Machine Learning

Model Performance & Analytics

Connected
Last Update: --:--:--
UTC
LIVE

Model Performance Overview

Model Performance Metrics

Meta-Learner Model Weights

XGBoost28.0%
Transformer35.0%
AdaBoost15.0%
LSTM12.0%
Random Forest10.0%

Real-time Prediction Confidence

Training Status

XGBoostTraining
75%ETA: 2h 15m
TransformerComplete
100%ETA: Done
LSTMQueued
0%ETA: 4h 30m

Feature Importance

Price Momentum89%
Volume Profile76%
Market Sentiment64%
Technical Indicators58%
News Sentiment42%

Model Health

Data Quality
94%
Prediction Drift
12.0%
Model Stability
87%
Latency
78%

Transformer Architecture Deep Dive

Architecture Specs

Model Dimension512
Attention Heads8
Encoder Layers6
Sequence Length1440
Dropout Rate0.1

Training Progress

Training Loss0.0847
Validation Loss0.0923
Learning Rate3e-4

Attention Analysis

1-min patterns34%
5-min momentum28%
15-min trends22%
Hourly cycles16%
© Paul Archer, Imperial College London