Peking University Mini Workshop on Financial Mathematics and Engineering
时间：2025 年 5 月 29 日 14:00 – 17:00
地点：智华楼 丁石孙教室
报告人：徐靖（中国人民大学），张霜剑（复旦大学）
部慧（北京航空航天大学），宋颖达（上海交通大学）
腾讯会议：735-761-541
会议链接：//meeting.tencent.com/dm/jxxmZvBhk1DB
■ 会议日程
|
时间
|
内容
|
报告人
|
单位
|
|
14:00–14:40
|
Sharpe Ratio Timing with Stop Loss
|
徐 靖
|
中国人民大学
|
|
14:40–15:20
|
A free boundary approach to the monopolist problem
|
张霜剑
|
复旦大学
|
|
15:20–15:30
|
休息
|
—
|
丁石孙教室外
|
|
15:30–16:10
|
When Size Fails: The Emergence of Information Leadership in Global Commodity Markets
|
部 慧
|
北京航空航天大学
|
|
16:10–16:50
|
Fast Simulation by Euler Approximation: From Diffusion to Queueing Network
|
宋颖达
|
上海交通大学
|
注：每场学术报告时长约 30 分钟，含提问与讨论环节。
报告人简介：
徐靖（中国人民大学）
中国人民大学财政金融学院副教授，应用金融系副主任。主要研究方向是具有摩擦因素的金融市场中的资产配置、期权定价、信息获取和风险管理问题。研究成果发表于Management Science, Operations Research, JFQA等十多种国际学术期刊。担任China Finance Review International的青年编委。
张霜剑（复旦大学）
复旦大学麻豆视频
青年研究员，博士生导师。2018年博士毕业于多伦多大学数学系；博士毕业后曾在巴黎高科国立统计与经济管理学院、巴黎高等师范学院、滑铁卢大学从事博士后研究。2023年9月至今任职于复旦大学麻豆视频
，研究方向为最优输运理论在经济金融中的应用，在垄断定价等领域取得重要研究成果。相关研究成果发表在Communications on Pure and Applied Mathematics、Mathematical Models and Methods in Applied Sciences、Advances in Mathematics、Economic Theory、Journal of Mathematical Economics、Journal of Convex Analysis、Conference on Learning Theory等应用数学、经济以及机器学习领域的国际权威期刊和会议上。曾入选国家级高层次青年人才计划。
部慧（北京航空航天大学）
北京航空航天大学经济管理学院金融系主任，副教授、博士生导师。研究领域为金融科技和监管科技、行为金融、实证资产定价、风险管理。主持国家自然科学基金重点项目1项，主持国家重点研发计划重点专项课题1项，主持面上和青年项目4项，主持多项部委和企业委托项目，作为副组长参与国家242课题和信息安全项目2项。已发表学术论文40余篇，出版教材1部、专著2部，已获技术发明专利授权6项，获得学术会议优秀论文奖12项；3篇论文获得“中国知网2024年年度高被引论文、高下载论文，前1%论文”。在金融监管科技领域取得了一系列创新成果并产生重要社会影响。牵头自主研发了5套金融监管系统服务金融监管和监察机构；参与债券风险监测预警产品，服务众多金融机构；参与多个业界金融产品研发。近五年报送6篇政策研究报告，1篇被中办采纳、3篇被司局级单位采纳，支撑了系列金融监管政策出台、对风险事件的定调和应对策略。担任3个学会专委会理事，1个教育委员会分会理事；担任多个学术会议程序委员会成员或组委会成员；担任多个项目评审专家。
宋颖达（上海交通大学）
上海交通大学安泰经济与管理学院副教授、博士生导师。于香港科技大学工业工程及物流管理系获博士学位，清华大学数学科学系获学士学位。主要研究方向包括金融工程、风险管理、随机模型和仿真计算等，主持和参与多项国家自然科学基金项目，研究成果发表在Operations Research、Mathematical Finance、INFORMS Journal on Computing等国际期刊上。目前担任《系统管理学报》编辑部副主任，中国“双法”研究会和中国运筹学会二级分会副秘书长、理事等。
报告摘要：
徐靖（中国人民大学）
We propose a new approach to market timing by targeting the Sharpe ratio, which nests volatility timing as a special case when the expected market return is constant. We further develop a theoretical model and introduce a stop-loss strategy, a novel addition to market timing, to enhance performance. Empirically, we find that Sharpe ratio timing substantially outperforms volatility timing, and incorporating the stop-loss strategy improves the results even further. Our study highlights the joint importance of the risk-return tradeoff and risk control in profitable trading, captured by the Sharpe ratio and stop-loss strategy, respectively.■
张霜剑（复旦大学）
The monopolist problem is a class of variational problems with convexity constraints. It has been shown to possess deep connections with several important areas of modern analysis, including optimal transport and free boundary problems. Although substantial progress has been made in the study of existence, uniqueness, and regularity of optimal solutions, their analytical structure in multidimensional settings remains far from fully understood. Rochet and Choné (Econometrica, 1998) reformulated this problem to a concave maximization over the set of convex functions. We characterize solutions to this problem, giving the first analytical description of an overlooked market segment, where the regularity built by Caffarelli-Lions plays a crucial role —— an extension of their regularity work to the quasilinear case is also recently studied. We use techniques from the study of the Monge–Ampére equation, the obstacle problem, and localization for measures in convex-order. This is a joint work with Robert J. McCann and Cale Rankin.
部慧（北京航空航天大学）
In complex economic networks, physical scale is traditionally assumed to dictate structural influence. Global commodity figures markets, however, present a striking paradox: despite accounting for the world’s largest trading volumes in several key commodities, Chinese futures markets predominantly remain price takers rather than price drivers. To decode this anomaly, we construct weighted, directed causal networks of price information diffusion. We reveal a persistent “size-power decoupling,” demonstrating that market scale fails to translate into topological leadership. By mapping network influence into a dynamic phase space, we find that global pricing benchmarks do not evolve linearly. Instead, information leadership follows systemic dynamical laws, undergoing topological phase transitions only when triggered by exogenous macro-shocks. Furthermore, we prove that trading volume is insufficient for pricing dominance; a market’s true influence relies on its capacity to efficiently encode policy uncertainties, market risks, and geopolitical conflicts. Integrating econometrics and complex systems science, this study uncovers the hidden mechanisms behind the emergence of pricing leadership, demonstrating that global influence is driven by the quality of signals a market broadcasts, rather than its sheer physical size.
宋颖达（上海交通大学）
The efficient management of large-scale queueing networks is critical for sectors such as healthcare, logistics, and customer service, where system performance directly impacts operational effectiveness and cost. To address this challenge, we introduce simulation techniques for complex, large-scale Markovian queueing networks, drawing direct inspiration from the Euler scheme widely used for simulating diffusion processes in financial modeling. Specifically, we develop two simulation schemes based on Euler-type approximation—termed the backward and forward schemes—which naturally accommodate time-varying dynamics and are optimized for vectorized implementation. Assuming a feedforward network structure, we prove that the two schemes yield stochastic upper and lower bounds on the system state, with approximation error remaining bounded over the simulation horizon. Under a recommended time step, we show that our schemes exhibit asymptotically diminishing relative error as system scale increases, while achieving dramatically lower computational complexity than traditional discrete-event simulation—offering speedups of up to tens of thousands of times. This work highlights the potential of adapting financial engineering’s workhorse Euler method to the simulation of large-scale discrete systems.
