Approach for retrieving similar stock price patterns using dynamic programming method
| Autor: | Yoshihisa Udagawa |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Dynamic time warping
Candlestick chart Computer science 02 engineering and technology 01 natural sciences 010305 fluids & plasmas Longest common substring problem Longest common subsequence problem Dynamic programming Chart Technical analysis 0103 physical sciences 0202 electrical engineering electronic engineering information engineering Econometrics 020201 artificial intelligence & image processing Stock (geology) |
| Zdroj: | iiWAS |
| DOI: | 10.1145/3151759.3151820 |
| Popis: | With the widespread use of the internet, online stock trading becomes popular. A huge amount of stock trading data are accumulated in the internet. Now, stock price prediction is a challenging research subject of the data mining techniques. Because stock price can vary according to uncontrollable factors such as interest ratios, investors' sentiment, or political actions, the fluctuation of stock prices moves seemingly random. However, technical analyses of stock prices recognize that there are chart patterns occurred repeatedly known as "Japanese candlestick chart patterns" for examples. In this paper, we propose a dynamic programming approach to retrieve similar stock price patterns. The longest common substring (LCS) algorithm is improved to deal with similar numeric sequences. The proposed LCS algorithm is compared with the Dynamic Time Warping (DTW) measure through experiments using the Nikkei stock average. Results on a morning star pattern being known as a powerful reversal pattern show that the proposed LCS algorithm finds the results as expected. However, from the viewpoint of investors, the proposed algorithm has room for improvements. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|