If you have read about trading systems you will have heard about strategy builders. What these constructors or builders like StrategyQuant do is search for systems based on conditions that we set.
Imagine that you want to create a strategy that only buys on Mondays in the Sp&500 index. That it operates with an entry indicator and an exit indicator. From this simple configuration, there are tools with which you can automatically search for systems or strategies that have been profitable.
This has a major drawback, finding strategies that do not make much sense but are winning and It seems to work but it is the result of chance (combination of conditions that may not make sense).
Well applied, we can get a lot out of it. In fact, these tools are used by our community in the city to create trading systems without programming.
One of the most advanced and with the most integrations with other platforms is StrategyQuant. Lately I have created most of my strategies here, so it deserved an article talking about this software, which in my opinion is the best on the market. you also have a full course on StrategyQuant with some templates and filters that I use to search for systems.
Well, let’s get down to business.
In this post I will explain everything you need to know to start using Strategy Quant and that you understand how it works. Aim:
1. What is StrategyQuant?
StrategyQuant is software that uses techniques to create trading systems without the need to know programming.
In addition to generating and finding algorithmic systems, StrategyQuantis an analysis tool to test our trading ideas (robustness test), analyze strategies and portfolios, find optimizations, perform targeted searches based on seasonal patterns, etc.
2. How does StrategyQuant work?
StrategyQuant It consists of three well-differentiated blocks or sections:
- Generation of strategies.
- Robustness test.
Let’s see what each of these stages consists of, so that you can easily develop your strategies.
3. Generation of strategies with StrategyQuant
There are basically two ways of generating systems on this platform:
3.1. Random Generation
It consists of generating strategies by brute force, as its name indicates, randomly depending on the parameters and conditions that you have configured.
3.2. Genetic Generation
Here the idea is to create the best strategy or the one closest to it, within a universe of combinations of base strategies, called parent strategies, using machine learning. Genetic generation is the fastest approach.
Let’s see an example to understand it better. Suppose we have two base strategies. One strategy uses the stochastic indicator and the RSI to generate the entry signal and the other strategy uses the ATR indicator and the MACD as the entry signal.
From these base strategies, new combinations are created by exchanging their parts. These new strategies are called child strategies and contain characteristics of their parent strategies.
This process of genetic generation through machine learning allows us to gradually improve the strategies until the best strategy, for example, is a strategy that uses the Stochastic indicator and the MACD to give the entry signal. That is, a combination of the base strategies.
I recently wrote an article about genetic algorithms in tradingthere I explain more in detail all this.
3.3. How to create strategies with StrategyQuant X software
The first thing we must do before starting is to download and add the data of the asset where we want to create the strategy in the “Data Manager” section.
You may have noticed that this data manager is QuantData Managera platform that is integrated into the SQ Software (both created by the same company) and that allows us to download data very comfortably.
The first thing we need to validate and create our strategies is asset data, that’s why it’s so important. I also leave you a video about QuantData Manager where I explain how it works. Within the city
Let’s get back. You have the entire process of configuring parameters for the generation of strategies in the “Builder” section tab. Clicking on “Builder” takes us to a main window with three tabs: Progress, Full Settings and Results.
In the tab “full settings”, we will find all the detailed options of the initial parameters that we need to configure to generate our strategy.
Among the parameters that we must configure in the case of a Forex strategy, we have:
Although StrategyQuant X offers us the option to generate Multi-Timeframe or Multi-Symbol strategies, if you are a beginner using the platform I recommend you start using the “Simple strategy” option.
- Generation mode (Build mode).
There are two options here: Genetic evolution and Random generation. Choosing one or the other depends on the goal you have set for yourself.
- Conditions or rules of entry and exit.
StrategyQuant X offers a large number of predefined indicator-based base conditions. During the generation process the software mixes the basic components that we have selected and looks for the best combination of input and output conditions.
You can choose to use a Fixed SL or TP, using indicator levels and based on ATR (indicator of volatility) changes. The latter is an option that you should take into account.
- Managing our money
The initial starting capital. In my case, I usually put in $10,000 and then adjust my account based on the funds I want to allocate. To simplify.
- Historical backtesting time.
- Trading platform.
- Filters or acceptance criteria.
- And many more.
5. Robustness Test
after you trading strategy has passed the acceptance filters, you can make small changes and measure how well they have worked. By doing , you increase the probability that you will generate profit in the future.
StrategyQuant X also gives you the option to perform the robustness test immediately during the strategy generation process, which you configure in the “Cross Check (Robustness)”. You can also perform the robustness test after the generation process in the section “retester”, the latter is the option that I recommend.
Now you can apply simpler tests, changing by changing the active or initial timeframe. Of all the test options available, I consider these three key:
5.1. Higher Backtest Precision
When you select the timeframe to generate your strategy, the software uses its historical data to generate candles in the selected time frame, so you are analyzing the behavior of the closing price of each candle ignoring what happened inside the candle.
What this test does is retest the previously generated strategy, but now with more precise candles of one minute M1 that are within the longer temporal candle, for example, one hour H1.
5.2. Monte Carlo – Evenings Manipulation
Through this test, the trades that make up the newly generated trading system are modified, thus evaluating whether the performance or profitability of our system is the product of the vast majority of trades or if, on the contrary, it depends only on a small portion of trades. The latter should be avoided.
The simulations could be, for example: remove a percentage of trades and see how our profits vary, reorder a percentage of trades and see how the Drawdown changes, etc.
5.3. Monte Carlo–Retest Methods.
Depending on the market that we decide to operate, we will be exposed to more or less volatility. This volatility can cause market conditions to change abruptly and make our trading system unprofitable.
Through this method you expose the system to extreme situations and evaluate how it responds. For example: changing the value of the historical data of the original simulation by a certain percentage, increasing the spread, slippage as usually happens in periods of high volatility, modifying the parameters of the indicators used by the trading system, etc.
6. Optimization of strategies
We have reached the phase of polishing the last details of our strategy. Here it is not about optimizing the strategy to get the most benefits, it is about trying a good number of optimizations and finding the most robust one.
StrategyQuant X has a very interesting option to perform the optimization, Walk Forward Matrix.
As you know, past earnings do not guarantee future earnings, it is because of them that our objective is not to predict what is going to happen but to find the scenario that is most likely to occur based on what has happened in the past.
The Walk Forward Matrix It allows you to segment the historical backtesting time, in which we already know that our strategy works, into smaller time periods.
Suppose you have chosen a 10-year historical. So, you can test optimizations in periods smaller than 1 year (In Sample IS – In sample) called “runs” and choose the best parameter optimization in each runs. This optimization is then tested in the 3 months (Out of Sample OOF) following each runs and see the results of each runs.
The results of each runs are passed through the acceptance filters and are displayed in the form of a matrix (runs vs %OOF). If they pass the filters they will appear in green or otherwise in red.
Select the matrix box optimization that has the most green boxes around it because this increases the chances that your strategy will continue to be profitable if the market changes in the near future.
StrategyQuant Right now it is one of the most versatile software out there because it allows us to create strategies, evaluate and optimize them, evaluating their robustness. You can save and export the code to major trading platforms such as MetaTrader4 (MT4), MetaTrader5 (MT5) Y Tradestation.
Although with these tools you can apply trading systems automatically without the need to know programming, you are going to throw your money away if you do not know how trading works, you do not know how to evaluate and test a strategy correctly or you do not have a clear idea of what you want to build from scratch.
That said and yes, here you can download a trial version (demo) to use it free for two weeks.
You also have here a podcast where I talk about StrategyQuant and how I take advantage of this platform.