Combining Maths + Tech to trade the FX market

Combining maths + behavorial psychology + tech to auto-trade the FX market profitably

Inspired by the strides of billionaire mathematician - Jim Simons in algorithmic trading, technical lead - Dr. Vincent Shinzo(PhD) initiated a pet project stemming from his dissertation in predictive modeling from The University of Tokyo in collaboration with expert economists and psychologists from the University of Amsterdam, together ViNx Finance Research Club was formed.

We conducted a comprehensive research in the worlds largest financial market and successfully invented mathematical models that uses historical price data, technical and economic indicators, volatility measures and IMPORTANTLY sentiment and human behavioral analysis to accurately predict short term price direction in the FX market.

These models + processes was developed into an easy to use auto-trading solution - ViNxAutoT.

On the backend, It works by monitoring and comparing real time market situation to a large set of data. Should the market condition match a certain criterium, a prediction is made and trade automatically activated to profit off it.(Liking it to how tesla autonomous driving works - it gets data from different sensors in the vehicle, processes it and executes).

Thing is, just as tesla, the models doesn't 100% get it right, understandably so, but when results were analyzed, we figured a pattern that can be used to efficiently profit off this predictions.

The pattern:

Logic 1 (Execute only one trade per day): Statistically, focusing on a single trade per day lowers the risk/exposure.

Logic 2 (Trade duration of 1-7 hours): Our research determined that keeping trades active between 1 to 7 hours strikes a balance between capturing profitable opportunities and minimizing exposure to market fluctuations.

Logic 3 (Maximum profit of 3% per trade): Close trade if it hits 3% profit and call it a good day - The best time to leave the stage is when the ovation was the loudest. Result data showed 3% profit was that mark.

Logic 4 (Maximum loss of 2% per trade): Close trade if it hits 2% loss - "cut your losses early" and call it a bad prediction day.

This approach ALWAYS yielded a net profit monthly.

Note: More often, trade closes due to the duration factor(1-7hours) and most times the trade is already on profit, sometimes on a smaller profit margin eg. 0.01%, 0.9% or on a higher margin ex. 1% to 2.9% or in fewer instances on a loss ex. -1.2%, - 0.9%...,

However, when daily result are computed over a month, it always returned a net profit of between 3% to 32%. It NEVER recorded a net loss.

These principles has kept our solution 100% efficient and consistent. It WORKS!

We are revolutionizing the financial market by developing cutting edge solutions that helps the average person partake in the exciting world of trading.

To get started, we recommed signing up for the demo to experiment the hour band that best works for you.

Fx auto-trading software-as-a-service

Experience the ease of forex trading on autopilot with our advanced software, allowing our subscribers to effortlessly profit without the hassle.

Overview

Combining maths + behavorial psychology + tech to auto-trade the FX market profitably

These models + processes was developed into an easy to use auto-trading solution - ViNxAutoT.

Thing is, just as tesla, the models doesn't 100% get it right, understandably so, but when results were analyzed, we figured a pattern that can be used to efficiently profit off this predictions.

The pattern:

Logic 1 (Execute only one trade per day): Statistically, focusing on a single trade per day lowers the risk/exposure.

Logic 4 (Maximum loss of 2% per trade): Close trade if it hits 2% loss - "cut your losses early" and call it a bad prediction day.

This approach ALWAYS yielded a net profit monthly.

However, when daily result are computed over a month, it always returned a net profit of between 3% to 32%. It NEVER recorded a net loss.

These principles has kept our solution 100% efficient and consistent. It WORKS!

We are revolutionizing the financial market by developing cutting edge solutions that helps the average person partake in the exciting world of trading.

To get started, we recommed signing up for the demo to experiment the hour band that best works for you.

Fx auto-trading software-as-a-service

Experience the ease of forex trading on autopilot with our advanced software, allowing our subscribers to effortlessly profit without the hassle.

Title: Development of a Forex Price Direction Prediction Model

Abstract

This project aims to develop a comprehensive model for predicting the direction of Forex market prices. The model integrates historical price data, technical indicators, economic indicators, sentiment analysis, and volatility measures using advanced machine learning techniques. The project outlines the components, architecture, and methodology used to create the prediction model.

1. Introduction:

The unpredictable nature of the Forex market has led to the exploration of predictive models that can guide trading decisions. This project seeks to create a sophisticated prediction model that leverages a combination of diverse data sources and indices. By integrating historical price data, technical and economic indicators, sentiment analysis, and volatility measures, the project aims to provide insights into potential price movements.

2. Literature Review:

Previous research in Forex price prediction has primarily focused on incorporating technical indicators and machine learning techniques. Our project builds on this foundation by considering a more comprehensive set of inputs, including economic indicators, sentiment analysis, and volatility measures. Existing studies have highlighted the potential of neural networks and LSTM for capturing complex relationships and temporal dependencies in financial data.

3. Methodology:

To create our Forex price direction prediction model, we follow a systematic methodology:

- Data Collection and Preprocessing: We collect historical price data for selected currency pairs and calculate various technical indicators. Economic indicators are sourced from reputable financial databases, while sentiment analysis is performed on relevant news articles and social media data. Volatility measures are computed using methods such as the Average True Range.

- Model Architecture: We design a multi-layer neural network that takes normalized historical price data and selected indicators as inputs. We also implement an LSTM network to capture temporal dependencies in the data. Ensemble learning techniques are applied to combine predictions from both neural network and LSTM models.

- Experimentation and Results: Our experiments involve training and testing the model on historical data. We evaluate the model's performance using metrics such as accuracy, precision, recall, and F1-score. We compare the individual predictions of the neural network and LSTM models with the ensemble predictions to assess their effectiveness.

4. Data Collection and Preprocessing:

- Historical Price Data: We collect historical price data for currency pairs (EUR/USD, GBP/USD, etc.) from reliable financial data providers. Each data point includes open, high, low, and close prices for a specific time interval.

- Technical Indicators: A comprehensive set of technical indicators, including Moving Averages, Relative Strength Index (RSI), Moving Average Convergence Divergence (MACD), and Bollinger Bands, is calculated using historical price data.

- Economic Indicators: Key economic indicators like GDP growth rate, inflation rate, interest rate, and trade balance are collected from reputable financial databases.

- Sentiment Analysis: We employ Natural Language Processing techniques to analyze sentiment in news articles and social media data related to the Forex market. Sentiment scores are generated for each data point.

- Volatility Measures: The Average True Range (ATR) is computed to measure market volatility over a specific period.

5. Model Architecture:

- Neural Network: Our neural network consists of input, hidden, and output layers. The input layer accepts normalized historical price data and selected indicators as features. Hidden layers capture intricate relationships between inputs, while the output layer produces a probability distribution for price movement.

- Long Short-Term Memory (LSTM): The LSTM network is designed to process sequential data and capture temporal patterns. It takes the normalized historical price data and indicators as input sequences to predict future price direction.

- Ensemble Learning: We combine predictions from the neural network and LSTM models using ensemble techniques such as weighted averaging or stacking. This approach aims to enhance prediction accuracy by leveraging the strengths of both models.

6. Experimentation and Results:

- Experimental Setup: We split the historical data into training, validation, and test sets. The models are trained on the training set and validated on the validation set. The test set is used to evaluate the models' performance in a real-world scenario.

- Performance Metrics: We evaluate the models using accuracy, precision, recall, F1-score, and possibly other relevant metrics. These metrics provide insights into the models' ability to predict different price movements accurately.

- Comparative Analysis: We compare the individual predictions of the neural network and LSTM models with the ensemble predictions. This analysis helps us determine the effectiveness of the ensemble approach in improving prediction accuracy.

7. Discussion:

- Interpretation of Model Outputs: We interpret the probability distribution provided by the model to understand the likelihood of various price movements.

- Strengths and Limitations: We discuss the strengths of our model, such as its ability to incorporate a diverse range of indicators. We also address limitations, such as potential data noise and the complexity of the model.

- Future Improvements: We explore possible avenues for refining the model, including exploring additional indicators, fine-tuning hyperparameters, and implementing more sophisticated ensemble techniques.

8. Conclusion:

In this project, we have developed a comprehensive Forex price direction prediction model that leverages historical price data, technical and economic indicators, sentiment analysis, and volatility measures. Our findings shed light on the potential of integrating diverse data sources to enhance prediction accuracy and guide trading decisions in the Forex market.

Rules Of Engagement?

Every milli second, ViNxAutoT simultaneously mointors the different markets looking for an entry point that meets the requirements of either of the 7 models

Whichever condition comes first becomes the trading strategy to be adopted for the day.

It takes only one trade per day.

Exit trade when it hits 3% in PROFIT of traded capital.

Exit trade when it hits 2% in LOSS of traded capital.

Exit trade exactly on the 1-7th Hr. Either in PROFIT or LOSS OR NEUTRAL.

- Trade closes when it hits 3% in profit OR 2% in loss OR 1-7hours in duration. Whichever comes FIRST!

These principles has kept us 100% efficient. It took years of research to get here!

Research Brief

Title: Development of a Forex Price Direction Prediction Model

Abstract

1. Introduction:

2. Literature Review:

3. Methodology:

To create our Forex price direction prediction model, we follow a systematic methodology:

4. Data Collection and Preprocessing:

- Economic Indicators: Key economic indicators like GDP growth rate, inflation rate, interest rate, and trade balance are collected from reputable financial databases.

- Volatility Measures: The Average True Range (ATR) is computed to measure market volatility over a specific period.

5. Model Architecture:

6. Experimentation and Results:

7. Discussion:

- Interpretation of Model Outputs: We interpret the probability distribution provided by the model to understand the likelihood of various price movements.

8. Conclusion:

Rules Of Engagement?

Every milli second, ViNxAutoT simultaneously mointors the different markets looking for an entry point that matches either of the 7 strategies

Whichever condition comes first becomes the trading strategy to be adopted for the day.

It takes only one trade per day.

Exit trade when it hits 3% in PROFIT of traded capital.

Exit trade when it hits 2% in LOSS of traded capital.

Exit trade exactly on the 1-7th Hr. Either in PROFIT or LOSS OR NEUTRAL.

- Trade closes when it hits 3% in profit OR 2% in loss OR 1-7hours in duration. Whichever comes FIRST!

These principles has kept us 100% efficient. It took years of research to get here!

Model Components:

1. Historical Price Data:

P_{t = (O_t, H_t, L_t, C_t)}

2. Technical Indicators:

MA_t, RSI_t, MACD_t, BB_t

3. Economic Indicators:

GDP_t, Inflation_t, InterestRate_t, TradeBalance_t

4. Sentiment Analysis:

Sentiment_t

5. Volatility Measures:

ATR_t

Model Architecture:

1. Data Preprocessing:

P_{normalized, t} = (P_t - mean(P)) / std(P)

2. Feature Selection:

X_{selected, t} = [MA_t, RSI_t, MACD_t, BB_t, GDP_t, Inflation_t, InterestRate_t, TradeBalance_t, Sentiment_t, ATR_t]

3. Weighted Aggregation:

W_MA, W_RSI, ..., W_ATR

4. Neural Network:

NN(P_{normalized, t}, X_{selected, t}; Θ)

5. Long Short-Term Memory (LSTM):

LSTM(P_{normalized, t}, X_{selected, t}; Φ)

6. Ensemble Learning:

Ensemble(NN, LSTM)

Model Output:

The model's output is a probability distribution indicating the likelihood of price movement (upward, downward, or neutral) within a specified time frame.

Why Subscribe to ViNxAutoT?

In today’s changing economic landscape, reaching your financial goals is often a challenge. Individuals and families have a lot of choices when making a decision on where to put their money that will yield long term profit. ViNxAutoT is that vehicle that will balance out this equation. It has shown consistency beyond reasonable doubt and has met all the key risk management criteria.

Model Components