Essays on deep learning and asset pricing
This dissertation includes two essays. In the first essay, we take advantage of deep learning and utilize both the price and fundamental information to separate stocks’ winners from losers. For the first model, we used a 2-layer Long Short-Term Memory (LSTM) neural network with past 80 days’ return information as inputs to predict the next day’s return and find a before-trading-cost monthly return of 29.58% with t-statistic of 26.81. For the second model, we design a novel 2-layer LSTM and Multi-layer Perceptron (MLP) hybrid neural network and utilize the monthly return and annual accounting data to predict the returns in the next month. We achieve a monthly return of 2.37% with a t-statistic of 8.97 before trading cost from 1993 through 2017. We use TAQ intraday data to explicitly estimate the trading cost and find that the profits of the daily trading strategy in the first model turn negative. However, the trading strategy utilizing both the price and fundamental information in the second model keeps significantly positive with a monthly return of 1.57% and t-statistic of 6.03 after the trading cost. We show that the 2-layer LSTM and MLP hybrid model performs better than the MLP-only and the hand-engineering double sort trading strategy. In the second essay, we present an empirical test that casts doubt on the widely accepted belief that cash-flow beta can partly explain the value premium. We double sort the stocks with their value and quality dimension and obtain four corner portfolios: (A) expensive quality, (B) cheap junk, (C) cheap quality and (D) expensive junk stocks. Prior research has shown that the value premium concentrates on cheap quality minus expensive junk (i.e., undervalued minus overvalued) but is not significant in cheap junk minus expensive quality stocks. If cash-flow beta is the source of the value premium, we would expect a larger cash-flow beta difference between the cheap quality and expensive junk portfolio. However, our empirical test shows that β_CF ((B) cheap junk) - β_CF ((A) expensive quality) ≫ β_CF ((C) cheap quality) - β_CF ((D) expensive junk). Therefore, our result may serve as evidence that the cash-flow beta may only spuriously explain the value premium.