Görüntüleme
8
İndirme
1
Bu makalede ChatGPT’nin geçmiş finansal tabloları ve sektörle ilgili güncel önemli gelişmeleri yorumlayarak finansal değerleri ne ölçüde doğru tahmin edebildiği analiz edilmiştir. Araştırma kapsamında BIST100 endeksinde işlem gören 73 şirkete ait finansal veri ve Microsoft BING Chat uygulamasından temin edilen sektör spesifik gelişmelerin özetlendiği metinlerden faydalanılmıştır. İlk aşamada ChatGPT’nin 2019-2021 dönemine ait finansal değerleri yorumlayarak 2022 yılı için satış, net kâr ve temettü dağıtımı tahminleri yapması istenmiştir. İkinci aşamada ChatGPT’ye sektör spesifik gelişmelerle ilgili metin girilmiş ve tahminini yenilemesi istenmiştir. Daha sonrasında bu iki aşamada ChatGPT tarafından yapılan tahminler gerçekleşmeler ile kıyaslanmış ve ChatGPT’nin bu tahminlerde ne ölçüde başarılı olduğu anlaşılmaya çalışılmıştır. Araştırma sonuçlarına göre ChatGPT satış değerlerini yalnızca finansal tablolardan faydalandığı ilk aşamada medyan değerler olarak %18 sapma ile tahmin edebilmiş, sektörle ilgili güncel gelişmeleri kullandığı ikinci aşamada ise %18,9 sapma ile tahmin edebilmiştir. Net kâr için ilk aşamada yapılan tahminlerin medyan sapması %57,8 olurken ikinci aşamada %64,9 olmuştur. Kâr payları için medyan sapmalar ilk aşamada %54,6 iken ikinci aşamada %51,8 olmuştur. Araştırma bulgularına göre ChatGPT’nin finansal ve haber analiz yetenekleri sınırlı olmakla birlikte ciddi bir gelişme süreci yaşadığı gözlemlenmiştir.
This article analyzes how correct ChatGPT can predict financial values by interpreting past financial statements and current significant developments related to the industry. In the framework of the study, the financial data of 73 companies trading in the BIST100 index and the Microsoft BING Chat application have used the text that summarizes specific industry developments. In the first phase, ChatGPT was asked to make sales, net profit and forecast for the year 2022 by interpreting the financial values for the period 2019-2021. In the second phase, ChatGPT was asked to update the text on specific developments in the sector. Later, in these two stages, the forecasts made by ChatGPT were compared with the events, and it was tried to understand how successful ChatGPT was in these forecasts. According to the research results, ChatGPT was able to estimate sales values with a disadvantage of 18% as median values in the first phase when it only benefited from the financial statements, while in the second phase it was able to estimate sales values with a disadvantage of 18.9%. The median deviation of the forecast for net profit in the first stage was 57.8%, while in the second stage it was 64.9%. The median disadvantages for profits were 54.6 percent in the first stage, while 51.8 percent in the second stage. Research findings show that ChatGPT’s financial and news analysis capabilities are limited, but it has a serious development process.
This article analyzes to what extent ChatGPT can accurately estimate financial values by interpreting past financial statements and current important developments in the sector. Within the scope of the research, the financial data of 73 companies traded in the BIST100 index and the texts that summarize the sector-specific developments obtained from the Microsoft BING Chat application were used. In the first stage, ChatGPT was asked to interpret the financial values for the 2019-2021 period and make estimates for sales, net profit, and dividend distribution for 2022. In the second stage, text about sector-specific developments was entered into ChatGPT and it was asked to revise its estimate. Afterward, the estimations made by ChatGPT in these two stages were compared with the realizations and it was tried to understand how successful ChatGPT was in these estimations. According to the results of the research, ChatGPT was able to estimate the sales values with 18% deviation as median values in the first stage, when it used only the financials, and with 18.9% deviation in the second stage, when it used both financials and the current developments in the sector. The median deviation of the estimates for net profit in the first phase was 57.8%, while it was 64.9% in the second phase. The median deviations for dividends were 54.6% in the first stage and 51.8% in the second stage. According to the research findings, it has been observed that ChatGPT has experienced a serious development process, although its financial and news analysis capabilities are limited.
Alan : Sosyal, Beşeri ve İdari Bilimler
Dergi Türü : Uluslararası
Makale Detay 
Benzer Makaleler
PDF Görüntüle
Dergi Bilgisi
Eseri Dinleyin
Alıntı Yap
Bu Sayfayı Yazdırın
Paylaş
Mendeley
