Shoriful025/market_carrier_trend_predictor

Market Carrier Trend Predictor

Overview

The Market Carrier Trend Predictor is a time-series forecasting model focused on identifying the primary Carrier (বাহক) wave of asset prices. In financial modeling, the "Carrier" refers to the underlying momentum or trend that supports localized price fluctuations. This model is specifically trained on high-frequency cryptocurrency data to predict future price directions ($t+24$ hours).

Model Architecture

This repository implements the Informer architecture, which is optimized for long-sequence time-series forecasting:

• ProbSparse Self-Attention: Reduces the computational complexity from $O(L^2)$ to $O(L \log L)$.
• Self-Attention Distilling: Compresses the temporal resolution to handle long-range dependencies in the "Carrier" frequency.
• Generative Style Decoder: Predicts the entire trend sequence in one forward pass.

The model evaluates prediction accuracy using Mean Squared Error (MSE): $$MSE=\frac{1}{n}\sum _{i=1}^n(Y_i-{\widehat{Y}}_i{)}^{2}MSE\; =\; \backslash frac\{1\}\{n\}\; \backslash sum\_\{i=1\}^\{n\}\; (Y\_i\; -\; \backslash hat\{Y\}\_i)^2$$

Intended Use

• Trend Identification: Isolating the long-term "Carrier" wave in volatile markets.
• Risk Management: Providing early warning signals for trend reversals.
• Algorithmic Trading: Serving as a primary feature for automated decision-making engines.

Limitations

• Black Swan Events: The model cannot account for external shocks such as regulatory changes or major security breaches.
• Volatility Scaling: During periods of hyper-volatility, the "Carrier" signal may become obscured by stochastic noise.
• Lagging Indicator: As with all time-series models, there is an inherent lag in responding to instantaneous price spikes.