---
title: M3FT: Micro-Macro-Micro Reasoning for Multi-Agent Cryptocurrency Perpetual Futures Trading
url: https://www.ml-quant.com/papers/arxiv/2611.10586/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-11-14
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: arXiv:2611.10586
source_url: https://arxiv.org/abs/2611.10586
featured: 2026-11-13
citations: unknown
topic: Crypto & DeFi
---


# M3FT: Micro-Macro-Micro Reasoning for Multi-Agent Cryptocurrency Perpetual Futures Trading

A micro-macro-micro framework combining numerical states with linguistic market interpretations and cross-asset communication improves coordinated trading decisions in perpetual futures.

- Source: https://arxiv.org/abs/2611.10586
- Identifier: arXiv:2611.10586
- Released: 2026-11-10
- First featured: Quant Letter No. 139 (2026-11-13): https://www.ml-quant.com/issues/2026-11-13/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: Crypto & DeFi
- Authors: Jiashun Wang, Beile Wang, Guangyan Gan, Wentao Zhang, Siguang Li, Shuo Sun, Yunbo Lu, Lin William Cong

## Abstract (arXiv, CC0)

Cryptocurrency perpetual futures form a large-scale online financial market in which automated agents manage interacting asset positions under changing conditions. Policies relying solely on micro-level numerical states may overlook cross-asset relationships and become vulnerable to regime shifts, while macro-level linguistic judgments alone cannot ensure precise measurement or optimize positions under trading costs and risk constraints. Connecting market interpretation to coordinated multi-asset decisions remains challenging. We propose M3FT, a micro-macro-micro framework for multi-agent perpetual futures trading. The first micro stage constructs time-aligned numerical states from exchange and account observations through deterministic processing. The macro stage transforms these measured states into explicit market interpretations without adding external market information. Four asset-specific teams, each comprising a Market Interpreter, an Asset Communicator, and a Risk Reviewer, refine these interpretations through experience retrieval, cross-asset communication, and review of joint-position risks. The final micro stage combines encoded linguistic representations with the original numerical states to learn continuous target positions. We implement this framework with a language-conditioned multi-agent policy: asset actors use global and private numerical states together with linguistic context, while centralized critics evaluate joint actions using portfolio-wide information and shared-account rewards. Experiments on real cryptocurrency perpetual futures data, including comparisons across language models and policy-learning algorithms and component ablations, show that the best-performing M3FT configuration achieves a higher mean test return than the evaluated direct LLM, numerical RL, and rule-based baselines.

## Related

- [Factors Influencing Cryptocurrency Prices: Evidence from Bitcoin, Ethereum, Dash, Litcoin, and Monero](https://www.ml-quant.com/papers/arxiv/2511.22782/): The study examines the price factors influencing five major cryptocurrencies from 2010-2018, highlighting the roles of market conditions, long-term appeal, and the SP500 index.
- [The Impact of Bitcoin ETF Approval on Bitcoin's Hedging Properties Against Traditional Assets](https://www.ml-quant.com/papers/arxiv/2512.12815/): Changing Dynamics with Traditional Assets: In January 2024, the approval of a Bitcoin Spot ETF boosted interest from big investors and made Bitcoin more closely linked to stock markets, while its connections to gold remained steady and its relationship with the U.S. Dollar stayed negative.
- [Improving Cryptocurrency Pump-and-Dump Detection through Ensemble-Based Models and Synthetic Oversampling Techniques](https://www.ml-quant.com/papers/arxiv/2510.00836/): The study uses SMOTE and advanced ensemble learning models like XGBoost and LightGBM to effectively detect pump and dump manipulation in cryptocurrency markets, enhancing market transparency and stability.
- [How Exclusive are Ethereum Transactions? Evidence from non-winning blocks](https://www.ml-quant.com/papers/arxiv/2509.16052/): Dominant Source of Builder Revenues: The study shows that exclusive transactions, found only in blocks from one builder, make up 84% of total fees paid by transactions in winning blocks on Ethereum's blockchain. This makes them the main source of builder revenues.
- [Banking 2.0: The Stablecoin Banking Revolution - How Digital Assets Are Reshaping Global Finance](https://www.ml-quant.com/papers/arxiv/2508.11395/): Stablecoins, combining cryptocurrency and traditional finance, are poised to transform the global financial system by addressing flaws in current fiat currencies and promoting a more globally connected financial system.
- [Universal Patterns in the Blockchain: Analysis of EOAs and Smart Contracts in ERC20 Token Networks](https://www.ml-quant.com/papers/arxiv/2508.04671/): A study of 44 million Ethereum transactions reveals distinct patterns, showing differences between human and automated transaction behaviors in blockchain ecosystems.