Paper: arXiv 2507.20468
Authors: Antonino Castelli, Paolo Giudici, Alessandro Piergallini
Abstract
The rapid growth of crypto markets has opened new opportunities for investors, but at the same time exposed them to high volatility. To address the challenge of managing dynamic portfolios in such an environment, this paper presents a practical application of a multi-agent system designed to autonomously construct and evaluate crypto-asset allocations. Using data on daily frequencies of the ten most capitalized cryptocurrencies from 2020 to 2025, we compare two automated investment strategies. These are a static equal weighting strategy and a rolling-window optimization strategy, both implemented to maximize the evaluation metrics of the Modern Portfolio Theory (MPT), such as Expected Return, Sharpe and Sortino ratios, while minimizing volatility. Each step of the process is handled by dedicated agents, integrated through a collaborative architecture in Crew AI. The results show that the dynamic optimization strategy achieves significantly better performance in terms of risk-adjusted returns, both in-sample and out-of-sample. This highlights the benefits of adaptive techniques in portfolio management, particularly in volatile markets such as cryptocurrency markets. The following methodology proposed also demonstrates how multi-agent systems can provide scalable, auditable, and flexible solutions in financial automation.
Complexity vs Empirical Score
- Math Complexity: 6.0/10
- Empirical Rigor: 6.0/10 (capped from 7.5: transaction costs not modelled; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 7.5)
Why this score: The paper employs advanced mathematics like Modern Portfolio Theory (MPT), mean-variance optimization, and rolling Sharpe maximization, which are conceptually dense. It demonstrates high empirical rigor by using real 5-year daily data on 10 assets, conducting in-sample and out-of-sample backtests, and implementing a full multi-agent system for automation.
Research Flowchart
flowchart TD A["Research Goal<br>Optimize crypto portfolios in volatile markets"] --> B["Data Input<br>Top 10 crypto assets (2020-2025)"] B --> C["Methodology<br>Multi-Agent System (Crew AI)"] C --> D["Strategy 1<br>Static Equal Weighting"] C --> E["Strategy 2<br>Rolling-Window Optimization"] D --> F["Evaluation & Comparison<br>Modern Portfolio Theory MPT<br>Sharpe & Sortino Ratios"] E --> F F --> G["Key Outcomes<br>Dynamic Optimization wins<br>Risk-adjusted returns improved<br>Scalable & Auditable solution"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a multi-agent system to autonomously construct and evaluate crypto-asset allocation strategies, comparing a static equal-weighted approach with a rolling-window optimization.
Source sentence
this paper presents a practical application of a multi-agent system designed to autonomously construct and evaluate crypto-asset allocations.
Assumes 4
- No short-selling and full investment constraint is imposed on the optimization. model
Source sentence
under the constraint of no short-selling and full investment.
- Static models assume constant risk and correlation over time. market
Source sentence
challenge the assumptions of constant risk and correlation used in static models.
- Transaction costs such as spreads and fees are ignored in the performance results. execution
Source sentence
Transaction costs: Spreads and fees are ignored and could erode part of Crew B’s edge.
- Trading can be executed at daily closing prices without slippage. execution
Source sentence
Our study focuses on a panel of daily closing prices for the ten largest crypto assets
Data
2020-08 – 2025-03 · top‑10 cryptocurrencies · crypto · daily
- Daily closing prices of the ten largest cryptocurrencies market-prices unknown
Source sentence
The dataset is composed of 1667 daily closing prices for the ten largest cryptocurrencies, considering market capitalization, over the period from 2020-08-20 to 2025-03-13.
Costs
Not modeled
Source sentence
Transaction costs: Spreads and fees are ignored and could erode part of Crew B’s edge.
Finds 5
Sharpe 1.00 (benchmark 0.53) · annual return 9.9% · max drawdown 15.0%
- In‑sample, the rolling optimization (Crew B) achieves a Sharpe ratio of 1.00, compared to 0.60 for the static optimized portfolio (Crew A). performance
Source sentence
Crew B achieves a Sharpe ratio of 1.00, with reduced volatility (10.2% against 12.4%) and improved expected return (9.9% vs 7.5%).
- Out‑of‑sample, Crew B maintains a Sharpe ratio of 0.72 while Crew A’s Sharpe drops to 0.36. performance
Source sentence
Here, Crew A’s Sharpe ratio drops to 0.36, while Crew B maintains a strong 0.72, highlighting better generalization.
- Crew B reduces in‑sample volatility to 10.2% versus 12.4% for Crew A. performance
Source sentence
Crew B achieves a Sharpe ratio of 1.00, with reduced volatility (10.2% against 12.4%)
- Dynamic rebalancing consistently dominates the static approach in the risk‑return space, delivering higher returns and lower volatility. performance
Source sentence
Crew B consistently dominates Crew A in the risk-return space, both in terms of higher return and lower volatility.
- The authors recommend the rolling optimized weights strategy for future portfolio allocation. other
Source sentence
Based on the above analysis, I recommend the rolling optimized weights strategy for future portfolio allocation.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 58 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.