Mining trading patterns of pyramid schemes from financial time series data

The current studies relating to pyramid schemes are mostly about qualitative analysis, whereas the quantitative analysis is still rare owing to the insufficiency in knowledge of their specific trading modes. Often, the trading modes of pyramid schemes are inconspicuous in financial data, making it difficult to be identified in the data. In this study, we propose a quantitative framework for mining trading patterns of pyramid schemes from financial time series data. The framework includes two parts: Long Range Sequence De-noising (LoRSD) algorithm and Contrast Trading Pattern Mining (Contrast TPM) algorithm. LoRSD distinguishes noise items by folding the statistical frequent items and removes the infrequent items recursively. In Contrast TPM, we first identify the frequent oneitemset by comparing the pyramid-related samples with the general samples. Subsequently, a random model is added in the comparative analysis to generate the frequency conditions for mining pyramid scheme patterns. Instead of setting user-defined support thresholds, we adopt contrastive samples as benchmarks in determining the frequency conditions. Our extensive experiments on the financial data set including behaviour of a real-world pyramid scheme demonstrate the effectiveness of our framework in sequence de-noising and mining trading patterns of pyramid schemes from financial time series data.(C) 2022 Elsevier B.V. All rights reserved.

Automated summary

This study proposes a quantitative framework for mining trading patterns of pyramid schemes from financial time series data. The framework includes the LoRSD algorithm for sequence de-noising and the Contrast TPM algorithm for mining patterns. The effectiveness of the framework is demonstrated through extensive experiments on financial data.