Updated: Oct 5
Why use trading system builders in the first place? #Diversification. If you have one idea to trade one instrument, then sooner or later you will be in a drawdown that you don’t know when you will get out of. To bring trading idea to an actual robust trading system that you can put money behind is not as easy as you think. You need hundred of ideas and thousands of iterations to reach a handful of robust systems. Like everything else, technology and automation is the solution.
Software that does genetic evolution is nothing new and has been used in research for many years in different sectors. Genetic software for trading, however, have been frowned up on and considered data mining. In the past two years, many advancements in software/hardware and the abundant availability of data made it possible to easily vet any system for its robustness, regardless of how the system was generated, be it a human or a computer. In fact, many hedge fund grade back testing engines and trading platform are available online for free to do whatever you dream. While you still need to know programming, but you don’t need to build the engine from scratch and test it for bugs and deploy it for use.
I am a semiprofessional trader and have been using TradeStation for the past 18 years since they were called Omega Research. I came across many software packages for the past years that does many things, but to build thousands of systems and vet them, that was left to hedge funds and firms with large capitals.
Traders today are so fortunate, because what use to cost millions of dollars only 5 years ago is available now for tens of thousands of dollars. In fact, you can start your hedge fund with 2-3 off the shelf software packages without any IT team on payroll. Even better the retail trader is now able to do what multimillion-dollar hedge funds use to do 5 years ago, all be it at a smaller scale, but still employing same robust concepts and using fast back test engines. I used TradeStation in the past to test ideas and build trading systems, then tried some professional software packages, but you always have to come up with your own ideas, stress test them and hope to find something usable. There are high grade packages that is suitable for hedge funds but accessible to professional traders, but you need to learn programming and to fork thousands of dollars.
This bring us to the current crop of #GeneticSystemBuilders. There are many and each of them have its advantages and disadvantages. Three is no perfect solution, but #StrategyQuantX is pulling way ahead in this race. Below are some of the features:
No programming knowledge is necessary
Build unlimited number of trading strategies
Develop strategies on any market, timeframe or multiple markets and timeframes
Filter all generated systems, using multiple criteria
Stress test filtered systems with multiple robustness tests to avoid over fitting systems
Export strategies to TradeStation, Ninja Trader, MetaTrader
Improve existing strategies
Run Projects, that does all the above automatically
The latest version of StrategyQuant X can generate and stress test any instrument. In fact since it generate systems based on multiple timeframes and multiple instruments, even external data such as commitment of traders, or market regime can be used in the generation of strategies.
I have been testing StrategyQuant X for couple of months now, and I have to say that I am really impressed. The software can research and test ideas, then generate strategies based on multiple conditions that can be customized by the user, then back test those strategies, optimize them, stress test them, and come up with final systems that can be formatted for #MetaTrader, #TradeStation or #NinjaTrader. The code generated for these platforms is open and you can modify/add/remove any part as you like.
Since there is no programming needed, everyone can design and test their own ideas, generate multiple systems based on those ideas and vet them using multiple stress tests. SQX can generate thousands of iterations, to be stress tested later for vetting.
While you can test your own ideas, SQX have ready-made concepts for building systems like trend following and mean reversion, but each time you run it, even on the same time frame and same instrument, it will come up with different way to trade it.
StrategyQuant X consist of six sections:
This is where you build your strategies. You can choose a pre-configured concepts, like Trend Following, Mean Reversion, Daily, etc. or you can customize the pre-configuration, or you can do it randomly.
Once you picked a model, then you pick the instrument, timeframe, 2nd instrument or more, 2nd timeframe or more.
Define your data, session times, OOS windows, commissions, slippage, spread, signal time range allowance, profit target, stop loss, maximum trades per day, exits.
Chose your building blocks. Here you can choose your favorite indicators/signal generator, order type, exit type.
Configure capital and position sizing
Choose your fitness that will be used to rank your systems. This can be a fixed value like Net Profit, or a Weighted value like (50% net profit, annual return %)
Finally pick filters that will pass/fail systems, like (number of trades per month/year, IS net profit, OOS net profit, Return/Drawdown), etc. the possibilities are endless.
Once you have thousands of filtered systems you need to stress test them
Restest with higher precision data (look inside bar)
Monte Carlo trade manipulation
Noise and Variance test
System Parameter Permutation
Walk Forward Optimization
Walk Forward Matrix
This is more detailed version of 2.7 and 2.8 above and can be done on loaded individual strategies plus produce the next set of parameters for your next OOS.
SQX just recently added their own data feed, for currencies, stocks, futures. It is still in Beta. This is very helpful, however you can import your own data from any source. You can shift time zone of any time series. For example if you want to trade DAX and you live in Los Angeles, you can shift the data to your time zone.
If you load high resolution data, say like 2 minute, then you can create any bar from that data, like 15 minutes, 22 minutes, 240 minutes, etc.
This is what sold me on this software. Here you run any task in sequence or individually and copy, filter results to your heart content. So you instruct the program to generate soy bean strategies based on 30 minutes bars, filter generated systems, stress test them, and save remaining robust systems. Basically let the computer run, dedicate the number of CPU cores and RAM to SQX, so that there is enough for you to browse/email/youtube, etc. and it will do it’s thing and let you know when it’s done. Of course you can still look at the dashboard to see which step is working now and how many systems passed certain step.
This is the last section, where you can add your indicator, signal, or modify existing one and customize it to your liking.
As you can see the software is very powerful and the possibilities are endless, yet very easy to start with and as you learn more, you apply more.
The team behind it is very responsive and have been squashing bugs as they are reported. In fact they have a list where anyone can report a bug, and you can see it’s progress online.
Strategy Quant X is very sophisticated, yet very easy to use. It is full of features for beginners and advanced traders. That’s why I will be doing a series of videos about each section and its advanced features.
A fully functional trial version is available here: https://strategyquant.com/download/
Subscribe to be notified whenever I post a new video blog to learn about StrategyQuant X and other software packages and trading ideas.
If you want to learn how to achieve the holy grail portfolio, then check my Algo Trading Master Class, for building robust trading systems without any programming knowledge or previous experience, by using proven building blocks and state of the art software to test and verify every step.