Forecasting directional movements of stock prices for intraday trading using LSTM and random forests
Forecasting directional movements of stock prices for intraday trading using LSTM and random forests
复制标题
DOI：
10.1016/j.frl.2021.102280
复制
发表时间：
2022-04-09
影响因子：
10.4
通讯作者：
Sahoo, Jajati Keshari
中科院分区：
文献类型：
--
作者：
Ghosh, Pushpendu;Neufeld, Ariel;Sahoo, Jajati Keshari
我们采用随机森林和 LSTM 网络（更准确地说 CuDNNLSTM）作为训练方法来分析它们在预测 1993 年 1 月至 2018 年 12 月间标准普尔 500 指数成分股样本外定向变动方面的有效性。我们引入了一种多特征设置，不仅包括相对于收盘价的回报，还包括相对于开盘价和日内回报的回报。作为交易策略，我们使用 Krauss 等人。 （2017）和 Fischer 和 Krauss（2018）作为基准。在每个交易日，我们买入概率最高的 10 只股票，并卖空概率最低的 10 只股票，从而在日内收益方面跑赢大盘，所有股票的货币权重相同。我们的实证结果表明，在考虑交易成本之前，多特征设置使用 LSTM 网络提供的日回报率为 0.64%，使用随机森林的每日回报率为 0.54%。因此，我们优于 Fischer 和 Krauss (2018) 以及 Krauss 等人中的单特征设置。 (2017) 仅包含相对于收盘价的日回报率，LSTM 和随机森林的相应日回报率分别为 0.41% 和 0.39%。
We employ both random forests and LSTM networks (more precisely CuDNNLSTM) as training methodologies to analyze their effectiveness in forecasting out-of-sample directional movements of constituent stocks of the S&P 500 from January 1993 fill December 2018 for intraday trading. We introduce a mull-feature setting consisting not only of the returns with respect to the closing prices, but also with respect to the opening prices and intraday returns. As trading strategy, we use Krauss et al. (2017) and Fischer and Krauss (2018) as benchmark. On each trading day, we buy the 10 stocks with the highest probability and sell short the 10 stocks with the lowest probability to outperform the market in terms of intraday returns - all with equal monetary weight. Our empirical results show that the mull-feature setting provides a daily return, prior to transaction costs, of 0.64% using LSTM networks, and 0.54% using random forests. Hence we outperform the single-feature setting in Fischer and Krauss (2018) and Krauss et al. (2017) consisting only of the daily returns with respect to the closing prices, having corresponding daily returns of 0.41% and of 0.39% with respect to LSTM and random forests, respectively.