FAQ

What data is used in the EASY Trading AI strategy?

EASY Trading AI utilizes a comprehensive and sophisticated approach to data gathering and preparation, essential for the successful implementation of its trading strategies across various financial markets. The cornerstone of this data-driven strategy is the use of high-resolution tick data. Below, we detail the kinds of data used, how it’s prepared, and subsequently utilized in model training.

Tick Data

Definition and Importance: Tick data represent the most granular level of information in the trading world, capturing every single change in price, no matter how small. Each ‘tick’ includes the price at which the transaction occurred and the exact timestamp of the transaction. This data is paramount in markets like Forex and cryptocurrencies where price movements can be swift and sudden, requiring a high level of precision for effective trading strategies.

Data Collection

Tick data is collected real-time from trading platforms and includes:

Price Changes: Every alteration in price recorded at the moment it occurs.
Volume Data: Information on the volume traded at each tick, providing insight into the weight of each price change.
Bid/Ask Prices: The prices at which buyers are willing to buy (bid) and sellers are willing to sell (ask), offering insights into supply and demand dynamics.

Data Preparation and Cleaning

Before this granular data can be fed into machine learning models, it must be meticulously prepared and cleaned, a process that involves several crucial steps:

1. Data Cleaning:

Removing Outliers: Occasionally, erroneous data points arise from glitches in data transmission or processing. These outliers, which do not reflect actual market movements, are identified and removed.
Handling Missing Values: Gaps in tick data can occur due to various reasons, such as connectivity issues. Depending on the scenario, methods like interpolation might be used to estimate missing ticks based on nearby data points.

2. Data Transformation:

Normalization: Tick data, especially price and volume, are normalized to ensure that the model isn’t biased by absolute values which can vary significantly across different instruments and markets.
Feature Engineering: From raw tick data, additional features such as moving averages, price change velocity, and volatility indicators are computed to provide models with more context on market conditions.

Model Training

With the data prepared, the following methodologies are typically adopted to train EASY Trading AI’s models:

1. Feature Selection:

Utilizing statistical tests and machine learning algorithms to select the most predictive features, reducing the dimensionality of the data and enhancing model performance.

2. Algorithm Training:

Supervised Learning: Predictive models are trained using historical tick data where the outcomes are known (e.g., price up, price down), enabling the models to learn patterns associated with specific market movements.
Reinforcement Learning: The models are further trained in a simulated environment where they can learn from their actions based on reward mechanisms, adapting their strategies in response to dynamic market conditions.

3. Validation and Backtesting:

Once trained, the models undergo rigorous backtesting, using separate sets of historical tick data to ensure that they perform well across different time periods and under various market conditions.

EASY Trading AI doesn’t stop learning post-deployment. The models constantly receive new tick data, allowing them to refine and adjust their predictions and strategies based on fresh market insights. This continuous learning cycle is crucial for maintaining the efficacy of trading strategies in the highly volatile and ever-evolving financial markets.

Overall, the blend of high-resolution tick data, thorough preparation and cleaning processes, and robust model training methodologies ensure that EASY Trading AI remains precise, adaptive, and effective across all supported trading platforms and instruments.