Evaluating the Effectiveness of Candlestick Analysis in Forecasting U.S. Stock Market

Wang, Mengjiao and Wang, Yujing (2019) Evaluating the Effectiveness of Candlestick Analysis in Forecasting U.S. Stock Market. Proceedings of the 2019 3rd International Conference on Compute and Data Analysis - ICCDA 2019 . pp. 98-101.

Full text not available from this repository.

Official URL: https://doi.org/10.1145/3314545.3314555


Predicting the stock market trend has drawn wide attention from the public, industry, and academia. There are various strategies for investment, ranging from the buy-and-hold strategy that let time makes money to high-frequency trading based on complicated models of machine learning. For the general public, the candlestick analysis is a simple and straightforward way to predict the market trend and help make decisions on buy and sell. However, the general effectiveness and validity of such methodology are understudied, and many previous guidebooks on candlestick analysis were based on specific cases and the summary of prior experiences, but lack unbiased and rigorous confirmation. Here, we aim at evaluating the effectiveness of multiple famous candlestick patterns based on recent data of 20 U.S. stocks. By estimating the fraction of correct guesses and expected return, we were able to show that three of the four patterns under scrutiny were not effective in generating profitable outcomes, but one of them do provide valuable information on market trend reversal. Although testing more candlestick patterns and using more test data may lead to the discoveries of more patterns being significantly profitable, our results call for scrutiny and scientific guidance for the general public when applying the candlestick analysis, and questions the general validity and applicability of such analyses.

Repository Staff Only: item control page