λ-泛竞争差,竞争差," /> λ-泛竞争差,竞争差,"/> 未知价格变动信息下在线单向交易的数据驱动型策略设计
主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2026, Vol. 34 ›› Issue (10): 128-139.doi: 10.16381/j.cnki.issn1003-207x.2024.2170

• • 上一篇    

未知价格变动信息下在线单向交易的数据驱动型策略设计

张文明(), 舒娜   

  1. 西北大学经济管理学院,陕西 西安 710127
  • 收稿日期:2024-12-06 修回日期:2025-06-13 出版日期:2026-10-25 发布日期:2026-10-09
  • 通讯作者: 张文明 E-mail:wenming@nwu.edu.cn
  • 基金资助:
    陕西省自然科学基础研究计划项目(2026JC-YBMS-0845);国家自然科学基金项目(72271198)

Data-driven Strategy Designing for the Online One-way Trading Without Known Price Variation Information

Wenming Zhang(), Na Shu   

  1. School of Economics and Management,Northwest University,Xi'an 710127,China
  • Received:2024-12-06 Revised:2025-06-13 Online:2026-10-25 Published:2026-10-09
  • Contact: Wenming Zhang E-mail:wenming@nwu.edu.cn

摘要:

在线单向交易问题中,决策者面对价格序列未知且需做出不可逆的决策。现有竞争分析策略依赖预知价格变动边界[m,M]的强假设,实际应用却难以满足。为消除这一限制,本研究针对未知价格变动下的在线单向交易问题,提出一种免预知价格边界的数据驱动型策略。首先,在竞争差准则基础上提出了一种新的竞争性策略评价指标“λ-泛竞争差”(其中,λ为风险偏好参数);然后,在“随意”给定m、M情形下设计了一个最优在线策略OSUCD;进一步地,把OSUCD升级为可以通过历史数据来选取合适风险偏好λ¯(m,M)的数据驱动型在线策略D-OSUCD。仿真实验表明,该策略对参数m和M在合理范围内的设置具有强鲁棒性。最后,对碳排放交易所中HBEA、BEA和GDEA三种交易产品进行实例分析,验证了策略D-OSUCD的有效性与稳健性。

关键词: 在线单向交易, 数据驱动型策略, 竞争分析, λ-泛竞争差')">λ-泛竞争差, 竞争差

Abstract:

In practical applications, the problem of online one-way trading is widely encountered in areas, including foreign exchange, inventory procurement, advertising placement, product sale, leasing issues, etc. In the online one-way trading problem, 1 unit of some asset is to be sold in n periods by DM (decision makers) to maximize their returns. At period i, the DM must instantaneously determine whether to sell and the quantity to sell, solely relying on current price information pi without any foresight into future prices, pi+1, pi+2,…,pn . Traditional research typically assumes that DM knows the upper and lower bounds or their ratio of price to formulate effective strategy.In order to abandon the assumption that the upper and lower bounds of price changes need to be predicted in advance in the previous research, a new competitive strategy evaluation conception “λ- universal competitive difference” is proposed based on the competitive difference conception (where λ is the risk preference). From the seller's perspective, the optimal strategy for DM is delved into to successfully sell one unit of an asset in at most k transactions over n time periods, with the objective of maximizing returns. Then an optimal online strategy OSUCD is devised with the upper and lower bounds of price changes given by m and M. Then, it is elucidated that under specific case when λ=12, the λ-universal competitive difference degrades into the competitive difference proposed by Wang et al. (2016),the effectiveness of the strategy OSUCD was indirectly proved thereby. Furthermore, OSUCD is upgraded to a data-driven online strategy D-OSUCD that can select appropriate risk preferences λ¯(m,M) based on historical data. Simulation data corroborates the robustness of D-OSUCD's parameter settings. Notably, the DM does not need to anticipate the assumptions of the upper M and lower bounds m of the future price fluctuations in advance, and can instead flexibly set parameters within a reasonable range.Additionally, the data-driven online strategy D-OSUCD is applied, grounded in the λ-universal competitive difference criterion, to the carbon emission trading products of the China Hubei Carbon Emission Exchange (HBEA), China Beijing Green Exchange (BEA), and CEEX China Emissions Exchange (GDEA). These applications further validate the efficacy and robustness of the strategy D-OSUCD. The data-driven online strategy D-OSUCD in this paper can be used to analyze not only online one-way trading but also other online decision-making problems.

Key words: online one-way trading, data-driven strategy, competitive analysis, λ-universal competitive difference, competitive difference

中图分类号: