Mid-LSTM for Anomaly-Aware Midterm Stock Prediction

Summary

The paper proposes Mid-LSTM, a deep learning model intended to predict stock prices over a midterm horizon while accounting for market anomalies. Its design represents market trends as hidden states. The described approach first formulates a midterm ARMA model that incorporates hidden states and CAPM, then combines an LSTM, a hidden Markov model, and linear regression networks. The stated goal is to limit accumulated prediction errors and explain factors associated with stock prices.

Experiments on S&P 500 stocks are reported to show a 2–4% improvement in prediction accuracy. The authors also report portfolio allocation results of up to 120.16% annual return and an average Sharpe ratio of 2.99. These are claims from the paper summary; the excerpt does not specify the evaluation period, baselines, transaction costs, validation design, or risk controls. The results therefore cannot establish out-of-sample performance or expected live returns on their own.

Key ideas

• Mid-LSTM represents market trends as hidden states for midterm stock prediction.
• The model combines LSTM, hidden Markov, and linear regression components.
• Its formulation also draws on ARMA and CAPM.
• The reported S&P 500 experiments show improved prediction accuracy and portfolio metrics, but the excerpt omits evaluation details.

Tags

# Risk Management via Anomaly Circumvent: Mnemonic Deep Learning for Midterm Stock Prediction


# Risk Management via Anomaly Circumvent: Mnemonic Deep Learning for Midterm Stock Prediction









Midterm stock price prediction is crucial for value investments in the stock market. However, most deep learning models are essentially short-term and applying them to midterm predictions encounters large cumulative errors because they cannot avoid anomalies. In this paper, we propose a novel deep neural network Mid-LSTM for midterm stock prediction, which incorporates the market trend as hidden states. First, based on the autoregressive moving average model (ARMA), a midterm ARMA is formulated by taking into consideration both hidden states and the capital asset pricing model. Then, a midterm LSTM-based deep neural network is designed, which consists of three components: LSTM, hidden Markov model and linear regression networks. The proposed Mid-LSTM can avoid anomalies to reduce large prediction errors, and has good explanatory effects on the factors affecting stock prices. Extensive experiments on S&P 500 stocks show that (i) the proposed Mid-LSTM achieves 2-4% improvement in prediction accuracy, and (ii) in portfolio allocation investment, we achieve up to 120.16% annual return and 2.99 average Sharpe ratio.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.