All about MQL5 Wizard : create robots without programming. - page 11

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 76): Using Patterns of Awesome Oscillator and the Envelope Channels with Supervised Learning

From our last article, we introduced the indicator pairing of the Awesome-Oscillator and the Envelope-Channels, and from the testing of that pair 7-8 of the 10 patterns walked forward on a 2-year test window. We usually follow up the introduction of an indicator pair with an exploration of what impact, if any, machine learning can have on the performance of these indicator signals. This article is no exception, and thus we are going to examine how the patterns 4,8,and 9 can be affected/influenced if we supplement their signals with a supervised-learning network as a filter. For our network, we are using a CNN whose kernels/channels are sized by the dot product kernel with cross-time attention.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 77): Using Gator Oscillator and the Accumulation/Distribution Oscillator

Continuing with our exploration of indicator pairings in generating key trade signals, we now look at the Gator oscillator and the Accumulation/Distribution oscillator (AD). Unlike the AD, the gator oscillator is direction neutral, however we use both for their complimentary use of tracking momentum and volume as a signal confirmation. Direction neutrality for the gator means we will rely on another indicator or price action in order to establish the trend. For this article, we are using the latter.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 78): Using Gator Oscillator and the Accumulation/Distribution Oscillator

In our last article we introduced 5 signal patterns from the pairing of the Gator and the Accumulation/Distribution oscillators, in this article we consider the final 5 of this set of 10. We have been looking at 10 signal patterns for each indicator pairing, and we will be maintaining this format. From the testing in the last article which was done on the pair GBP JPY on the 30-minute timeframe, our results indicated that the patterns 0, 3, and 4 struggled to forward walk, however before we can choose what to improve with supervised learning, let's complete examining and testing of patterns 5 to 9.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 79): Using Gator Oscillator and Accumulation/Distribution Oscillator with Supervised Learning

In our last 2 articles, we tested, as usual, ten signal patterns while using the indicator pairing of the Gator oscillator and Accumulation/Distribution oscillator. In doing so, as has been our practice, we found three consistent laggard patterns: 0, 3, and 4. Rather than discard or disregard these, this article aims to explore whether supervised learning can revive their performance. We employ a CNN enhanced with kernel regression and dot product similarity and examine if networks with such architecture can extract hidden value from signals that sometimes initially appear weak. As per the last two articles all testing is on the pair GBP JPY and on the 30 minute time frame.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 80): Using Patterns of Ichimoku and the ADX-Wilder with TD3 Reinforcement Learning

In the family of RL algorithms, Twin Delayed Deep Deterministic Policy Gradient (TD3) is emerging, in some circles, as a solid candidate for financial applications. TD3 is designed for continuous action spaces, making it particularly well-suited to trading problems where position sizing and timing are not binary but require fine-grained control. If we compare it to its predecessor, DDPG, TD3 brings in crucial stability improvements such as using more than one critic, adding noise smoothing to the target actions and also delaying policy updates to prevent overfitting to transient fluctuations.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 81): Using Patterns of Ichimoku and the ADX-Wilder with Beta VAE Inference Learning

Almost every trader has come to accept that markets move in cycles of optimism and pessimism, and yet we have very few out-of-the-box tools that capture these cycles with the consistency required for profitable trading. Recently, global markets have leaned toward bearish sentiment, with sudden selloffs and shallow rebounds becoming more frequent. In such a setting, mechanical strategies based on lagging indicator rules are bound to produce false signals, as volatility shakes out trades that would otherwise have followed through in calmer conditions.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 82): Using Patterns of TRIX and the WPR with DQN Reinforcement Learning

When trading with robots or Expert Advisors, the pursuit for structured and repeatable trading rules usually starts with technical indicators that are familiar. The tendency is often to dabble in oscillators, moving averages, and price based patterns in order to construct strategies that can outlive changing market regimes. Among these, the TRIX, aka Triple Smoothed Exponential Moving Average, as well as the WPR which is also Williams Percent Range, are a classic pair. TRIX usually captures momentum by sieving out short-term price noise, while WPR highlights overbought or oversold situations. Both indicator patterns, therefore, when combined, can complement each other and help spot turning points or continuations in price.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 83): Using Patterns of Stochastic Oscillator and the FrAMA — Behavioral Archetypes

For this article, we are once again doing a deep plunge into a created custom Signal Class that brings together the Stochastic Oscillator and the Fractal Adaptive Moving Average. This infusion of two well known indicators is meant to give us a powerful hybrid system whose indicator-end-goal is to provide a ‘pipelined’ binary representation of these indicators’ logic. This primarily serves as a digital noise filter for market patterns. As usual, we are going to consider 10 indicator patterns that cater to different sorts of markets, with the market types under consideration being trending/mean-reverting, correlated/decoupled, and low-volatility/high-volatility. Each of these ten patterns that we’ll define in the custom signal class stands for a particular setup of either FrAMA slope or Stochastic positioning or price-pattern.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 84): Using Patterns of Stochastic Oscillator and the FrAMA - Conclusion

From the last article, we examined the first 5 signal patterns of the Indicator pairing Stochastic-Oscillator and Fractal Adapting Moving Average. From our small test window, all appeared to have profitable forward walks, with training done over a year and the validation performed over the subsequent year.

[Sergey Golubev]

MQL5 Wizard Techniques you should know (Part 85): Using Patterns of Stochastic-Oscillator and the FrAMA with Beta VAE Inference Learning

Within MetaTrader 5’s ecosystem, the MQL5 Wizard does stand as a solid tool that enables traders to rapidly prototype and deploy new trade ideas. As we have covered in past articles, this all happens without getting into low-level coding. At its core, the wizard utilizes a modular framework that allows traders to choose from predefined signal classes, or money management strategies, or trailing stop mechanisms. The ability to have a plug and play approach in assembling an Expert Advisor has the unintended consequence of democratizing algorithmic trading, which makes trading more accessible to individuals with varying level of expertise, and this should have the long term effect of boosting market liquidity, all else being equal.