انتخاب سبد سهام بهینه بر اساس ریزش مورد انتظار چندافقی پارامتریک و ناپارامتریک

ﭘﮋوﻫﺶ ﺣﺎﺿﺮ ﻃﺮاﺣﯽ دو ﻣﺪل ﺳﺒﺪ ﺳﻬﺎم ﺑﺮ اﺳﺎس رﯾﺰش ﻣﻮرد اﻧﺘﻈﺎر ﭼﻨﺪاﻓﻘﯽ ﮐﻪ اوﻟﯽ ﺑﺮ اﺳﺎس ﺳﻨﺎرﯾﻮﺳﺎزی ﺗﺎرﯾﺨﯽ و دوﻣﯽ ﺑﻪﺻﻮرت ﭘﺎراﻣﺘﺮﯾﮏ اﺳﺖ. روش ﺑﺮ اﺳﺎس ﺗﻮزﯾﻊ ﻣﺨﻠﻮط ﻧﺮﻣﺎل - ﻻﭘﻼس ﭼﻨﺪﮔﺎﻧﻪ ﺑﺎﻫﺪف ﺑﺮازش ﻣﻨﺎﺳﺐ در دﻧﺒﺎﻟﻪﻫﺎی ﺗﻮزﯾﻊ ﺑﺎزده اﺳﺖ. ﺑﺮای ﻣﺤﺎﺳﺒﻪ ﻋﺪدی رﯾﺰش ﻣﻮرد اﻧﺘﻈﺎر ﭼﻨﺪاﻓﻘﯽ در ﺷﮑﻞ ﭘﺎراﻣﺘﺮﯾﮏ از اﻣﯿﺪ ﭼﻨﺪﮐﯽ اﺳﺘﻔﺎده ﻣﯽﺷﻮد. ﺳﺒﺪ ﺳﻬﺎم ﻧﻤﻮﻧﻪای ﭘﮋوﻫﺶ از 8 ﺷﺎﺧﺺ ﯾﺎ ﺻﻨﻌﺖ ﻋﻤﺪه از ﺑﻮرس اوراق ﺑﻬﺎدار ﺗﻬﺮان ﺷﺎﻣﻞ ﺷﺎﺧﺺ ﺑﺎﻧﮏﻫﺎ، ﻣﻮاد ﺷﯿﻤﯿﺎﯾﯽ، ﻣﻮاد داروﯾﯽ، ﺧﻮدرو، ﮐﺎﺷﯽ و ﺳﺮاﻣﯿﮏ، ﮐﺎﻧﯽ ﻓﻠﺰی، ﺳﯿﻤﺎن و ﻓﻨﯽ - ﻣﻬﻨﺪﺳﯽ در ﺑﺎزه 1390/01/6 ﺗﺎ 1399/12/26 ﺗﺸﮑﯿﻞ ﺷﺪه اﺳﺖ. ﻧﺘﯿﺠﻪ ﻣﻄﺎﻟﻌﻪ ﺗﺠﺮﺑﯽ ﺑﺮ روی ﯾﮏ ﺳﺒﺪ ﺳﻬﺎم ﺑﺎ ﻫﺸﺖ ﺷﺎﺧﺺ از ﺑﻮرس اوراق ﺑﻬﺎدار ﺗﻬﺮان ﺑﺎ ﮐﺪﻧﻮﯾﺴﯽ در ﻣﺤﯿﻂ ﻧﺮماﻓﺰار ﻣﺘﻠﺐ و ﺑﺎ اﺳﺘﻔﺎده از اﻟﮕﻮرﯾﺘﻢ ﺑﻬﯿﻨﻪﺳﺎز ﺗﺠﻤﻌﯽ ذرات در ﺑﺎزه 1390 ﺗﺎ 1399 ﻧﺸﺎن ﻣﯽدﻫﺪ ﮐﻪ روﯾﮑﺮد ﭘﺎراﻣﺘﺮﯾﮏ در دادهﻫﺎی ﺗﺴﺖ و در ﺳﻪ ﻣﻌﯿﺎر ﻣﺘﻮﺳﻂ ﺑﺎزده ﮐﺴﺐ ﺷﺪه، ﻧﺴﺒﺖ ﺷﺎرپ و ﺧﻄﺎی ﻧﺴﺒﯽ ﺑﯿﻦ رﯾﺰش ﻣﻮرد اﻧﺘﻈﺎر ﭼﻨﺪاﻓﻘﯽ ﭘﯿﺶﺑﯿﻨﯽ ﺷﺪه ﺗﻮﺳﻂ ﻣﺪل و ﻣﻘﺪار واﻗﻌﯽ ﻣﺸﺎﻫﺪه ﺷﺪه، ﻋﻤﻠﮑﺮد ﺑﻬﺘﺮی ﻧﺴﺒﺖ ﺑﻪ روﯾﮑﺮد ﺳﻨﺎرﯾﻮﺳﺎزی ﺗﺎرﯾﺨﯽ دارد. درﻧﺘﯿﺠﻪ در ﺻﻮرت ﺗﺮدﯾﺪ ﯾﺎ ﺗﺮس از مقوله افق زﻣﺎﻧﯽ ﻣﯽﺗﻮان از روﯾﮑﺮد رﯾﺰش ﻣﻮرد اﻧﺘﻈﺎر ﭼﻨﺪاﻓﻘﯽ ﭘﺎراﻣﺘﺮﯾﮏ اﺳﺘﻔﺎده ﮐﺮد.

optimal portfolio selection based on parametric and non-parametric multi-horizon expected shortfall

the aim of the current research is to design two stock portfolio models based on the expected multi-horizon drop, the first is based on historical scenario creation and the second is parametric. the method is based on the multiple laplace-normal mixed distribution with the aim of fitting properly in the return distribution sequences. quantitative hope is used to numerically calculate the expected multi-dimensional drop in parametric form. a sample research portfolio of 8 major indexes or industries from the tehran stock exchange, including banks, chemicals, pharmaceuticals, automobiles, tiles and ceramics, metal minerals, cement, and technical-engineering in the period 01/01/2013 to 12/26 it was formed in 2019. the result of the experimental study on a stock portfolio with eight indices from the tehran stock exchange by coding in the matlab software environment and using the particle cumulative optimizer algorithm in the period from 1390 to 1399 shows that the parametric approach in the test data and in the three average performance measures , the sharpe ratio and the relative error between the model-predicted polygonal expected fall and the actual observed value have a better performance than the historical scenario approach. as a result, in case of doubt or fear of the time horizon category, the parametric multi-horizon expected decline approach can be used.