Thực hiện mơ hình ARIMA

Một phần của tài liệu TIỂU LUẬN môn học KINH tế LƯỢNG ỨNG DỤNG NGÀNH tài CHÍNH đề tài dự báo GIÁ CHỨNG KHOÁN TỔNG CTCP BIA RƯỢUNƯỚC GIẢI KHÁT sài gòn (Trang 44 - 53)

Mơ hình chuỗi thời gian đơn biến hay cịn được gọi với cái tên mơ hình ARIMA được sử dụng khi hành vi của biến số cần giải thích được quyết định bởi những thơng tin về giá trị của chính nó trong q khứ và/hoặc giá trị hiện tại và quá khứ của hạng nhiễu.

Mơ hình tự hồi qui là giá trị ước tính tương lai của mơ hình phân tích chuỗi thời gian chỉ phụ thuộc vào giá trị trong quá khứ.: Yt-1, Yt-2..., Yt-n và sai số ut. Mơ hình tự hồi quy bậc 1có dạng:

AR(1): Yt = µ + ϕ Yt-1 + ut

Trong mơ hình tự hồi qui q trình phụ thuộc vào tổng có trọng số của các giá trị quá khứ và số hạng ngẫu nhiên:

Mơ hình trung bình trượt (MA) là loại mơ hình đơn giản nhất trong nhóm các mơ hình chuỗi thời gian mà giá trị hiện tại của biến số Yt chỉ phụ thuộc vào sai số ut. Mơ

hình trung bình trượt bậc 1 có dạng:

MA(1): Yt = µ + θ1ut-1+ ut

Trong mơ hình trung bình trượt q trình phụ thuộc vào tổng có trọng số của các ngẫu nhiên hiện hành có độ trễ:

MA(q): Yt = µ + θ1ut-1 + θ2ut-2 + θ3ut-3 + …+ θqut-q+ ut Trong đó: ut là nhiễu trắng

µ (const) là hằng số t=1,2,3...n

ARIMA (1,1,1)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:12

Sample (adjusted): 12/08/2016 5/31/2021 Included observations: 1115 after adjustments Convergence achieved after 65 iterations MA Backcast: 12/07/2016

Variable Coefficient Std. Error t-Statistic Prob.

C -0.035059 0.142524 -0.245983 0.8057

AR(1) 0.804896 0.064169 12.54344 0.0000

MA(1) -0.767291 0.071414 -10.74420 0.0000

R-squared 0.029750 Mean dependent var 0.023303

Adjusted R-squared 0.028005 S.D. dependent var 4.020613

S.E. of regression 3.963914 Akaike info criterion 5.595028

Sum squared resid 17472.43 Schwarz criterion 5.608525

Log likelihood -3116.228 Hannan-Quinn criter. 5.600131

F-statistic 17.04832 Durbin-Watson stat 1.890245

Prob(F-statistic) 0.000000

Inverted AR Roots .80 Inverted MA Roots .77

SPSABt = -0.0351 + 0.8049AR(1) − 0.7673MA(1) + ȗt SPSABt = -0.0351 + 0.8049SPSABt-1 − 0.7673ȗt-1 + ȗt

SABt − SABt-1= -0.0351 + 0.8049 (SABt-1 − SABt-2) – 0.7673ȗt-1 + ȗt SABt = -0.0351 + 1.8049SABt-1 − 0.8049SABt-2 – 0.7673ȗt-1 + ȗt

ARIMA (1,1,1) (20,1,11)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 12:45

Sample (adjusted): 1/05/2017 5/31/2021 Included observations: 1096 after adjustments Convergence achieved after 11 iterations MA Backcast: 12/20/2016 1/04/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.023707 0.113483 -0.208903 0.8346

AR(1) -0.216060 0.243681 -0.886650 0.3755

AR(20) -0.064365 0.029020 -2.217957 0.0268

MA(1) 0.287898 0.239214 1.203518 0.2290

MA(11) -0.054049 0.029529 -1.830394 0.0675

R-squared 0.013438 Mean dependent var -0.025500

Adjusted R-squared 0.009821 S.D. dependent var 3.916682

S.E. of regression 3.897401 Akaike info criterion 5.563049

Sum squared resid 16572.00 Schwarz criterion 5.585856

Log likelihood -3043.551 Hannan-Quinn criter. 5.571679

F-statistic 3.715207 Durbin-Watson stat 1.980966

Prob(F-statistic) 0.005194

Inverted AR Roots .85-.14i .85+.14i .77+.40i .77-.40i .61+.62i .61-.62i .39+.78i .39-.78i .13-.86i .13+.86i -.15+.86i -.15-.86i -.41-.78i -.41+.78i -.63-.62i -.63+.62i -.79-.40i -.79+.40i -.87+.14i -.87-.14i Inverted MA Roots .74 .62+.41i .62-.41i .29+.69i

.29-.69i -.14+.75i -.14-.75i -.53+.58i -.53-.58i -.77+.21i -.77-.21i

SPSABt = -0.0237 - 0.2161AR(1) – 0.0644AR(20) + 0.2879MA(1) – 0.054MA(11) + ȗt

SPSABt = -0.0237 - 0.2161SPSABt-1 – 0.0644SPSABt-20 + 0.2879ȗt-1 – 0.054ȗt-11 + ȗt SABt − SABt-1 = - 0.0237 - 0.2161(SABt-1 − SABt-2) – 0.0644(SABt-20 − SABt-21) + 0.2879ȗt-1 – 0.054 ȗt-11 + ȗt

SABt = - 0.0237 + 0.7839SABt-1 + 0.261SABt-2 – 0.0644SABt-20 + 0.0644SABt-21 + 0.2879ȗt-1 – 0.054 ȗt-11 + ȗt

ARIMA (1,1,1) (20,1,20)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:18

Included observations: 1096 after adjustments Convergence achieved after 48 iterations MA Backcast: 12/07/2016 1/04/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.024624 0.119317 -0.206373 0.8365

AR(1) -0.077168 0.295093 -0.261503 0.7938

AR(20) -0.026943 0.112884 -0.238677 0.8114

MA(1) 0.154900 0.292505 0.529565 0.5965

MA(20) -0.039484 0.115826 -0.340894 0.7332

R-squared 0.010302 Mean dependent var -0.025500

Adjusted R-squared 0.006673 S.D. dependent var 3.916682

S.E. of regression 3.903591 Akaike info criterion 5.566223

Sum squared resid 16624.69 Schwarz criterion 5.589030

Log likelihood -3045.290 Hannan-Quinn criter. 5.574852

F-statistic 2.839120 Durbin-Watson stat 1.992527

Prob(F-statistic) 0.023311

SPSABt = -0.0246 – 0.077AR(1) – 0.0269AR(20) + 0.1549MA(1) – 0.0395MA(20) + ȗt

SPSABt = -0.0246 – 0.077SPSABt-1 – 0.0269SPSABt-20 + 0.1549ȗt-1 – 0.0395ȗt-20 + ȗt SABt − SABt-1 = -0.0246 – 0.077(SABt-1 − SABt-2) – 0.0269(SABt-20 − SABt-21) + 0.1549ȗt-1 – 0.0395ȗt-20 + ȗt

SABt = -0.0246 – 0.923SABt-1 + 0.077SABt-2 – 0.0269SABt-20 + 0.0269SABt-21 + 0.1549ȗt-1 – 0.0395ȗt-20 + ȗt

ARIMA (1,1,1) (20,1,26)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:19

Sample (adjusted): 1/05/2017 5/31/2021 Included observations: 1096 after adjustments Convergence achieved after 12 iterations MA Backcast: 11/29/2016 1/04/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.022006 0.122969 -0.178951 0.8580

AR(1) -0.045971 0.288418 -0.159390 0.8734

AR(20) -0.061262 0.029386 -2.084726 0.0373

MA(1) 0.122349 0.286965 0.426357 0.6699

MA(26) 0.033691 0.030265 1.113198 0.2659

R-squared 0.011292 Mean dependent var -0.025500

Adjusted R-squared 0.007667 S.D. dependent var 3.916682

Sum squared resid 16608.06 Schwarz criterion 5.588029

Log likelihood -3044.742 Hannan-Quinn criter. 5.573852

F-statistic 3.115085 Durbin-Watson stat 1.990858

Prob(F-statistic) 0.014605

SPSABt = -0.22 – 0.046AR(1) – 0.0612AR(20) + 0.1223MA(1) + 0.0337MA(26) + ȗt SPSABt = -0.22 – 0.046SPSABt-1 – 0.0612SPSABt-20 + 0.1223ȗt-1 + 0.0337ȗt-26 + ȗt SABt − SABt-1 = -0.22 – 0.046(SABt-1 − SABt-2) – 0.0612(SABt-20 − SABt-21) + 0.1223ȗt-1 + 0.0337ȗt-26 + ȗt

SABt = -0.22 + 0.954SABt-1 + 0.046SABt-2 – 0.0612SABt-20 + 0.0612SABt-21 + 0.1223ȗt-1 + 0.0337ȗt-26 + ȗt

ARIMA (1,1,1) (21,1,11)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:20

Sample (adjusted): 1/06/2017 5/31/2021 Included observations: 1095 after adjustments Convergence achieved after 20 iterations MA Backcast: 12/21/2016 1/05/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.036523 0.133184 -0.274230 0.7840

AR(1) 0.370450 0.245818 1.507014 0.1321

AR(21) 0.043991 0.027576 1.595293 0.1109

MA(1) -0.295123 0.253004 -1.166474 0.2437

MA(11) -0.044853 0.029163 -1.537994 0.1243

R-squared 0.011517 Mean dependent var -0.025608

Adjusted R-squared 0.007890 S.D. dependent var 3.918470

S.E. of regression 3.902981 Akaike info criterion 5.565914

Sum squared resid 16604.26 Schwarz criterion 5.588738

Log likelihood -3042.338 Hannan-Quinn criter. 5.574551

F-statistic 3.175074 Durbin-Watson stat 1.988176

Prob(F-statistic) 0.013184

Inverted AR Roots .88 .85+.25i .85-.25i .73+.48i .73-.48i .56-.67i .56+.67i .33-.80i .33+.80i .08-.86i .08+.86i -.18+.84i -.18-.84i -.42+.74i -.42-.74i -.62-.58i -.62+.58i -.76+.37i -.76-.37i -.84+.13i -.84-.13i

Inverted MA Roots .79 .67+.40i .67-.40i .34-.68i .34+.68i -.08-.74i -.08+.74i -.47-.57i -.47+.57i -.70+.21i -.70-.21i

SPSABt = -0.0365 + 0.3704AR(1) – 0.044AR(21) – 0.2951MA(1) – 0.0449MA(11) + ȗt

SPSABt = -0.0365 + 0.3704SPSABt-1 – 0.044SPSABt-21 – 0.2951ȗt-1 – 0.0449ȗt-11 + ȗt SABt − SABt-1 = -0.0365 + 0.3704(SABt-1 − SABt-2) – 0.044(SABt-21 − SABt-22) – 0.2951ȗt-1 – 0.0449ȗt-11 + ȗt

SABt = -0.0365 + 1.3704SABt-1 − 0.3704SABt-2 – 0.044SABt-21 + 0.044SABt-22 – 0.2951ȗt-1 – 0.0449ȗt-11 + ȗt

ARIMA (1,1,1) (21,1,20)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:20

Sample (adjusted): 1/06/2017 5/31/2021 Included observations: 1095 after adjustments Convergence achieved after 13 iterations MA Backcast: 12/08/2016 1/05/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.053093 0.146142 -0.363296 0.7165

AR(1) 0.693157 0.147291 4.706039 0.0000

AR(21) 0.067591 0.025618 2.638368 0.0084

MA(1) -0.642041 0.157749 -4.070016 0.0001

MA(20) -0.064225 0.029639 -2.166925 0.0305

R-squared 0.014120 Mean dependent var -0.025608

Adjusted R-squared 0.010502 S.D. dependent var 3.918470

S.E. of regression 3.897841 Akaike info criterion 5.563278

Sum squared resid 16560.55 Schwarz criterion 5.586102

Log likelihood -3040.895 Hannan-Quinn criter. 5.571915

F-statistic 3.902695 Durbin-Watson stat 1.944884

Prob(F-statistic) 0.003745

Inverted AR Roots .94 .89+.25i .89-.25i .77+.48i .77-.48i .58-.67i .58+.67i .35-.81i .35+.81i .09+.87i .09-.87i -.17+.85i -.17-.85i -.41+.76i -.41-.76i -.62+.59i -.62-.59i -.77+.38i -.77-.38i -.85+.13i -.85-.13i

Inverted MA Roots .92 .88+.26i .88-.26i .75+.50i .75-.50i .55-.69i .55+.69i .30-.82i .30+.82i .03-.86i .03+.86i -.24-.82i -.24+.82i -.49+.70i -.49-.70i -.68+.51i -.68-.51i -.80-.27i -.80+.27i -.85

SPSABt = -0.0531 + 0.6932AR(1) + 0.0676AR(21) – 0.642MA(1) – 0.0642MA(20) + ȗt

SPSABt = -0.0531 + 0.6932SPSABt-1 + 0.0676SPSABt-21 – 0.642ȗt-20 – 0.0642ȗt-20 + ȗt SABt − SABt-1 = -0.0531 + 0.6932(SABt-1 − SABt-2) + 0.0676(SABt-21 − SABt-22) – 0.642ȗt-20 – 0.0642ȗt-20 + ȗt

SABt = -0.0531 + 1.6932SABt-1 – 0.6932SABt-2 + 0.0676SABt-21 − 0.0676SABt-22 – 0.642ȗt-20 – 0.0642ȗt-20 + ȗt

ARIMA (1,1,1) (21,1,26)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:21

Sample (adjusted): 1/06/2017 5/31/2021 Included observations: 1095 after adjustments Convergence achieved after 12 iterations MA Backcast: 11/30/2016 1/05/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.049224 0.174179 -0.282604 0.7775

AR(1) 0.734381 0.143553 5.115743 0.0000

AR(21) 0.031136 0.020992 1.483224 0.1383

MA(1) -0.690854 0.153162 -4.510620 0.0000

MA(26) 0.036864 0.025655 1.436889 0.1510

R-squared 0.012945 Mean dependent var -0.025608

Adjusted R-squared 0.009323 S.D. dependent var 3.918470

S.E. of regression 3.900161 Akaike info criterion 5.564469

Sum squared resid 16580.27 Schwarz criterion 5.587293

Log likelihood -3041.547 Hannan-Quinn criter. 5.573105

F-statistic 3.573825 Durbin-Watson stat 1.933770

Prob(F-statistic) 0.006638

Inverted AR Roots .92 .87-.23i .87+.23i .74+.46i .74-.46i .57+.65i .57-.65i .34-.77i .34+.77i .09+.83i .09-.83i -.16+.82i -.16-.82i -.40+.73i -.40-.73i -.60-.57i -.60+.57i -.74-.36i -.74+.36i -.81+.13i -.81-.13i

Inverted MA Roots .92-.10i .92+.10i .86-.30i .86+.30i .76-.49i .76+.49i .61+.65i .61-.65i .44+.77i .44-.77i .24+.85i .24-.85i .02+.87i .02-.87i -.19-.85i -.19+.85i -.39+.77i -.39-.77i -.56+.66i -.56-.66i -.70+.50i -.70-.50i -.80-.31i -.80+.31i -.85-.11i -.85+.11i

SPSABt = -0.4922 + 0.7344AR(1) + 0.0311AR(21) – 0.6909MA(1) – 0.0367MA(26) + ȗt

SPSABt = -0.4922 + 0.7344SPSABt-1 + 0.0311SPSABt-21 – 0.6909ȗt-1 – 0.0367ȗt-26 + ȗt SABt − SABt-1 = -0.4922 + 0.7344(SABt-1 − SABt-2) + 0.0311(SABt-21 − SABt-22) – 0.6909ȗt-1 – 0.0367ȗt-26 + ȗt

SABt = -0.4922 + 1.7344SABt-1 – 0.7344SABt-2 + 0.0311SABt-21 − 0.0311SABt-22 – 0.6909ȗt-1 – 0.0367ȗt-26 + ȗt

ARIMA (1,1,1) (35,1,11)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:21

Sample (adjusted): 2/02/2017 5/31/2021 Included observations: 1081 after adjustments Convergence achieved after 8 iterations MA Backcast: 1/11/2017 1/25/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.044353 0.114135 -0.388602 0.6976

AR(1) -0.328075 0.219991 -1.491314 0.1362

AR(35) -0.064246 0.028565 -2.249141 0.0247

MA(1) 0.395051 0.214133 1.844885 0.0653

MA(11) -0.055986 0.028599 -1.957620 0.0505

R-squared 0.013347 Mean dependent var -0.048940

Adjusted R-squared 0.009679 S.D. dependent var 3.919430

S.E. of regression 3.900416 Akaike info criterion 5.564658

Sum squared resid 16369.45 Schwarz criterion 5.587718

Log likelihood -3002.698 Hannan-Quinn criter. 5.573389

F-statistic 3.638847 Durbin-Watson stat 1.967761

Prob(F-statistic) 0.005933

Inverted AR Roots .91-.08i .91+.08i .88+.25i .88-.25i .82-.40i .82+.40i .74-.54i .74+.54i .63+.67i .63-.67i .50-.77i .50+.77i .35+.85i .35-.85i .20+.90i .20-.90i .03-.92i .03+.92i -.13-.91i -.13+.91i -.30-.88i -.30+.88i -.45+.81i -.45-.81i -.59-.72i -.59+.72i -.71+.61i -.71-.61i -.80-.47i -.80+.47i -.88-.32i -.88+.32i -.92-.16i -.92+.16i -.94

Inverted MA Roots .74 .62+.41i .62-.41i .29-.69i .29+.69i -.14+.75i -.14-.75i -.54-.57i -.54+.57i -.78-.21i -.78+.21i

SPSABt = -0.4444 – 0.3281AR(1) – 0.0642AR(35) + 0.3951MA(1) – 0.056MA(11) + ȗt

SPSABt = -0.4444 – 0.3281SPSABt-1 – 0.0642SPSABt-35 + 0.3951ȗt-1 – 0.056ȗt-11 + ȗt SABt − SABt-1 = -0.4444 – 0.3281(SABt-1 − SABt-2) – 0.0642(SABt-35 − SABt-36) + 0.3951ȗt-1 – 0.056ȗt-11 + ȗt

SABt = -0.4444 + 0.6719SABt-1 + 0.3281SABt-2 – 0.0642SABt-35 + 0.0642SABt-36 + 0.3951ȗt-1 – 0.056ȗt-11 + ȗt

ARIMA (1,1,1) (35,1,20)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:23

Sample (adjusted): 2/02/2017 5/31/2021 Included observations: 1081 after adjustments Convergence achieved after 11 iterations MA Backcast: 12/28/2016 1/25/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.044442 0.113917 -0.390125 0.6965

AR(1) -0.194310 0.238776 -0.813775 0.4160

AR(35) -0.059035 0.029229 -2.019710 0.0437

MA(1) 0.268201 0.234460 1.143912 0.2529

MA(20) -0.065391 0.029929 -2.184864 0.0291

R-squared 0.014214 Mean dependent var -0.048940

Adjusted R-squared 0.010549 S.D. dependent var 3.919430

S.E. of regression 3.898701 Akaike info criterion 5.563779

Sum squared resid 16355.06 Schwarz criterion 5.586839

Log likelihood -3002.222 Hannan-Quinn criter. 5.572510

F-statistic 3.878698 Durbin-Watson stat 1.979968

Prob(F-statistic) 0.003908

SPSABt = -0.4444 – 0.1943AR(1) – 0.0590AR(35) + 0.2682MA(1) – 0.0654MA(20) + ȗt

SPSABt = -0.4444 – 0.1943SPSABt-1 – 0.0590SPSABt-35 + 0.2682ȗt-1 – 0.0654ȗt-20 + ȗt SABt − SABt-1 = -0.4444 – 0.1943(SABt-1 − SABt-2) – 0.0590(SABt-35 – SABt-36) + 0.2682ȗt-1 – 0.0654ȗt-20 + ȗt

SABt = -0.4444 + 0.8057SABt-1 + 0.1943SABt-2 – 0.0590SABt-35 + 0.0590SABt-36 + 0.2682ȗt-1 – 0.0654ȗt-20 + ȗt

ARIMA (1,1,1) (35,1,26)

Dependent Variable: SPSAB Method: Least Squares Date: 06/10/21 Time: 11:23

Sample (adjusted): 2/02/2017 5/31/2021 Included observations: 1081 after adjustments Convergence achieved after 20 iterations MA Backcast: 12/20/2016 1/25/2017

Variable Coefficient Std. Error t-Statistic Prob.

C -0.044944 0.124411 -0.361253 0.7180

AR(1) -0.071462 0.291883 -0.244833 0.8066

AR(35) -0.055811 0.029484 -1.892931 0.0586

MA(1) 0.147489 0.289771 0.508986 0.6109

MA(26) 0.033658 0.030320 1.110081 0.2672

R-squared 0.010509 Mean dependent var -0.048940

Adjusted R-squared 0.006831 S.D. dependent var 3.919430

S.E. of regression 3.906021 Akaike info criterion 5.567530

Sum squared resid 16416.53 Schwarz criterion 5.590590

Log likelihood -3004.250 Hannan-Quinn criter. 5.576261

F-statistic 2.857009 Durbin-Watson stat 1.988348

Prob(F-statistic) 0.022626

SPSABt = -0.0449 – 0.0715AR(1) – 0.0558AR(35) + 0.1475MA(1) + 0.0337MA(26) + ȗt

SPSABt = -0.0449 – 0.0715SPSABt-1 – 0.0558SPSABt-35 + 0.1475ȗt-1 + 0.0337ȗt-26 + ȗt SABt − SABt-1 = -0.0449 – 0.0715(SABt-1 − SABt-2) – 0.0558(SABt-35 − SABt-36) + 0.1475ȗt-1 + 0.0337ȗt-26 + ȗt

SABt = -0.0449 + 0.9285SABt-1 + 0.0715SABt-2 – 0.0558SABt-35 + 0.0558SABt-36 + 0.1475ȗt-1 + 0.0337ȗt-26 + ȗt

Một phần của tài liệu TIỂU LUẬN môn học KINH tế LƯỢNG ỨNG DỤNG NGÀNH tài CHÍNH đề tài dự báo GIÁ CHỨNG KHOÁN TỔNG CTCP BIA RƯỢUNƯỚC GIẢI KHÁT sài gòn (Trang 44 - 53)

Tải bản đầy đủ (PDF)

(69 trang)