indicator

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!

Crypto currency Bitcoin bot development with Python over Javascript PHP

 

Crypto currency development with open source

This is a video explaining my view of crypto currency bot development. This includes Bitcoin with a focus on using Python over Javascript or PHP.

Advantages of using Python over web development programming languages

After my various videos on based on trading bots for Bitcoin. See below for that. The main hate on from what I see is that people don’t like negative references back to web development languages (e.g. Javascript or PHP) are substandard for more advanced languages including Python. I explain all of my rationale in my first 13 minute video below.

Working Python gives you many advantages to integrate with other awesome packages including backtrader. It seems that a widely evolved package like ccxt  enables you to wrap to 90 global exchanges. Some will be valid and safe while others will drop off with bad news around them. As a result, you only to maintain one interface with the latest exchange changes. You will not need to pull your hair out by maintaining each exchange you are interested in working with.

Here are some past videos I have done regarding Bitcoin bots. This includes Gekko or other bot projects.

There is no lazy way

If you plan to use some black box bot, you are pretty well wasting money. I say that because you most likely are looking for the quick easy way to make money at crypto. Guess again? Those days are over. Not only that, it means that you will need to work. You will need to full understand ‘under the approach’ of how these instrument flow. Yes they are valid. It is the volatility where you will make some great coin if you forecast the correct market direction. As a result, you will need to understand this with a good amount of learning. Sorry but I don’t think there is no other way.

 

Add OHLC Bar Data To Your Algo Strategy

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!

Technical Analysis trading with TA-Lib with Python

How to use TA-Lib with Python for technical analysis trading

This video will highlight the need of using a hopeful reliable technical analysis indicator library. My video explains the history of how this library was built and used in other popular projects.

Correct technical analysis indicators 

I could go on and on about technical analysis but I do find it useful for timing your positions including entry and exit. In past videos on testing visually various technical indicators with JForex, you will clearly see why I use this process. For instance, I thought linear regression would be useful. By checking visually on charts, it clearly showed  clearly how this indicator was not right. It was too sensitive on potential market direction at low levels e.g. 1 minute timeframe.

Technical indicator lessons

Using the chosen indicators it clearly  showed how they are not as sensitive/noisy nor lagging. This was an important eureka moment.  It is unfortunate I spent too many weeks to figure this out. Logging your positions in full detail of the indicators results you want to use is very helpful.  This is a critical process you need to understand instead of guessing where your trading positions results will end up. Many retail traders will never to get to this point. As a result, it does help to have this point of view when testing your trading ideas.

Python faster to develop with 

Also, the difference between developing 100% in the JForex API is very time consuming. As you get more confident with Python. you will be able to bang out trading scripts much faster. You will n longer be bogged down by the limitation of a broker or their API. As hinted in my video, this is why I would prefer to have all the market and trading data in one central hub (eg. Redis or CSV).  Once again, i cover this in my video.

Here is one benefit of using Python for my own development!

I now have the ability to fully backtest with the backtrader Python package as opposed to what I did in the past.

Here are the links from the video

https://github.com/mrjbq7/ta-lib

https://mrjbq7.github.io/ta-lib/func_groups/momentum_indicators.html

Create powerful trading robots with JForex API

#1 goal in forex trading is not to lose money but profit is 2nd

 

 

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!

Financial forecasting with probabilistic programming and Pyro

Financial forecasting with probabilistic programming and Pyro

Lots of tutorials here

  1. Simple time series forecasting (and mistakes done)
  2. Correct 1D time series forecasting + backtesting
  3. Multivariate time series forecasting
  4. Volatility forecasting and custom losses
  5. Multitask and multimodal learning
  6. Hyperparameters optimization
  7. Enhancing classical strategies with neural nets
  8. rn recognition and machine learning
  9. Bayesian methods for hackers
  10. Documentations of libraries below

This current tutorial is here:

medium.com/@alexrachnog/financial-forecasting-with-probabilistic-programming-and-pyro-db68ab1a1dba

View story at Medium.com

View story at Medium.com

 

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!

Gold Price Forecasting Using Python Machine Learning

Gold Price Prediction Using Python Machine Learning

This is a basic way to apply machine learning Python packages like SciKit Learn. As gold is now becoming the number 1 asset of interest, here is a way to analyze movements on various gold ETFs. This is nothing revolutionary but it gives you the high level on how to implement these type of analysis and forecasting on popular market assets of the time

Links here

http://quantlabs.net/blog/2018/01/gold-price-prediction-using-python-machine-learning/

 

Also, I have posted a crucial item on understanding the bid and ask spread. If you never get the fundamentals on this, you will pretty well be failing with no knowledge why. Knowing this will help my forex strategy analysis much more. I will be posting a new video on this pointing back to this article. Let me know if you are interested in this by responding.

Thanks Bryan

Gold Price Prediction Using Python Machine Learning

http://quantlabs.net/blog/2018/01/bitcoin-price-forecasting-using-monte-carlo-simulation-forex-analysis/

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!

Gold Price Prediction Using Python Machine Learning

 

A deep learning framework for financial time series using stacked autoencoders and long-short term memory

Gold Price Prediction Using Python Machine Learning

This is a basic way to apply machine learning Python packages like SciKit Learn. As gold is now becoming the number asset of interest, here is a way to analyze movements on various gold ETFs. This is nothing revolutionary but it gives you the high level on how to implement these type of analysis and forecasting on popular market assets of the time.

Gold Price Prediction Using Machine Learning In Python

Bitcoin and Doge crypto currency the joke or still serious ?

A deep learning framework for financial time series using stacked autoencoders and long-short term memory

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!

OpenBSD install with GUI video

 

OpenBSD install with GUI video

Go back to the effing basics as they say in this video. Also, this was triggered by a Twitter follower who sent a list of the insecure ‘features’ of FreeBSD so this was recommended instead. After further investigation, this seems to be valid!

Caveats: It seems slower than FreeBSD which is faster. As a result, this is more important to stick with as a priority since I can configure or harden myself. Or I could always use HardenedBSD as well which worth investigating.

https://www.quora.com/Is-FreeBSD-faster-than-Linux

 

And if you were doing Java, you’d want highly tuned threads. FreeBSD gave up a lot of stability for a long time for their kernel threads.

Is OpenBSD slow on servers? from openbsd

See

Death of Apple started while crypto currency event tonite

Dude behind Netflicks who did FreeBSD

http://www.bsdnow.tv/episodes/2014_11_26-8000000_mogofoo-ops

 

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!