Literature highlighted that financial time series data pose significant challenges for accurate stock price prediction, because these data are characterized by noise and susceptibility to news; traditional statistical methodologies made assumptions, such as linearity and normality, which are not suitable for the non-linear nature of financial time series; on the other hand, machine learning methodologies are able to capture non linear relationship in the data. To date, neural network is considered the main machine learning tool for the financial prices prediction. Transfer Learning, as a method aimed at transferring knowledge from source tasks to target tasks, can represent a very useful methodological tool for getting better financial prediction capability. Current reviews on the above body of knowledge are mainly focused on neural network architectures, for financial prediction, with very little emphasis on the transfer learning methodology; thus, this paper is aimed at going deeper on this topic by developing a systematic review with respect to application of Transfer Learning for financial market predictions and to challenges/potential future directions of the transfer learning methodologies for stock market predictions.

Shivam has over 10 years of experience in the investment industry and in applying Artificial Intelligence techniques. Artificial intelligence (AI) and machine learning (ML) models are mathematical models that find pre-existing relationships in data. These are powerful techniques successful across industries, but when it comes to predicting financial markets, professionals have mixed opinions. In the past 10 years, the financial industry has spent a lot of resources to utilize complex models in stock prediction, but unfortunately, the question remains the same: Are these complex models good enough for predicting financial markets? The mathematical models try to find pre-existing relationships between output variables and input variables, but if a relationship does not exist, then it does not matter which model you use; the prediction would be wrong.

Modern machine learning models (such as deep neural networks and boosting decision tree models) have become increasingly popular in financial market prediction, due to their superior capacity to extract complex non-linear patterns. However, since financial datasets have very low signal-to-noise ratio and are non-stationary, complex models are often very prone to overfitting and suffer from instability issues. Moreover, as various machine learning and data mining tools become more widely used in quantitative trading, many trading firms have been producing an increasing number of features (aka factors). Therefore, how to automatically select effective features becomes an imminent problem. To address these issues, we propose DoubleEnsemble, an ensemble framework leveraging learning trajectory based sample reweighting and shuffling based feature selection. Specifically, we identify the key samples based on the training dynamics on each sample and elicit key features based on the ablation impact of each feature via shuffling. Our model is applicable to a wide range of base models, capable of extracting complex patterns, while mitigating the overfitting and instability issues for financial market prediction. We conduct extensive experiments, including price prediction for cryptocurrencies and stock trading, using both DNN and gradient boosting decision tree as base models. Our experiment results demonstrate that DoubleEnsemble achieves a superior performance compared with several baseline methods.

Convolutional neural networks (CNN) and long short-term memory (LSTM) networks have become a staple of sequence learning. Due to the well-established fact that financial time series data exhibit exceptionally noisy characteristics, capital market anomalies are virtually impossible to detect. We deploy CNN networks for predicting out-of-sample stock movements for 200 high-volume European stocks from 1994 until 2014, and compare its overall performance with a modified LSTM model as in Fischer, Krauss (2017). Specifically, we compare empirical training and validation accuracies of both model architectures and reveal portfolio performance characteristics in terms of return and risk metrics for different portfolio sizes, trying to derive common patterns within the top and flop stocks. Thus, we unveil sources of long-term profitability and demonstrate, that both LSTM and CNN networks are able to extract meaningful information from such noisy financial time series.