13 The Rational Trading by Using Economic Fundamentals AOSHIMA Kentaro 14 2 26 ( ) : : :
The Rational Trading by Using Economic Fundamentals AOSHIMA Kentaro abstract: Recently Artificial Markets on which machine agents trade virtual stock have recieved scholarly attention with hard movement of actual financial markets. Researchers try analyzing the complexity of actual markets by using Artificial Markets. U-Mart is also virtual futures market,which uses existent price data as spotprice. Until now, agents on U-Mart use technical analyses for trading or predicting price. But movement of stock prices are affected by economic fundamentals. In this paper I propose the agent which uses news on U-Mart. The agent predicts spotprice with Neural Network which inputdata are technical data and numeric calculated from frequency of specific keywords from news. And I compare the proposed agent to a agent which uses only technical data. As a consequence the proposed agent trade rationally than technical agent. 1,,.,,,,,, [1].,. [2] [4] U-Mart U-Mart if-then 1
, U-Mart,.,,.,. 2, 3 U-Mart. 4, 5, 6,,., 7 2,U-Mart.,,,.,,,,, ( ),.[5] 3 U-Mart, U-Mart. U-Mart, 1.,.,,. U-Mart Server Programmed 1: U-Mart Human U-Mart,,,., ( ).,,U- Mart.,U-Mart,.,U-Mart,. U-Mart..,(if-then),,,, [6],,,.,,. 2
4 U-Mart Server News Server U-Mart, U-Mart U-Mart,,.,U-Mart. 1 PostgreSQL,, ( ). Date Tag DateOfNews News 1999/1/2 23 1999/1/2 newsa 1999/1/3 24 1999/1/2 newsb 1999/1/4 25 1999/1/4 newsc 1999/1/5 26 1999/1/5 newsd 1:,,U-Mart. 2,U-Mart,U-Mart,. 1. 2. U-Mart 3. Date Tag Date News 21/9/11 51 21/9/12 aaaa 21/9/12 52 21/9/12 bbbb 21/9/13 53 21/9/14 cccc 2:,U-Mart 4. String,,,, 5,.,,.,, 5.1,.,, 1 2 3
.,U-Mart J3. J3,.,, 5.2,.,, [7], [8],.,,, 1, 1 9.( 2 ), 1.,. Good(+1) Bad(-1) Keyword List,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, 2: keyword list,. 5.3, [9]., [1] [11] F Input = m k=1 G apps k n l=1 B apps l m k=1 G apps k + n l=1 B apps l (1) m,n, G appsk k,b appsl l. F Input F Input 5.3.1, ( ), 3. 4
, 13 13 (F Input), 1 2. 5.4 3: 5.3.2,, 4,,. 4. 4:, 3 1 3 3 3,, 5,,.., 4,,..,,, (p) (p1),p1 (p2).. p p1 p1 p2,p1,p2 p p1 p1 p2,p2,p1 p p1 p1 p2,p1,p2 p p1 p1 p2,p2,p1, 5
,. U-Mart ( )., thelatest spotprice p trading day predict-1 p1 next day predict-2 p2 case1 (p<p1,p1<p2) SELL > p2, BUY < p1 case2 (p<p1,p1>p2) SELL > p1, BUY < p2 case3 (p>p1,p1<p2) SELL > p2, BUY < p1 case4 (p>p1,p1>p2) SELL > p1, BUY < p2 5:,U-Mart 6., U-Mart,,,U-Mart., 4 KeywordList,,,. 6 U-Mart 6.1,. 6.2 6: 6.2.1 7. 2 5.. 1st Term 2nd Term 3rd Term 4th Term 5th Term 3 1 Learning Term Experiment Term 2 7: 6.2.2,,,21 11 1 2( A: 3 ), U-Mart21 25 21 45( A: 4 ) 6
.,. ( 2), ( 3) 38. 6.3 5, (2) (3). 8 12. (1 9 ). 13 5.,, 1st term, 1st term, 5, 5 1st term.,,,,,,. 5th term. 14,,. 15,16.y []. 6.,.,,,,.,.,, ( ),,,, 1st term 17. 1st term 5 F Input. ( ),5 F Input,, ( ).,F Input ( ), ( ).,, F Input, if-then 7
3.5e+8 9e+8 3e+8 8e+8 2.5e+8 7e+8 Profit 2e+8 1.5e+8 1e+8 Profit 6e+8 5e+8 4e+8 3e+8 5e+7 2e+8 1e+8-5e+7 32 17 29 37 42 43 4 23 41 12 31 16 28 3 14 39 18 27 38 13 19 22 34 33 2 3 14 45 34 2 35 27 23 21 24 3 37 22 36 13 4 33 16 12 8: 1st term profit 11: 4th term profit 5e+8 1.6e+9 4.5e+8 1.4e+9 4e+8 1.2e+9 3.5e+8 3e+8 1e+9 Profit 2.5e+8 Profit 8e+8 2e+8 6e+8 1.5e+8 4e+8 1e+8 5e+7 2e+8 34 37 17 45 28 2 3 27 32 31 23 12 4 43 24 16 25 38 22 14 21 19 33 45 18 13 27 37 34 14 2 21 25 12 36 28 3 44 17 16 33 9: 2nd term profit 12: 5th term profit 1.6e+9 6e+8 1.4e+9 5e+8 1.2e+9 4e+8 1e+9 Profit 8e+8 Profit 3e+8 6e+8 2e+8 4e+8 1e+8 2e+8 15 45 18 27 34 24 28 14 2 21 3 25 17 4 37 22 33 26 16 19 12 23 45 34 14 27 37 2 28 13 18 21 3 12 17 25 24 16 44 4 36 23 22 33 31 1: 3rd term profit 13: Average of profit 8
44 F-Input spotprice 1.8 1 42.6 44.8 4.4 price 42 4 38 36 34 32 3.6.4.2 -.2 -.4 -.6 -.8 f-input spot price 38.2 36 -.2 34 -.4 32 -.6 3 -.8 28 2 4 6 8 1-1 trading times F-Input 28-1 2 4 6 8 1 trading times 17: F Input SpotPrice (1st term) f-input spotprice futreprice F-BuyOrderPrice F-SellOrderPrice T-BuyOrderPrice T-SellOrderPrice 14: 5th term 7 5 45 F-position, U-Mart,,. SellQuant-BuyQuant 4 35 3 25 2 15 1 5 1 2 3 4 5 6 7 8 9 1 trading times 15: SellQuant-BuyQuant 5 T-position 4 3 2 1-1 -2 1 2 3 4 5 6 7 8 9 1 trading times 16:, ( 9 ),.,,.,,,., (+1-1),,,,., 13 3,,. 9
,,.,.,,,,,,,,,,.,,,,,,,.,, OB, [1],,,,,,,,, Vol.15, No. 6, pp 982-989, 2. [2],,, 2-ICS- 119, Vol.99, pp1-8,2. [3],,,, 21-ICS- 123, pp37-42,21. [4] K.Izumi,K.Ueda, Analysis of dealers s processing financial news based on an artificial market approach, Journal of Computational Intelligence in Finance Vol.7, pp.23-33, 1999. [5],,,21. [6],, 21. [7] J. Thomas and K. Sycara, Integrating Genetic Algorithms and Text Learning for Financial Prediction, Proceedings ofthe GECCO-2 Workshop on Data Mining with Evolutionary Algorithms, July, 1999. [8] Y.Matsumoto,A.Kitauchi,T.Yamashita, Y.Hirano,H.Matsuda,K.Takaoka,M.Asahara, Japanese Morphological Analysis System ChaSen version 2.2.1, Dec, 2. [9] N.Baba,H.Handa,M.Hayashi, Utilization of neural networks and GAs for cnstruction an intelligent intelligent decision support system to deal stocks,proceedings of SPIE Conference,Vol.276,pp.164-174,1996. 1
[1] N.Baba et al. A Hybrid Algorithm for Finding Global Minimum of Error Function of Neural Networks and Its Applications,Neural NetWorks Vol.7,pp.1253-1265,1994. [11],,,,1999. 11
A : 3 1. 5 11, 12 5 13 14., 15. 2. 1 16. 17.,, 18., 1., 19.. 1 2. 1. 3: ( ) 21 22,23 24 32 33 34,35 36,37 38 39 4 41,42 43 44 45. GA., 24 26 2,4,8.27 32 26., 34 33 ( ).35. 36,.37. 4 +2-2 41.5.42 12 RSI U/(U+D) U D=.7.3 4: (U-Mart21) 12
B B.3 (5 ),,,. B.1 5 B.2 5 36 5,,,, 5 18,5,.,.,,,,,,,,,., Average Profit 2.5e+8 2e+8 1.5e+8 1e+8 5e+7-5e+7 2 3 8 7 5 4 6 agent4 ±2 agent5 ±3 agent6 ±4 agent7 ±6 agent8 ±1 18: ( ) 5: 13