Abstract
Portfolio management is essential for any investment decision. Yet, traditional methods in the literature are ill-suited for the characteristics and dynamics of cryptocurrencies. This work presents a method to build an investment portfolio consisting of more than 1500 cryptocurrencies covering 6 years of market data. It is centred around Topological Data Analysis (TDA), a recent approach to analyze data sets from the perspective of their topological structure. This publication proposes a system combining persistence landscapes to identify suitable investment opportunities in cryptocurrencies. Using a novel and comprehensive data set of cryptocurrency prices, this research shows that the proposed system enables analysts to outperform a classic method from the literature without requiring any feature engineering or domain knowledge in TDA. This work thus introduces TDA-based portfolio management of cryptocurrencies as a viable tool for the practitioner.
Introduction
Managing a cryptocurrency portfolio presents unique challenges that traditional financial models often fail to address. Unlike established asset classes, cryptocurrencies lack tangible underlying assets or centralized oversight, leading to extreme volatility and a high noise-to-signal ratio. Traditional optimization techniques, such as the Markowitz model, are highly sensitive to estimation errors in this environment. Furthermore, most existing research is limited to a handful of major assets over short timeframes. This paper introduces a robust, automated selection method capable of handling thousands of assets by analyzing the "shape" of price movements through Topological Data Analysis (TDA).
Methodology: The TDA-RFM Pipeline
The core of this research is a multi-stage pipeline that transforms raw price data into a topological "score" for each asset. By treating price movements as geometric shapes rather than just statistical distributions, the system can identify structural patterns that traditional metrics miss.
1. Topological Feature Extraction
The process begins with Time Delay Embedding (TDE), which converts a single-dimensional price time series into a high-dimensional point cloud. From this cloud, the researchers construct a Vietoris-Rips Complex to identify 1-dimensional homology ($H_1$)—essentially looking for "loops" or cycles in the data that represent recurring patterns or structural shifts. These features are then encoded into Persistence Landscapes, and the $L^2$-norm of these landscapes is calculated to quantify the strength of the topological signal.
2. The RFM Adaptation
In a novel cross-domain application, the authors adapt the RFM (Recency, Frequency, Monetary) framework from marketing to financial signals. In this context, a positive topological signal is treated as a "purchase event":
- Recency: Days since the topological signal was last positive.
- Frequency: How often the signal was positive within a specific window.
- Monetary: The cumulative value of those positive signals.
These metrics are normalized and summed to create a final score, which determines the asset's allocation within the portfolio.
Dataset and Market Scope
The study utilizes one of the most comprehensive datasets in TDA-financial research, covering the market's evolution from its early stages through the 2017 bull run and the subsequent "crypto winter."
- Scope: Daily market data for 1,561 cryptocurrencies.
- Duration: 6 years, from 2013 to July 2019.
Key Results and Insights
The TDA-based portfolio was benchmarked against a Naive Allocation (1/N) strategy. The results demonstrate that capturing the "shape" of the market provides a significant advantage over raw statistical returns:
- Resilience in Bear Markets: The TDA portfolio showed remarkable stability during market downturns. For instance, in January 2019, while the Naive portfolio lost 10.93%, the TDA portfolio achieved a gain of 0.05%.
- Robustness to Noise: By focusing on persistent structural features, the model is inherently less sensitive to the small price fluctuations (noise) that often plague cryptocurrency data.
- Practical Viability: The system outperformed classic methods without requiring manual feature engineering or deep domain expertise in TDA, making it an accessible tool for financial practitioners.
Conclusion
This research demonstrates that Topological Data Analysis, when combined with an adapted RFM framework, offers a powerful alternative to traditional portfolio management in highly volatile markets. While computational costs and parameter sensitivity remain areas for future optimization, the ability to identify valuable price trends through their topological signatures marks a significant advancement in automated investment strategies.
Summary updated