当前位置：首页 > news >正文

qlib Alpha360 因子列表解读

news 2025/7/26 21:09:15

Alpha360因子列表:
['CLOSE59', 'CLOSE58', 'CLOSE57', 'CLOSE56', 'CLOSE55', 'CLOSE54', 'CLOSE53', 'CLOSE52', 'CLOSE51', 'CLOSE50', 'CLOSE49', 'CLOSE48', 'CLOSE47', 'CLOSE46', 'CLOSE45', 'CLOSE44', 'CLOSE43', 'CLOSE42', 'CLOSE41', 'CLOSE40', 'CLOSE39', 'CLOSE38', 'CLOSE37', 'CLOSE36', 'CLOSE35', 'CLOSE34', 'CLOSE33', 'CLOSE32', 'CLOSE31', 'CLOSE30', 'CLOSE29', 'CLOSE28', 'CLOSE27', 'CLOSE26', 'CLOSE25', 'CLOSE24', 'CLOSE23', 'CLOSE22', 'CLOSE21', 'CLOSE20', 'CLOSE19', 'CLOSE18', 'CLOSE17', 'CLOSE16', 'CLOSE15', 'CLOSE14', 'CLOSE13', 'CLOSE12', 'CLOSE11', 'CLOSE10', 'CLOSE9', 'CLOSE8', 'CLOSE7', 'CLOSE6', 'CLOSE5', 'CLOSE4', 'CLOSE3', 'CLOSE2', 'CLOSE1', 'CLOSE0', 'OPEN59', 'OPEN58', 'OPEN57', 'OPEN56', 'OPEN55', 'OPEN54', 'OPEN53', 'OPEN52', 'OPEN51', 'OPEN50', 'OPEN49', 'OPEN48', 'OPEN47', 'OPEN46', 'OPEN45', 'OPEN44', 'OPEN43', 'OPEN42', 'OPEN41', 'OPEN40', 'OPEN39', 'OPEN38', 'OPEN37', 'OPEN36', 'OPEN35', 'OPEN34', 'OPEN33', 'OPEN32', 'OPEN31', 'OPEN30', 'OPEN29', 'OPEN28', 'OPEN27', 'OPEN26', 'OPEN25', 'OPEN24', 'OPEN23', 'OPEN22', 'OPEN21', 'OPEN20', 'OPEN19', 'OPEN18', 'OPEN17', 'OPEN16', 'OPEN15', 'OPEN14', 'OPEN13', 'OPEN12', 'OPEN11', 'OPEN10', 'OPEN9', 'OPEN8', 'OPEN7', 'OPEN6', 'OPEN5', 'OPEN4', 'OPEN3', 'OPEN2', 'OPEN1', 'OPEN0', 'HIGH59', 'HIGH58', 'HIGH57', 'HIGH56', 'HIGH55', 'HIGH54', 'HIGH53', 'HIGH52', 'HIGH51', 'HIGH50', 'HIGH49', 'HIGH48', 'HIGH47', 'HIGH46', 'HIGH45', 'HIGH44', 'HIGH43', 'HIGH42', 'HIGH41', 'HIGH40', 'HIGH39', 'HIGH38', 'HIGH37', 'HIGH36', 'HIGH35', 'HIGH34', 'HIGH33', 'HIGH32', 'HIGH31', 'HIGH30', 'HIGH29', 'HIGH28', 'HIGH27', 'HIGH26', 'HIGH25', 'HIGH24', 'HIGH23', 'HIGH22', 'HIGH21', 'HIGH20', 'HIGH19', 'HIGH18', 'HIGH17', 'HIGH16', 'HIGH15', 'HIGH14', 'HIGH13', 'HIGH12', 'HIGH11', 'HIGH10', 'HIGH9', 'HIGH8', 'HIGH7', 'HIGH6', 'HIGH5', 'HIGH4', 'HIGH3', 'HIGH2', 'HIGH1', 'HIGH0', 'LOW59', 'LOW58', 'LOW57', 'LOW56', 'LOW55', 'LOW54', 'LOW53', 'LOW52', 'LOW51', 'LOW50', 'LOW49', 'LOW48', 'LOW47', 'LOW46', 'LOW45', 'LOW44', 'LOW43', 'LOW42', 'LOW41', 'LOW40', 'LOW39', 'LOW38', 'LOW37', 'LOW36', 'LOW35', 'LOW34', 'LOW33', 'LOW32', 'LOW31', 'LOW30', 'LOW29', 'LOW28', 'LOW27', 'LOW26', 'LOW25', 'LOW24', 'LOW23', 'LOW22', 'LOW21', 'LOW20', 'LOW19', 'LOW18', 'LOW17', 'LOW16', 'LOW15', 'LOW14', 'LOW13', 'LOW12', 'LOW11', 'LOW10', 'LOW9', 'LOW8', 'LOW7', 'LOW6', 'LOW5', 'LOW4', 'LOW3', 'LOW2', 'LOW1', 'LOW0', 'VWAP59', 'VWAP58', 'VWAP57', 'VWAP56', 'VWAP55', 'VWAP54', 'VWAP53', 'VWAP52', 'VWAP51', 'VWAP50', 'VWAP49', 'VWAP48', 'VWAP47', 'VWAP46', 'VWAP45', 'VWAP44', 'VWAP43', 'VWAP42', 'VWAP41', 'VWAP40', 'VWAP39', 'VWAP38', 'VWAP37', 'VWAP36', 'VWAP35', 'VWAP34', 'VWAP33', 'VWAP32', 'VWAP31', 'VWAP30', 'VWAP29', 'VWAP28', 'VWAP27', 'VWAP26', 'VWAP25', 'VWAP24', 'VWAP23', 'VWAP22', 'VWAP21', 'VWAP20', 'VWAP19', 'VWAP18', 'VWAP17', 'VWAP16', 'VWAP15', 'VWAP14', 'VWAP13', 'VWAP12', 'VWAP11', 'VWAP10', 'VWAP9', 'VWAP8', 'VWAP7', 'VWAP6', 'VWAP5', 'VWAP4', 'VWAP3', 'VWAP2', 'VWAP1', 'VWAP0', 'VOLUME59', 'VOLUME58', 'VOLUME57', 'VOLUME56', 'VOLUME55', 'VOLUME54', 'VOLUME53', 'VOLUME52', 'VOLUME51', 'VOLUME50', 'VOLUME49', 'VOLUME48', 'VOLUME47', 'VOLUME46', 'VOLUME45', 'VOLUME44', 'VOLUME43', 'VOLUME42', 'VOLUME41', 'VOLUME40', 'VOLUME39', 'VOLUME38', 'VOLUME37', 'VOLUME36', 'VOLUME35', 'VOLUME34', 'VOLUME33', 'VOLUME32', 'VOLUME31', 'VOLUME30', 'VOLUME29', 'VOLUME28', 'VOLUME27', 'VOLUME26', 'VOLUME25', 'VOLUME24', 'VOLUME23', 'VOLUME22', 'VOLUME21', 'VOLUME20', 'VOLUME19', 'VOLUME18', 'VOLUME17', 'VOLUME16', 'VOLUME15', 'VOLUME14', 'VOLUME13', 'VOLUME12', 'VOLUME11', 'VOLUME10', 'VOLUME9', 'VOLUME8', 'VOLUME7', 'VOLUME6', 'VOLUME5', 'VOLUME4', 'VOLUME3', 'VOLUME2', 'VOLUME1', 'VOLUME0', 'LABEL0']
因子总数: 361

Alpha360 因子库包含 361 个因子（实际比名称中的360多1个），主要由以下几类历史数据组成：

1. 时间序列结构

所有因子都是基于过去60个交易日(约3个月)的历史数据构建，数字后缀表示天数：

59 = 59天前(约3个月前)
0 = 当天
中间数字表示连续的时间序列

2. 因子分类

(1) 价格类因子 (CLOSE/OPEN/HIGH/LOW/VWAP)

CLOSE0-CLOSE59: 过去60个交易日的收盘价序列
OPEN0-OPEN59: 过去60个交易日的开盘价序列
HIGH0-HIGH59: 过去60个交易日的最高价序列
LOW0-LOW59: 过去60个交易日的最低价序列
VWAP0-VWAP59: 过去60个交易日的成交量加权平均价序列

(2) 成交量类因子

VOLUME0-VOLUME59: 过去60个交易日的成交量序列

(3) 目标变量

LABEL0: 预测目标(通常是未来某段时间的收益率)

3. 特点分析

原始数据特征：与Alpha158不同，Alpha360主要提供原始价格和成交量数据，而不是计算好的技术指标
时间序列建模：
- 提供完整的60天历史序列，适合RNN、Transformer等时序模型
- 每个时间点包含5个价格维度(OPEN/HIGH/LOW/CLOSE/VWAP)+1个成交量维度
灵活性：
- 用户可以根据这些原始数据自行构造技术指标
- 适合需要自定义特征工程的场景
数据量：
- 每个股票每天有6×60=360个原始特征
- 加上LABEL0共361列