web analytics

Profitable Trading Strategies Devised Through Machine Learning

Published:
Lead Inventor: Dacheng Xiu
Print or Save to PDF

SUMMARY

  • Sentimental analysis for financial forecasting has been impeded by the ultra-high dimensional challenge of numerically representing huge troves of text.
  • Prior efforts circumvented this bottleneck by reusing word scoring devised for earlier studies, utilizing sentiment values imprecisely matched to context, significantly reducing the value of the results.
  • The faculty inventor utilized machine learning to perform sentimental analysis of the archives and text feed of the Dow Jones Newswires over the past several decades. Words in these texts were scored positive, negative, or neutral, as per their correlation with profitable asset returns. These scores were summed for entire articles or feeds to produce sentiment values highly predictive of asset performance.
  • The close correlation shown between news releases and price values enabled devising a trading strategy operating through buying and selling an equal ratio of assets predicated to perform well or poorly, respectively.
  • The results demonstrated robust sentimental analysis capable to being expertly tailored for specific forecasting that both produced accurate financial predictions and surmounted the higher dimensional challenge.

FIGURE

Table 1 reports the performance of Exponentially-Weighted calendar Time (EWCT) portfolios as we vary turnover limits from mild (γ = 0.9) to heavily restricted (γ = 0.1). Moving down the rows we see that a more severe turnover restriction drags down the gross Sharpe ratio of the trading strategy, indicating a loss in predictive information due to signal smoothing. This drag is offset by a reduction in trading costs. As a result, the net Sharpe ratio peaks at 2.3 when γ = 0.5. That is, with a moderate amount of turnover control (and concomitant signal smoothing), the gain from reducing transaction costs outweighs the loss in predictive power. In sum, Table 5 demonstrates the attractive risk-return tradeoff to investing based on news sentiment even after accounting for transaction costs.

 

ADVANTAGES

ADVANTAGES

  • Accurate forecasts of price changes
  • Non-computationally intensive
  • Sentimental scoring expertly tuned to specific contexts

APPLICATIONS

  • Trading and hedging strategies
  • Portfolio management 
  • Economic planning

 

PUBLICATIONS

https://www.nber.org/system/files/working_papers/w26186/w26186.pdf