<![CDATA[Regression based estimation or prediction on the stock price can be done traditionally and in a modern way. In this research it is done traditionallywith regard to instrinsic/ fundamental values of the company. Traditionallysome literature uses PER as a model of estimation, but this model have weaknesses if the Analyst or the share owners do not know how to read the financial report or they just trust the numbers on the report. The alternative way to use a model of estimation in this research considered the DER, ROA and ROE variables. This research used a quantitative research. The population of this research is 45 companies. The kind of data used is secondary data. The data analysis used is Double Linier Regression Analysis. The result of the research is the validity test show that the Communality number is higher than Rule of Thumbs in the point of 0,5 (table 4.4) which identifies that the regression equation made do not have multikoliniarity and the regression is able to estimate correctly (table 4.5). it has F-test and it is significant in 0,000, so that  it can be concluded that the model made does not have independent variable which influence the dependent one. In this research, the researcher deleted the free variable DER. This is because the researcher wants a model with estimation that can predict as exact as possible. Therefore, the regression equation is: the stock price = 1097.090 + 1141.427 ROA + 5407.516 ROE. The conclusions of this research are, (1) the result of F test shows that the regression model can be used to predict the stock price. In another word, the independent variable consists of ROA and ROE, some or all of its independent variables influence their dependent variable which is the stock price. (2) t Test of ROA and ROE have a positive effect on the ROE stock price. (3) ROA and ROE can explain 86, 7% effect on the stock share.]]>
