fund_apply_redeem dataset

AFAC2023-金融场景理解-赛题1 金融营销场景中的时序预测 数据集
蚂蚁财富面向海量用户，每天都存在大量的基金申购和赎回行为。作为平台，既需要联合机构为基金设置短期库存额度，避免短期内单只基金进量过大，从而增加基金经理投资难度、带来投资风险，影响用户投资产品收益率和稳定性，也需要保障基金库存供给，最大化满足市场需求，因此，基金的申赎预测及库存运筹规划对风控管理和基金销售都具有重要意义。
与普通的时序预测任务不同的是，基金申赎预测不仅需要考虑基金收益表现等产品特征变化，更需要考虑到金融营销场景下的特殊属性，财富场景碎片化，用户行为周期长且受到金融行情的波动影响，产品间存在着挤压和替代等复杂关系。

AFAC2023 - Financial Scenario Understanding - Track 1: Time Series Forecasting in Financial Marketing Scenarios Dataset
Ant Fortune serves a vast user base, with a large number of fund subscription and redemption transactions occurring every day. As a platform, it needs to collaborate with financial institutions to set short-term inventory limits for funds, so as to prevent excessive inflows of a single fund in a short period, which would increase the investment difficulty for fund managers, introduce investment risks, and undermine the return rate and stability of users' investment products. Meanwhile, it also needs to guarantee the supply of fund inventories to maximize the satisfaction of market demand. Therefore, fund subscription and redemption forecasting and inventory operation planning are of great significance to both risk management and fund sales.
Unlike conventional time series forecasting tasks, fund subscription and redemption forecasting not only needs to consider changes in product characteristics such as fund return performance, but also needs to account for the special attributes of financial marketing scenarios. The wealth management scenario is fragmented, user behavior cycles are lengthy and affected by fluctuations in financial market trends, and there exist complex relationships such as competition and substitution among different fund products.

搜集汇总

背景与挑战

背景概述

该数据集是AFAC2023金融场景理解竞赛的赛题1数据集，专注于金融营销场景中的时序预测任务，旨在预测基金的申购和赎回金额。数据集包含基金申赎数据、行情信息和时间信息，特点在于结合了基金特征、用户行为和市场行情，并考虑了节假日导致的时序异步性，适用于开发复杂的时序预测模型以支持风控和销售决策。