Transformer Graph Convolved Dynamic Mode Decomposition for Trading Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating trading signals for diverse asset classes are inefficient in modeling dynamics and require substantial computational resources, leading to inaccurate and delayed signal generation.
Innovation Solution
The implementation of a transformer graph convolved dynamic mode decomposition (TGCDMD) layer, which combines transformer modules with graph convolution layers to automatically determine asset relationships, reducing model parameters and computational costs, enabling timely and accurate trading signal generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods are used for modeling asset dynamics, then model accuracy can be maintained, but computational resources and time requirements increase substantially
Solution Approach 1:
The patent transforms the asset price data into a different parameter space using dynamic mode decomposition, decomposing the data into mode coefficients that capture essential dynamics. This parameter transformation enables more efficient computation while preserving the critical information needed for accurate trading signal generation.
Solution Approach 2:
The patent segments the asset dynamics into multiple independent modes through dynamic mode decomposition. Each mode represents a distinct timescale or pattern of behavior, allowing the system to process and analyze different aspects of asset dynamics separately and efficiently, reducing overall computational burden.
2Measurement precision
If comprehensive asset data is processed to improve signal accuracy, then trading signal precision increases, but computational complexity increases
Solution Approach 1:
The patent extracts only the essential dynamic modes from the comprehensive asset data using dynamic mode decomposition. Instead of processing all raw data points, the system identifies and extracts the dominant modes that capture the essential behavior, significantly reducing model complexity while maintaining signal accuracy.
Solution Approach 2:
The patent employs dynamic mode decomposition to capture the time-varying dynamics of asset prices in a computationally efficient manner. The method adapts to changing market conditions by identifying evolving modes dynamically, allowing the model to remain simple yet responsive to complex market behaviors.
3Reliability
If more model parameters are used to capture asset relationships, then modeling accuracy improves, but computational costs increase
Solution Approach 1:
The patent applies partial action by using only the most significant dynamic modes to model asset relationships. Instead of incorporating all possible parameters and modes, the system selects and uses only the dominant modes that contribute most to modeling accuracy, reducing computational costs while maintaining reliability.
4Speed
If real-time processing is implemented for timely signal generation, then response speed increases, but processing accuracy may deteriorate
Solution Approach 1:
The patent performs preliminary decomposition of asset data into dynamic modes in advance. This preprocessing step organizes the data into a structured format that can be rapidly processed in real-time without sacrificing accuracy, as the essential dynamics are already captured in the mode coefficients ready for quick analysis and signal generation.
Data Source
AI summary
A computer-implemented method of generating trading signals for one or more financial instruments by an electronic device. The method includes: receiving, by a processor of the electronic device, time series data of one or more assets from a data source; pre-processing, by the processor, the time series data; extracting, by the processor, features from the pre-processed time series data to obtain a plurality of time series of feature vectors; calculating, by the processor, an affinity matrix representing pairwise similarity between each pair of the assets; calculating, by the processor, a Transformer Graph Convolved Dynamic Mode Decomposition of data matrix assembled from extracted features; determining, by the processor, a prediction of the feature vectors for the assets; extracting, by the processor, trading signals from the predicted feature vectors; feeding the trading signals to the financial instruments.