Schnick_Demo
This EA is a demonstration script for a Support Vector Machine (SVM) learning tool. It generates synthetic training data based on a predefined 'Schnick' classification function, trains two SVMs (one with clean data and one with randomly inserted errors), then tests and compares their classification accuracy. It does not execute any actual trades.
Open the interactive view (equity curve, symbol switch)Key facts
Backtest results (H1, about 1 year)
Not backtested yet — this EA could not be converted automatically.
Strategy analysis
The strategy concept of this EA is based on Support Vector Machine (SVM) binary classification, using 7 input features (height, weight, number of legs, number of eyes, arm length, average speed, call frequency) to determine whether a fictional creature 'Schnick' belongs to a specific species. Entry conditions do not involve real market trading; instead, 5000 training samples are generated once during OnInit, each containing 7 randomly generated feature values. Exit conditions are not applicable since the EA only performs a classification test and terminates during initialization. Position sizing and money management are completely absent, as this is a purely machine learning demonstration tool with no financial trading logic. The EA demonstrates SVM robustness to noisy data by comparing classification accuracy between clean training data and training data with 500 randomly inserted errors. The classification decision boundary is defined by the hard-coded multi-dimensional range checks in the isItASchnick function, which is essentially a multi-dimensional hyperrectangle classification problem.
Strengths
- Clearly demonstrates how to integrate SVM usage within MQL5
- Intuitively shows SVM robustness to noise by comparing training results with and without errors
- Clean code structure with well-separated functions and ample comments for understanding
Weaknesses
- Does not execute any actual trades; it is merely a proof-of-concept demonstration
- Classification criteria are hard-coded in the isItASchnick function, lacking flexibility and real-market applicability
- Training and testing are performed once in OnInit, with no capability for continuous learning and classification on live market data
Possible improvements
- Apply SVM classification to real market indicator data to implement machine-learning-based trade signal generation
- Add real-time classification capability, using the trained SVM in OnTick to classify incoming data and generate trading signals
Risk assessment
Overall risk: Safe
- This EA does not execute trades, posing no market risk
- All computations are completed during initialization, with no impact on runtime performance
Evaluation scores
Parameters
| Parameter | Default | Description |
|---|---|---|
N_Inputs | 7 | Number of SVM input features |
N_TrainingPoints | 5000 | Number of training samples |
N_TestPoints | 5000 | Number of test samples |
OP_TOLERANCE | 0.01 | SVM training error tolerance parameter |
N (insertRandomErrors) | 500 | Number of random errors inserted into training data |
Similar Machine Learning EAs
Backtests are historical simulations and do not guarantee future results. Not financial advice.