µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)
ÇѱÛÁ¦¸ñ(Korean Title) |
¿Â¶óÀÎ ´º½º¿Í ±â¼úÀû ÁöÇ¥¸¦ ÀÌ¿ëÇÑ Å׸¶ÁÖ µî¶ô ¿¹Ãø |
¿µ¹®Á¦¸ñ(English Title) |
Price Fluctuation Prediction for Theme Stocks Using Online News and Technical Indicators |
ÀúÀÚ(Author) |
¹Ú´ö¿µ
Deog-Yeong Park
À̱âÈÆ
Ki-Hoon Lee
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 35 NO. 03 PP. 0066 ~ 0076 (2019. 12) |
Çѱ۳»¿ë (Korean Abstract) |
ƯÁ¤ ÁÖÁ¦ÀÇ ´º½º¿¡ ÀÇÇØ ÁÖ°¡°¡ ¿µÇâÀ» ¹Þ´Â ±â¾÷±ºÀ» Å׸¶ÁÖ¶ó°í ÇÑ´Ù. ÃÖ±Ù ¿Â¶óÀÎ ´º½º¿Í ±â¼úÀû ÁöÇ¥¸¦ È°¿ëÇÑ ÁÖ°¡¿¹ÃøÀÌ ¿¬±¸µÇ°í ÀÖÁö¸¸, ÅõÀÚ À§ÇèÀÌ Å« Å׸¶ÁÖ¿¡ ´ëÇÑ ¿¬±¸´Â ºÎÁ·ÇÏ´Ù. º» ³í¹®¿¡¼´Â °¨¼ºÁöÇ¥¿Í ±â¼úÀû ÁöÇ¥¸¦ ÀÌ¿ëÇÏ¿© Å׸¶ÁÖ¿¡ ´ëÇØ ÁÖ°¡ µî¶ôÀ» ¿¹ÃøÇÏ´Â ¸ðµ¨À» Á¦¾ÈÇÑ´Ù. Á¦¾ÈÇÑ ¸ðµ¨Àº Gated Recurrent Unit (GRU)¸¦ »ç¿ëÇØ ´º½º·ÎºÎÅÍ °¨¼ºÁöÇ¥¸¦ °è»êÇÏ°í, °¨¼ºÁöÇ¥¿Í ±â¼úÀû ÁöÇ¥¸¦ ¹ÙÅÁÀ¸·Î eXtreme Gradient Boosting (XGBoost)À» ÅëÇØ ÁÖ°¡ µî¶ôÀ» ¿¹ÃøÇÑ´Ù. ºÏÇÑ Å׸¶¿¡ ¼ÓÇÏ´Â 10°³ ±â¾÷À» ÀÌ¿ëÇÏ¿© ½ÇÇèÇÑ °á°ú, Á¦¾ÈÇÑ ¸ðµ¨ÀÌ Ãֽſ¬±¸ÀÇ ¸ðµ¨º¸´Ù Æò±Õ Á¤È®µµ°¡ ÃÖ´ë 19.0%P ´õ ³ô¾Ò´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
Theme stocks are a group of companies whose stock prices are affected by news of a certain subject. Recently, there have been research efforts to predict stock prices using online news and technical indicators, but little attention has been paid so far to theme stocks, which have high investment risks. In this paper, we propose a model that predicts stock price fluctuations for theme stocks using a sentiment indicator and technical indicators. The proposed model calculates a sentiment indicator from news using Gated Recurrent Unit (GRU) and predicts stock price fluctuations through eXtreme Gradient Boosting (XGBoost) based on the sentiment indicator and technical indicators. The experimental results using 10 companies belonging to the North Korea theme show that the proposed model improves the average accuracy up to 19.0%P compared with the state-of-the-art model.
|
Å°¿öµå(Keyword) |
Å׸¶ÁÖ
ÁÖ°¡¿¹Ãø
µö·¯´×
GRU
XGBoost
Theme Stocks
Stock Price Prediction
Deep Learning
GRU
XGBoost
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|