Locally Linear Embedding for High-Frequency Trading Marker Discovery

High-frequency trading (HFT) has been challenging fintech and data science. HFT trading marker prediction is a rarely investigated but important problem in finance and data science. In this study, we first propose locally linear embedding based HFT marker prediction algorithm to tackle this problem and HFT trading marker evaluation algorithms for validation. Our results demonstrate locally linear embedding (LLE) outperform its peers in capturing trading markers in terms of accuracy and complexity for its local data structure keeping mechanism in embedding.